This is your personal invite to join our machine learning group initiative, a collective experiment to unite ML practitioners.
Our goal is to establishing a community with a shared culture of best practices, methodologies, and greater principles applied to wider ML projects.
CEO, SingularityNET
Dr. Ben Goertzel is the CEO of the decentralized AI network SingularityNET, a blockchain-based AI platform company, and the Chief Science Advisor of Hanson Robotics, where for several years he led the team developing the AI software for the Sophia robot. Dr. Goertzel also serves as Chairman of the Artificial General Intelligence Society, the OpenCog Foundation, the Decentralized AI Alliance and the futurist nonprofit Humanity+. Dr. Goertzel is one of the world’s foremost experts in Artificial General Intelligence, a subfield of AI oriented toward creating thinking machines with general cognitive capability at the human level and beyond. He also has decades of expertise applying AI to practical problems in areas ranging from natural language processing and data mining to robotics, video gaming, national security and bioinformatics. He has published 20 scientific books and 140+ scientific research papers, and is the main architect and designer of the OpenCog system and associated design for human-level general intelligence. He obtained his PHD in mathematics from Temple University in 1989.
Talk: Neural-Symbolic AI for Creativity, Generalization and Transfer Learning
Head of Customer Experience, Yle
Jaakko Lempinen works as a Head of Customer Experience at Yle – Finnish public broadcaster. For the last 14 years Jaakko has been developing data and advanced analytics (ie.ML/AI) - solutions for the media industry for both commercial and public media. Jaakko is a well-known speaker on this topic and is always eager to learn more about the future of artificial intelligence & customer experience in media sector.
Talk: How Finnish Public Broadcaster Yle is the Only Streaming Service Beating Out Netflix
Snorkel AI - Machine Learning Engineer & Open Source Lead
Chip Huyen works to bring the best practices to machine learning production. She’s built AI applications at Snorkel AI, Netflix, NVIDIA, and Primer. She graduated from Stanford, where she taught TensorFlow for Deep Learning Research. She’s also the author of four bestselling Vietnamese books.
Talk: Design, Data, Development, Deployment: Breaking Down the Machine Learning Production Pipeline
Melanie Mitchell is the Davis Professor of Complexity at the Santa Fe Institute, and Professor of Computer Science (currently on leave) at Portland State University. Her current research focuses on conceptual abstraction, analogy-making, and visual recognition in artificial intelligence systems.
Melanie is the author or editor of six books and numerous scholarly papers in the fields of artificial intelligence, cognitive science, and complex systems. Her book Complexity: A Guided Tour (Oxford University Press) won the 2010 Phi Beta Kappa Science Book Award and was named by Amazon.com as one of the ten best science books of 2009. Her latest book is Artificial Intelligence: A Guide for Thinking Humans (Farrar, Straus, and Giroux).
Talk: Abstraction and Analogy in Natural and Artificial Intelligence
Director, Recommendations Systems, Twitch
Ashish is a Director of Recommendations at Twitch where he works on building scalable recommendation systems across a variety of product surfaces, connecting content to people. He has worked on recommendations systems at multiple organizations, most notably Twitter where he led Trends and Events recommendations and at Capital One. Ashish was also a co-founder of GALE Partners, a full-service digital agency in Toronto. He has over 20 years of experience with over a decade in building ML systems. Ashish is a guest lecturer at IIT BHU teaching Deep Learning and also writing a book on advanced NLP techniques.
Talk: Key Design Patterns for Building Recommendation Systems At Scale
Co-Founder / CTO – maiot
I'm a seasoned SRE/Opsguy with 11+ years experience in data-heavy companies (ecommerce, SaaS, advertising). For over two years now I'm one of two CTOs of maiot, a Munich-based AI startup. Originally focused on predictive maintenance / asset optimization of industrial assets and commercial vehicles, we're now making our internal tech stack available to a broad audience.
Talk: A Tale of A Thousand Pipelines
Chief Analytics Officer, Banorte
Jose Murillo established and leads the Analytics Business Unit at Banorte –the second largest financial group in Mexico. His group has been recognized as a success story within the data and analytics industry. The value created with analytics on 2019, the fifth year of operations of his group, was equivalent to 1 billion USD –during his tenure the value created exceeds 3 billion USD. This case was published at Harvard Business Review as an example of a company which made its analytics investments pay-off (2018) and by Forbes (2020) as an example of a firm that has developed high ROI artificial intelligence applications. In addition, the Harvard Business School has written and taught a case study on Jose’s analytics and digital transformation leadership. He obtained the Lafferty Global Award on Credit Card Excellence for the impact of analytics on customer equity (2016). Jose has been named repeatedly among the top global analytics and data professionals by Corinium. He has been published in several industry and refereed journals, he is an international speaker at data science and artificial intelligence forums, and he is a member of Queen’s University Advisory Board for the Master’s in Management Analytics.
Prior to Banorte, Jose was a top ranking official at Mexico’s Central Bank and participated for more than a decade at the Monetary Policy Committee holding the staff’s view on inflation –the key variable for policy decision. Also, he was an advisor to the International Monetary Fund and led missions to several Latin American countries. At the academia he taught economics at Rice University, ITAM and El Colegio de Mexico. He holds a PhD in economics from Rice University and a BA from ITAM (Cum Laude and national prize winner from the National Chamber of Commerce and Tlacalel).
Talk: Banorte’s AI Transformation Journey: How the Analytics Team of this Bank Yielded 3 Billion USD Revenue During the Past Five Years
Applied Research Scientist, Pinterest
Nadia Fawaz is an applied research scientist at Pinterest and the tech lead for Inclusive AI. Her research and engineering interests include machine learning for personalization, AI fairness and data privacy. Her work leverages techniques from AI including deep learning, information theory, fairness and privacy theory, and aims at bridging theory and practice. She was a winner of the ACM RecSyS challenge on Context-Aware Movie Recommendations CAMRa2011, her 2012 UAI paper "Guess Who Rated This Movie: Identifying Users Through Subspace Clustering" was featured in an MIT TechReview article as “The Ultimate Challenge For Recommendation Engines”, and her work on inclusive AI was featured many press outlets, including FastCompany and Vogue Business. Earlier, she was a Staff Software Engineer in Machine Learning and the tech lead for the job recommendations team at LinkedIn, a principal research scientist at Technicolor Research lab, Palo Alto, and a postdoctoral researcher at the Massachusetts Institute of Technology, Research Laboratory of Electronics. She received her Ph.D. in EECS in 2008 and her Diplome d'ingenieur (M.Sc.) in EECS in 2005 both from Ecole Nationale Superieure des Telecommunications de Paris and EURECOM, France. She is a Member of the IEEE and of the ACM.
Talk: Inclusive Search and Recommendation
Senior Data Scientist, Precisely
Brandy Freitas is a senior data scientist at Precisely (formerly Pitney Bowes Software and Data), where she works with clients in a wide variety of industries to develop analytical solutions for their business needs. Brandy is a research physicist-turned-data scientist based in Boston, MA. Her academic research focused primarily on protein structure determination, applying machine learning techniques to single-particle cryoelectron microscopy data. Brandy is a National Science Foundation Graduate Research Fellow and a James Mills Pierce Fellow. She holds an undergraduate degree in physics and chemistry from the Rochester Institute of Technology and did her graduate work in biophysics at Harvard University.
Talk: Harnessing Geospatial Data for Machine Learning
Cognitive Data Specialist, Macadamian Technologies
Having traversed academia, healthcare and industry, and trained as a cognitive scientist, Roxana tackles research questions from various lenses striving to provide comprehensive, unique, and data-driven insights. Sought-after public speaker and facilitator, she’s counting over 30 conference presentations and speaking invitations, and 4 undergraduate courses on topics centered around human cognition and research approaches. When not in front of the computer, she enjoys the solitude of her garden, trying to recreate a bit of the magic from “the old world” where she spent summers walking across vineyards with a whistle to send away starlings, and falls picking hectares of grapes to make magic.
Talk: The Algorithm is Not Enough: UX Meets Data Science
Chief Strategy Officer, Akendi
Over the past twenty-five years, Scott has worked in the areas of business strategy, product design and development in the high tech sector. He brings with him cross-sector expertise and experience working with clients in industries such as aviation, telecom, and finance. His primary area of focus over the last several years has been in healthcare. Scott has a master’s degree in Physics from Queen’s University. Prior to joining Macadamian in 2006, Scott co-founded Maskery and Associates, a UX design consultancy. His focus the past several years has been the improved integration of multi-discipline teams that include clinical, business, technical, design, and data expertise; and the enhancement of the tools and practices these teams apply.
Talk: The algorithm is not enough: UX meets Data Science
Director of Data Science, The Washington Post
Patrick is the Director of Data Science at the Washington Post. His team of data scientists and software engineers builds personalization algorithms, advertisement targeting platforms, and products at the intersection of journalism and machine learning. He also lead the development of Zeus Technology, a publisher focused suite of advertising technologies that utilize NLP technology to drive contextual ad targeting. Before joining the Washington Post, he lead engineering teams at Amazon Web Services building cloud networking solutions used by some of the largest companies in the world.
Talk: Raising the Quality of Online Conversations with Machine Learning
Senior Data Scientist, Cineplex
Valerii joined Cineplex last year to drive the best practice in development and serving of ML models. Previously he was a Senior Data Scientist at Real Tech, an IoT company in the Water Industry. Prior to that he has contributed to the development of innovative solutions for a variety of brands such as LG Electronics, Panasonic, SAMSUNG, Toyota, Scotiabank. Author of 20 patented inventions in Signal Processing, Electronics and Computing. He has a University Degree in Telecom Engineering and PhD in Automated Control Systems. Lifelong learner and researcher by nature.
Talk: Movie Attendance Forecasting: Machine Learning in Post-COVID Market
Vice President, Data Science and Advanced Analytics
Dr. Mamdani is Vice President of Data Science and Advanced Analytics at Unity Health Toronto. Dr. Mamdani’s team bridges advanced analytics including machine learning with clinical and management decision making to improve patient outcomes and hospital efficiency. Dr. Mamdani is also Professor in the Department of Medicine of the Faculty of Medicine, the Leslie Dan Faculty of Pharmacy, and the Institute of Health Policy, Management and Evaluation of the Dalla Lana Faculty of Public Health. Dr. Mamdani is the Director of the University of Toronto Faculty of Medicine Centre for Machine Learning in Medicine. He is also adjunct Senior Scientist at the Institute for Clinical Evaluative Sciences (ICES) and a Faculty Affiliate of the Vector Institute, which is a leading institution for artificial intelligence research in Canada. Further, Dr. Mamdani is a member of the Human Drug Advisory Panel of the Patented Medicine Prices Review Board (PMPRB). Previously, Dr. Mamdani founded the Ontario Drug Policy Research Network (ODPRN), which is among the world’s most impactful collaborations between researchers and drug policy decision-makers. He was also the Founding Director of the Li Ka Shing Centre for Healthcare Analytics Research and Training (LKS-CHART) of Unity Health Toronto and the Founding Director of the Applied Health Research Centre (AHRC) of the Li Ka Shing Knowledge Institute of Unity Health Toronto, which is Toronto’s leading academic research organization focused on the design and implementation of multicentre clinical research initiatives. In 2010, Dr. Mamdani was named among Canada’s Top 40 under 40. Prior to joining the Li Ka Shing Knowledge Institute and Unity Health Toronto, Dr. Mamdani was a Director of Outcomes Research at Pfizer Global Pharmaceuticals in New York. Dr. Mamdani’s research interests include pharmacoepidemiology, pharmacoeconomics, drug policy, and the application of advanced analytics approaches to clinical problems and health policy decision-making. He has published approximately 500 research studies in peer-reviewed medical journals, including leading journals such as the New England Journal of Medicine, the Lancet, the Journal of the American Medical Association, the British Medical Journal, and Annals of Internal Medicine. His research has been cited over 34,000 times and has an h-index of 90. Dr. Mamdani obtained a Doctor of Pharmacy degree (PharmD) from the University of Michigan (Ann Arbor) in 1995 and subsequently completed a fellowship in pharmacoeconomics and outcomes research at the Detroit Medical Center in 1997. During his fellowship, Dr. Mamdani obtained a Master of Arts degree in Economics from Wayne State University in Detroit, Michigan with a concentration in econometric theory. He then completed a Master of Public Health degree from Harvard University in 1998 with a concentration in quantitative methods, focusing on biostatistics and epidemiological principles.
Talk: Applied Machine Learning in Healthcare - Practical and Legal Considerations
Talk: A Cookbook for Deep Continuous-Time Predictive Models
Abstract:
Do you want to make predictions based on irregularly-sampled, sparse time series? This tutorial will outline a series of approaches to this task based on neural networks. We'll start with simple feedforward approaches, and gradually build towards latent-variable stochastic differential equation models. This talk will also highlight some recent research on regularizing differential equation-based models to be more computationally efficient to solve.
What You'll Learn
You'll learn about the main existing approaches for building flexible time series models, and their strengths and weaknesses.
Talk: Neural-Symbolic AI for Creativity, Generalization and Transfer Learning
Abstract:
Neural-Symbolic methods, combining neural ML tools with symbolic logical reasoning methods, have greater capability than current purely subsymbolic ML methods for transfer learning, generalization beyond the training space, and creative hypothesis and invention. This talk will give examples of neural-symbolic AI implemented using the OpenCog AI framework, including semantics-preserving hypergraph embeddings and probabilistic logic based explanations of ML-identified data patterns. Practical applications will be discussed, including personalized medicine, humanoid robotics and grammar learning.
What You'll Learn:
How to think about, and build, neural-symbolic AI systems that generalize, transfer knowledge and perform creative hypothesis generation.
Talk: A Tale of A Thousand Pipelines
Abstract:
This is an Ops-guy's story about we had to build scalable deep learning pipelines for hundreds of model trainings on giant timeseries datasets. Ah, and how we saved 80% of our cost along the way.
What You'll Learn:
Real-world learnings from putting deep learning models rapidly from research to production through solid Ops and orchestration.
Talk: Abstraction and Analogy in Natural and Artificial Intelligence
Abstract:
In 1955, John McCarthy and colleagues proposed an AI summer research project with the following aim: “An attempt will be made to find how to make machines use language, form abstractions and concepts, solve kinds of problems now reserved for humans, and improve themselves.” More than six decades later, all of these research topics remain open and actively investigated in the AI community. While AI has made dramatic progress over the last decade in areas such as vision, natural language processing, and robotics, current AI systems still almost entirely lack the ability to form humanlike concepts and abstractions.
Some cognitive scientists have proposed that analogy-making is a central mechanism for conceptual abstraction and understanding in humans. Douglas Hofstadter called analogy-making “the core of cognition”, and Hofstadter and co-author Emmanuel Sander noted, “Without concepts there can be no thought, and without analogies there can be no concepts.” In this talk I will reflect on the role played by analogy-making at all levels of intelligence, and on how analogy-making abilities will be central in developing AI systems with humanlike intelligence.
What You'll Learn:
You'll learn how modern AI and ML are approaching the problem of conceptual abstraction and analogy-making, and how these approaches compare with human abilities in these areas.
Talk: Key Design Patterns for Building Recommendation Systems At Scale
Abstract:
Description of key design patterns useful for building scalable recommendation systems, based on learnings from deploying several such systems in the field.
What You'll Learn:
Practical considerations in building real life recommendation systems
Talk: How Finnish Public Broadcaster Yle is the Only Streaming Service Beating Out Netflix
Abstract:
I’ll talk about how AI will shape the future of media experience and how Yle is shaping it’s operations around this change. I’ll give few examples of how ideas are scaled into products across the whole organisation. I’ll also talk about how the culture changes within organisations as they start to benefit more from progressive data solutions – what are the future skills that every organisation should have and how to get started with the change.
What You'll Learn:
Case studies from media sector, How to drive the change in your organisation and what do you actually need to make the change.
Talk: Design, Data, Development, Deployment: Breaking Down the Machine Learning Production Pipeline
Abstract:
Machine learning has found increasing use in the real world, and yet a framework for productionizing machine learning systems is not well understood.
This talk outlines the challenges and approaches to designing, developing, and deploying ML systems. It starts with the gap between ML in research and ML in production. It examines how ML applications differ from traditional software engineering applications, the scaling challenge, and the rise of MLOps.
The next part covers the four main stages in the iterative process of ML systems design. For each stage, it breaks down the steps needed, the tradeoffs of different solutions at each step.
The talk ends with a survey of the MLOps landscape by analyzing over 200 available tools, where they fit into the ecosystem, and what’s missing in the ecosystem.
What You'll Learn:
Attendees will gain an understanding of principles of knowledge translation in applied machine learning in healthcare and understand issues related to privacy and ethics as well as legal considerations.
Talk: Applied Machine Learning In Healthcare - Practical And Legal Considerations
Abstract:
Applied machine learning has the potential to transform healthcare, particularly in the areas of automation, prediction, and optimization. However, numerous challenges to the acquisition, storage, and utilization of data as well as the development of practical machine learning algorithms and change management principles need to be considered. This talk will provide an overview of the process of applying ML into healthcare and the legal and ethical considerations needed for data access and application.
What You'll Learn:
Attendees will gain an understanding of principles of knowledge translation in applied machine learning in healthcare and understand issues related to privacy and ethics as well as legal considerations.
Talk: Raising the Quality of Online Conversations with Machine Learning
Abstract:
The quality of online comments is critical to the Washington Post. Learn how they built a machine learning system for automatically moderating comments from millions of readers. We will share the technical challenges with building the comment moderation platform and how we raised the quality of online conversations with machine learning.
What You'll Learn:
How to build a system that utilizes both human and machine learning moderation to efficiently scale to millions of reader comments.
Talk: Banorte's AI Transformation Journey: How the Analytics Team of this Bank Yielded 3 Billion USD Revenue During the Past Five Years.
Abstract:
There are high expectatios about AI initiatives across different industries in North America. However, too often results have been disappointing producing some backlash against digital transformation efforts. The capacity to implment and demonstrate high ROI AI projects changes this dynamic. This talk will delve into Banorte's transformation journey into an AI enhanced organization with data science projects yielding a net revenue that exceeds 3 billion USD during the past five years and avoiding the transformational fatigue.
What You'll Learn:
1. How to measure AI contribution to the bottom line
2. What are key prerequisites to focus yield high ROI on AI projects
3. Where to focus AI inititiatives to have a large organizational impact: revenue or cost?
Talk: Inclusive Search and Recommendations
Abstract:
Machine learning powers many advanced search and recommendation systems, and user experience strongly depends on how well ML systems perform across all data segments. This performance can be impacted by biases, which can lead to a subpar experience for subsets of users, content providers, applications or use cases. Biases may arise at different stages in machine learning systems, from existing societal biases in the data, to biases introduced by the data collection or modeling processes. These biases may impact the performance of various components of ML systems, from offline training, to evaluation and online serving in production systems. Specific techniques have been developed to help reduce bias at each stage of an ML system. We will describe sources of bias in ML technology, why addressing bias matters, and techniques to mitigate bias, with examples from our work on inclusive AI at Pinterest. Mitigating bias in machine learning systems is crucial to successfully achieve our mission to "bring everyone the inspiration to create a life they love".
What You'll Learn:
We will describe sources of bias in ML technology, why addressing bias matters, and techniques to mitigate bias, with examples from our work on inclusive AI at Pinterest.
Talk: Harnessing Geospatial Data for Machine Learning
Abstract:
Working with and analyzing geospatial data requires a different and often nuanced approach from most data types, especially to derive spatial predictions and detect patterns using machine learning applications. Many data scientists and analysts are not used to fully leveraging the power of geospatial data, and often don't know what business questions to ask, aren't aware of which algorithms are available to them to enrich their models, or resort to eliminating spatial variables entirely in order to use the data with common machine learning algorithms.
What You'll Learn:
How to maximize the value of geospatial data using machine learning and artificial intelligence techniques, business problems that can be tackled in a variety of industries using this type of data, and how to utilize algorithms specific to spatial data.
Talk: The Algorithm is not Enough: UX Meets Data Science
Abstract:
We live in an age of data; so much data that it’s overwhelming. We also live in the age of UX where user centricity is no longer the exception or a market differentiator - it is now the norm. Many product companies have an established team of data science experts; many have an established team of UX experts. However, it’s not very common to come across companies that have both. And if they do, each team often works in a vacuum, siloed from each other.
What You'll Learn:
Cross-disciplinary artefacts and processes generally not applied in digital product development Human centered data science
Talk: The Algorithm is not Enough: UX Meets Data Science
Abstract:
We live in an age of data; so much data that it’s overwhelming. We also live in the age of UX where user centricity is no longer the exception or a market differentiator - it is now the norm. Many product companies have an established team of data science experts; many have an established team of UX experts. However, it’s not very common to come across companies that have both. And if they do, each team often works in a vacuum, siloed from each other.
What You'll Learn:
Cross-disciplinary artefacts and processes generally not applied in digital product development Human centered data science
Senior Data Scientist,The Washington Post
Ling Jiang is a data scientist at the Washington Post. She enjoys working on data mining and knowledge discovery from large volume of data. She is skilled in various machine learning and data mining techniques, and using them to tackle business problems. At the Post, she has successfully built several data-powered products using machine learning and NLP techniques. She graduated from Drexel University with a PhD degree in Information Science before joining The Washington Post.
Talk: Raising the Quality of Online Conversations with Machine Learning
Talk: Raising the Quality of Online Conversations with Machine Learning
Abstract:
The quality of online comments is critical to the Washington Post. Learn how they built a machine learning system for automatically moderating comments from millions of readers. We will share the technical challenges with building the comment moderation platform and how we raised the quality of online conversations with machine learning.
What You'll Learn:
How to build a system that utilizes both human and machine learning moderation to efficiently scale to millions of reader comments.
Founder, Arima
Winston is the founder of Arima, a synthetic database that captures individual consumer-level behavioural and demographic attributes across Canada. Arima aims to be a full-stack solution for data scientists to easily acquire individual-level consumer intelligence, connecting those who want better data to build more robust ML models and those who have data, without compromising data privacy. Prior to founding Arima, Winston was the Director of Data Science at PwC and Omnicom Mediacom. Winston is also a part time faculty member at Northeastern University Toronto, and sits on the advisory board of the Master of Analytics program.
Talk: A Machine Learning based Privacy-Preserving Framework for Generating Synthetic Data from Aggregated Sources
Talk: A Machine Learning based Privacy-Preserving Framework for Generating Synthetic Data from Aggregated Sources
Abstract:
A synthetic dataset is a data object that is generated programmatically, and it is often necessary for situations where data privacy is a concern, or when collecting data is difficult or costly. Although it is a fundamental step for many data science tasks, an efficient and standard framework is absent. In this presentation, we study a specific synthetic data generation task called downscaling, a procedure to infer high-resolution information (e.g., individual level records) from low-resolution variables (e.g., average of many individual records), and propose a multi-stage framework. Specifically, we discuss 1) how synthetic data is generated from aggregated sources like census, 2) why is this important from a application perspectives, and 3) two real world use cases demonstrating why using synthetic data generation can significantly improve model performances.
What You'll Learn:
I will present a novel method for generating synthetic datasets (which has not yet been published) as well as 2 real world case studies of Arima's partners on how synthetic data has improved their model performances.
Senior Business Development Manager - AI/ML & VC, Amazon Web Services
Ari Kalfayan is a Senior Business Development Manager at AWS in charge of AI/ML startups. Ari began his career in the AI/ML space in 2009, where he led sales at Figure Eight (sold for $300M to Appen) which pioneered data labeling technology. Ari then joined the founding team of Weights & Biases (W&B) in 2018 which is pioneering machine learning tools for deep learning. Ari focuses on helping early-stage co-founders, who are building machine learning startups, accelerate growth and achieve product market fit. He specializes in helping early stage startups accelerate their growth by helping them connect with internal resources at AWS/Amazon, with go-to-market strategy, introductions to enterprise accounts, and connecting startups with investors.
Talk: Winning Your First 50 Enterprise Customers: Practical Strategies to Successfully Launch a Machine Learning Startup
Talk: Winning Your First 50 Enterprise Customers: Practical Strategies to Successfully Launch a Machine Learning Startup
Abstract:
This talk is designed to help you land your first 50 enterprise machine learning customers. Whether you are developing your first machine learning application, creating an enterprise ML infrastructure startup, or creating new Machine/Deep Learning tools, this hands-on session is designed to share practical strategies, growth hacks, and specific techniques to use that will win you your first customers and scale.
This presentation will be broken up into three parts:
- Landing your first customer (0-1 customer)
- Validating your business model (1-10 customers)
- Scaling your business (10-50 customers)
This presentation is designed to leave you with practical tips to help you acquire new customers, no matter your funding stage.
What You'll Learn:
Practical advise and mistakes from having launched two top tier ML tools companies
Director, Amazon Alexa AI
Dr. Sujith Ravi is a Director at Amazon Alexa AI where he is leading efforts to build the future of multimodal conversational AI experiences at scale. Prior to that, he was leading and managing multiple ML and NLP teams and efforts in Google AI. He founded and headed Google’s large-scale graph-based semi-supervised learning platform, deep learning platform for structured and unstructured data as well as on-device machine learning efforts for products used by billions of people in Search, Ads, Assistant, Gmail, Photos, Android, Cloud and YouTube. These technologies power conversational AI (e.g., Smart Reply), Web and Image Search; On-Device predictions in Android and Assistant; and ML platforms like Neural Structured Learning in TensorFlow, Learn2Compress as Google Cloud service, TensorFlow Lite for edge devices.
Dr. Ravi has authored over 100 scientific publications and patents in top-tier machine learning and natural language processing conferences. His work has been featured in press: Wired, Forbes, Forrester, New York Times, TechCrunch, VentureBeat, Engadget, New Scientist, among others, and also won the SIGDIAL Best Paper Award in 2019 and ACM SIGKDD Best Research Paper Award in 2014. For multiple years, he was a mentor for Google Launchpad startups. Dr. Ravi was the Co-Chair (AI and deep learning) for the 2019 National Academy of Engineering (NAE) Frontiers of Engineering symposium. He was also the Co-Chair for ICML 2019, NAACL 2019, and NeurIPS 2018 ML workshops and regularly serves as Senior/Area Chair and PC of top-tier machine learning and natural language processing conferences like NeurIPS, ICML, ACL, NAACL, AAAI, EMNLP, COLING, KDD, and WSDM.
Website: www.sravi.org
Twitter: @ravisujith
LinkedIn: https://www.linkedin.com/in/sujithravi
Talk: Efficient AI: Building Efficient Neural Computing Machines on the Edge & Cloud
Talk: Efficient AI: Building Efficient Neural Computing Machines on the Edge & Cloud
Abstract:
Deep learning has changed the computing paradigm. Today, AI researchers & practitioners increasingly use deep neural networks for many applications across different modalities and areas such as NLP, Vision, Speech, Conversational and Multimodal AI. However, much of the Deep Learning revolution has been limited to the Cloud and highly specialized hardware. Recently the AI community has witnessed an increasing trend for training larger and larger neural models (e.g., GPT-3, T5, BERT) that achieve state-of-the-art results but require enormous computation, memory and energy resources on the Cloud. In order to enable AI experiences in real-time across all users and devices, ML models have to run efficiently on the Cloud and personal devices on the Edge (e.g., mobile phones, wearables, IoT) which have limited computing capabilities.
This talk will introduce our work on Neural Projection computing, an efficient AI paradigm, and a family of efficient Projection Neural Network architectures that yield fast (e.g., quadratic speedup for transformer networks) and tiny models that shrink memory requirements by upto 10000x while achieving near state-of-the-art performance powering vision and NLP applications on billions of mobile devices. Widespread increase in availability of connected “smart” appliances (e.g., conversational assistants) means that there is an ever-expanding surface area for mobile intelligence and ambient devices in homes. Our approach enables efficient ML to solve complex prediction tasks for such applications both on-device and on Cloud, keeping model size, compute and power usage low while simultaneously optimizing for accuracy.
What You'll Learn:
Cutting-edge technology & practical applications for efficient Deep Learning on the Edge & Cloud
Sr. Director - Robotics and Machine Learning, Rogers Communications
Gonzalo is the Sr. Director of Intelligent Automation at Rogers Communications. Gonzalo’s work focuses in identifying business problems that can be solved through new technologies like Machine Learning, Intelligent Automation and others. As well, he understands the challenges in operationalizing those solutions, having deployed and implemented several of them in ways that deliver measurable financial results for organization. Gonzalo hold a BSc in Electrical Engineering and an MBA from the Richard Ivey School of Business in Canada.
Talk: Predicting Which Customers Will Experience a Technical Issue Tomorrow and Will Call as A Result
CEO and Co-Founder, Paperspace
Dillon is the CEO and Cofounder of Paperspace, a 5-year-old cloud and machine learning infrastructure company backed by YCombinator, Battery Ventures, Sinewave Capital, Intel Capital, and more.
Talk: Productionizing Deep Learning Models at Scale
Lead A.I. Scientist, Coveo
Educated in several acronyms across the globe (UNISR, SFI, MIT), Jacopo was founder and CTO of Tooso, an A.I. company based in San Francisco: Tooso was named "Gartner Cool Vendor" in 2019 and acquired by Canadian unicorn Coveo in the same year. Jacopo is currently the Lead A.I. Scientist at Coveo, building A.I. products for a network of more than 400 customers, including several Fortune 500 companies. Together with his team, he combines product thinking with technical innovation on a variety of topics, presenting original findings at major conferences (KDD, HCOMP, ACL, ECAI, RecSys, etc.). In previous lives, he managed to get a Ph.D., do scienc-y things for a pro basketball team and simulate a pre-Columbian civilization.
Talk: AI in the Multiverse: Measuring ROI when A/B Tests are Not Possible
Vice President Data Strategy - North America, Mastercard
Joe’s career has been focused on building and scaling data-driven products, services, and businesses. Joe is Vice President of Data Strategy for North America at Mastercard with responsibility for realizing Mastercard’s global data strategy and Data Responsibility Imperative across all business lines in North America and enabling the growth of new services through region and product specific data strategies. Previously Joe was Lead Executive for Data at MaRS Discovery District in Toronto, North America’s largest urban innovation hub dedicated to growing Canada’s tech sector. At MaRS Joe founded and led the data practice, building strategic partnerships to scale Canadian data and AI businesses in sectors including retail, finance, energy, and healthcare. Joe has also held leadership roles in product development for location-based data-products and web services at Ordnance Survey in the UK and for secure real-time mobile data services at Blackberry. Joe holds a Masters degree in Geographical Information Science from the University of Edinburgh and an MBA from Manchester Business School and is a Certified Analytics Professional and a Privacy and Access by Design Ambassador.
Talk: Scaling Global Models with Regional Data Strategies and Model Governance
Senior Data Scientist, Scotiabank
Shirin is a senior data scientist at Artificial intelligent and machine learning team at Scotiabank. She has an engineering PhD from data science lab, Ryerson University, Canada. Her PhD was on applying Reinforcement leaning to prioritize software bugs in issue tracking system. Shirin's expertise includes but not limited to applying reinforcement learning, deep learning, classification, and clustering to real world problems. She also has experience working in an agile environment and been able to build machine learning solutions praised by internal clients and senior executives.
Talk: Customer Segmentation, Pricing, and Profit Optimization for international Banking
Talk: Predicting Which Customers Will Experience a Technical Issue Tomorrow and Will Call as A Result
Abstract:
Large telecom providers (and many other industries) spend tens of millions of dollars each year reacting to customer issues. This generally takes the form of large call center and repair technician workforces that are waiting for an issue to happen, in order to help solve it.
Utilizing machine learning and the power of robotic process automation (RPA), we have set out to determine a way to predict which customers are going to reach out with an issue, before they actually do – empowering us to take immediate action, to correct the issue, before a customer notices and before they have to spend their valuable time contact us.
This talk will focus on our journey to build this model, and how we are able to operationalize the findings quickly using RPA.
Finding a way to predict which customers will experience those issues, and taking and action BEFORE the customer calls has been a long sought after objective in the business. At Rogers, we have used Machine Learning to develop such a model with precision rates of over 90%. We have also combined this prediction with an action layer driven Robotics Automation, which takes the actions required to correct the technical issue "before the Customer notices it".
I would like to share how this ecosystem (ML, Robotics and process engineering) will result in significant benefits for the organization.
What You'll Learn:
Real world applications of ML & How we operationalize model findings quickly in an
Talk: Productionizing Deep Learning Models at Scale
Abstract:
The ecosystem for deploying SaaS applications includes countless tools for delivering an app to production, monitoring its performance, and deploying in real-time. By contrast, while we’ve seen explosive growth in the adoption of machine and deep learning (ML/DL) across industries, putting ML/DL models into production isn’t as well supported. During this talk, we’ll discuss the emerging patterns, state-of-the-art methods, and best practices leading companies are using to productionize ML/DL models.
What You'll Learn:
ML infrastructure and toolstacks are endlessly interesting and convoluted. Dillon has great clarity on macro trends within the infrastructure space while maintaining pragmatism about incorporating the latest open source tools.
Talk: AI in the Multiverse: Measuring ROI when A/B Tests are Not Possible
Abstract:
A.I. services are deployed to produce improvements to important business metrics, e.g. customer engagement, number of transactions, total profits. When evaluating the contribution of a new service, it is crucial to be able to answer the attribution question: how much of my target outcome would have been achieved even in the absence of the A.I. model? Causal assessments are usually done though A/B tests, which however are not always feasible: who would switch off Amazon recommendations entirely to do such an assessment? In this talk, we show how to use A.I. to assess A.I. contributions to revenues in eCommerce: in particular, we will show how deep learning models can be used to assess how much revenues in a digital shop comes from interactions with search and recommendation APIs. Our findings can be generalized to many other settings, to assess and monitor the performance of existing ML pipelines even in the absence of A/B testing.
What You'll Learn:
Attribution models for site search engines are stuck at "last-action" and Google Analytics-style reporting: since A/B testing the search bar is impossible, it is really hard to make informed business decisions involving the search experience. What would you do if you knew causation, not correlation, in the search behavior of your shoppers?
Talk: Scaling Global Models with Regional Data Strategies and Model Governance
Abstract:
AI-driven, including ML, models provide the capability to process a greater volume and variety of data to power new global platforms and products and to optimize global business operations. Given that the world and its data are ever more varied and dynamic, to take advantage of this power models need to be highly adaptable to represent the local diversity of events, people, markets, and operations. Models developed only with a global perspective can result in missing valuable insights, and potential harms from models that are biased in their results, or inadvertently exclude groups in society. This talk will outline the business imperative for robust and ethical model design and Mastercard's approach to leveraging a global data-strategy that sets the highest standards for the responsible use of data and AI though human-centered data-design while ensuring local compatibility and functionality through a regional approach to data sourcing and quality, model testing and governance, and internal data literacy. The benefits of scaling global models through regional data strategies will be illustrated with examples from fraud detection, credit decisioning, economic modeling, and understanding consumer preferences.
What You'll Learn:
How to set out an enterprise approach to responsible use of data and AI, how to translate that into global data strategy elements and frameworks and then how to use regional or country specific data and model building strategies
Talk: Customer Segmentation, Pricing, and Profit Optimization for international Banking
Abstract:
Background: Pricing is a famous business issue in many companies and organizations. The approach behind pricing analytics can be formulate as customer segmentation and constrained optimization problem in order to increasing sales and/or revenue.
Aim: Our main objectives is to design a pricing product that can help to:
1) Identify groups of elastic and inelastic customers,
2) Determine the optimal rate for each group of customers,
3) Be agonistic pipeline and can be reusable for other pricing use cases.
Methodology: We propose to use model based recursive partitioning (MOB) which use product characteristics and customer attributes as input and customer willingness to pay as output to segment customers. For each customer segmentation, we found the demand curve function and formulate the nonlinear optimization problem that maximize the sale or revenue using PYOMO and IPOPT.
Results: This pricing product has been used in three different countries, Peru, Coloumbia and Mexico in various products such as mortgage, SPL, and term deposit with great feedback. It helped Scotiabank to capture international banking customer behaviour and their price sensitivity more promptly .
What You'll Learn:
This is about applying cutting edge machine learning domain in the banking domain. As pricing is very critical, mainly companies do not reveal their methodology so google search will not help that much.
Partner & Co-Founder, INQ Data Law
Talk: Applied Machine Learning in Healthcare - Practical and Legal Considerations
Talk: Applied Machine Learning In Healthcare - Practical And Legal Considerations
Abstract:
Applied machine learning has the potential to transform healthcare, particularly in the areas of automation, prediction, and optimization. However, numerous challenges to the acquisition, storage, and utilization of data as well as the development of practical machine learning algorithms and change management principles need to be considered. This talk will provide an overview of the process of applying ML into healthcare and the legal and ethical considerations needed for data access and application.
What You'll Learn:
Attendees will gain an understanding of principles of knowledge translation in applied machine learning in healthcare and understand issues related to privacy and ethics as well as legal considerations.
Payments, Digital Banking and IT Managing Director, Banorte
In 2019 Francisco was appointed as Payments, Digital Banking and IT Managing Director. He has strong interest in new technology, innovation and delivery in the financial services sector with the main challenge of the digital transformation of the organization, products and services.
Previously, Francisco worked as the CIO focused on the development of an advanced organizational model that will allow management and operation of infrastructure and solutions, increase the capacity of project implementation and encourage innovation processes in the organization.
With close to thirty years of experience in the financial sector, he has developed his career in the Technology, Products, Payments, Digital Banking and Operations departments where he has successfully participated in numerous consulting projects, operational and organizational transformations, apps development, systems integration and infrastructure
Francisco has a degree of Chemical Engineering from Universidad Iberoamericana (Mexico) and has completed degrees in different Universities in Mexico, USA and Canada.
Francisco is married and has two children with whom he shares as much time as possible practicing sports.
Talk: Banorte’s AI Transformation Journey: How the Analytics Team of this Bank Yielded 3 Billion USD Revenue During the Past Five Years
Talk: Banorte's AI Transformation Journey: How the Analytics Team of this Bank Yielded 3 Billion USD Revenue During the Past Five Years.
Abstract:
There are high expectatios about AI initiatives across different industries in North America. However, too often results have been disappointing producing some backlash against digital transformation efforts. The capacity to implment and demonstrate high ROI AI projects changes this dynamic. This talk will delve into Banorte's transformation journey into an AI enhanced organization with data science projects yielding a net revenue that exceeds 3 billion USD during the past five years and avoiding the transformational fatigue.
What You'll Learn:
1. How to measure AI contribution to the bottom line
2. What are key prerequisites to focus yield high ROI on AI projects
3. Where to focus AI inititiatives to have a large organizational impact: revenue or cost?
Director of Machine Learning, Tubi
Jaya Kawale is the Director of Machine Learning at Tubi leading all of the machine learning efforts at Tubi encompassing homepage recommendations, content understanding and ads. Prior to Tubi, she has worked on different aspects of recommender systems at Netflix and Adobe research labs. She did her PhD from the University of Minnesota, Twin cities and her thesis won several awards including the Explorations in Science using computation award. She has published many top tier conference and journal papers.
Talk: Understanding Content Using Deep Learning for Natural Language Processing
Fellow The Paulson Institute
Matt Sheehan is a Fellow at the Paulson Institute’s think tank, MacroPolo, where he leads work on U.S.-China technology issues, with a specialization in artificial intelligence. His research maps and quantifies the key inputs to AI ecosystems globally. Matt is the author of the book The Transpacific Experiment: How China and California Collaborate and Compete for our Future.
From 2010-2016 Matt lived and worked in China, including as the first China correspondent for The World Post. In 2016 he returned to the Bay Area, working as an analyst and consultant on topics connecting China and California. In 2018, he was selected as a finalist for the Young China Watcher of the Year award.
Matt's research has been cited and he has been quoted in numerous media outlets, including The New York Times, The Wall Street Journal, The Financial Times, Reuters, The San Francisco Chronicle, and elsewhere.
Talk: Assessing China’s AI Capabilities
Principal Researcher, Microsoft Research
Rich Caruana is a Senior Principal Researcher at Microsoft. His focus is on intelligible/transparent modeling, machine learning for medical decision making, deep learning, and computational ecology. Before joining Microsoft, Rich was on the faculty in Computer Science at Cornell, at UCLA's Medical School, and at CMU's Center for Learning and Discovery. Rich's Ph.D. is from CMU. His work on Multitask Learning helped create interest in a subfield of machine learning called Transfer Learning. Rich received an NSF CAREER Award in 2004 (for Meta Clustering), best paper awards in 2005 (with Alex Niculescu-Mizil), 2007 (with Daria Sorokina), and 2014 (with Todd Kulesza, Saleema Amershi, Danyel Fisher, and Denis Charles), and co-chaired KDD in 2007 with Xindong Wu.
Talk: Friends Don’t Let Friends Deploy Black-Box Models: The Importance of Intelligibility in Machine Learning for Healthcare
Race Data Collection Expert, Toronto Police Service
Dr. Mai Phan is a data expert consultant currently supporting Toronto Police Service's ground-breaking anti-racism initiative. Mai works collaboratively with colleagues and partners to promote an evidence-based approach to anti-racism, human rights, and inclusion in public service organizations.
Mai was senior research/policy advisor at the Anti-Racism Directorate. She led the development and establishment of the Anti-Racism Data Standards and provided strategic advice and support to public sector organizations regulated to collect race-based data under the Anti-Racism Act. She contributed to the development of the Systemic Racial Barriers Identification and Removal Program to support advancement of workplace racial equity and inclusion within the Ontario Public Service.
Prior to that, as a Human Rights Advisor in the Ministry of Community Safety and Correctional Services, Mai supported initiatives to address systemic discrimination and remove barriers in employment and service delivery in correctional services.
Talk: Race Data As An Anti-Racism Tool
Founder & CEO, Beijing Rxthinking Inc.
Kan Deng PhD, graduated from Shanghai Jiaotong University with Bachelor and Master degree, then graduated from Carnegie Mellon University with Ph.D degree for Robotics and Machine Learning.
After graduation, Dr Deng worked with Oracle Inc as a Principal Architect for 6 years, worked with Telenav as the General Manager of its Beijing lab for 5 years, then worked with Baidu as a senior director in charge of its core business, web search engine.
In 2015, Dr Deng founded Beijing RxThinking Inc, applying deep reinforcement learning cutting-edge technology to solve healthcare problems.
RxThinking has been collecting more than 800 million of Electronic Health Records from top hospitals in China. We translate those 800 million EHRs into structured clinical routes, one by one, so that we have 800 million structure clinical routes. After then we compress them together to assemble a medical map.
With the medical map, we develop AI doctor assistant. Our AI doctor assistant can answer patients’ queries, recommend what-to-do-next to the community doctors, supervise the quality and cost for the hospitals, and improve the doctor's productivity hundreds of times especially for online remote diagnosis.
During the pandemic, our product is widely used in China serving over 200 million people, including the citizens in Wuhan. And we are awarded as the top 10 of the 100 best practices of AI medical solutions, by China Academy of Information and Communications Technology (CAICT) in April 2020.
Talk: AI. MD in the Time of Corona Virus in China
Professor of Computer Science, Electrical and Computer Engineering and Statistical Science, Duke University
Cynthia Rudin is a professor of computer science, electrical and computer engineering, and statistical science at Duke University. Previously, Prof. Rudin held positions at MIT, Columbia, and NYU. Her degrees are from the University at Buffalo and Princeton University. She is a three time winner of the INFORMS Innovative Applications in Analytics Award, was named as one of the "Top 40 Under 40" by Poets and Quants in 2015, and was named by Businessinsider.com as one of the 12 most impressive professors at MIT in 2015. She has served on committees for INFORMS, the National Academies, the American Statistical Association, DARPA, the NIJ, and AAAI. She is a fellow of both the American Statistical Association and Institute of Mathematical Statistics.
Talk: Interpretability vs. Explainability in Machine Learning
Talk: Understanding Content Using Deep Learning for Natural Language Processing
Abstract:
Tubi is an advertiser based video on demand service that allows its users to watch content online. For a lot of the content, there is a large amount of textual data in the form of user reviews, synopsis, title plots and even Wikipedia. Furthermore, there is a large amount of metadata in the form of actors, ratings, year of release, studio, etc. In this talk, I will present some of the challenges in understanding the data and present our platform for content understanding.
What You'll Learn:
Content understanding, deep learning for natural language processing, challenges in an industrial setting
Talk: Assessing China’s AI Capabilities
Abstract:
The talk will give an overview of China's AI/ML ecosystem, and a deep dive into its capabilities when it comes to leading edge research in neural networks.
What You'll Learn:
You'll learn about China's role in the global flows of AI research talent, and what implications this has for government policy in the US, Canada and Europe.
Talk: Friends Don’t Let Friends Deploy Black-Box Models: The Importance of Intelligibility in Machine Learning for Healthcare
Abstract:
In machine learning often a tradeoff must be made between accuracy and intelligibility: the most accurate models usually are not very intelligible, and the most intelligible models usually are less accurate. This often limits the accuracy of models that can safely be deployed in mission-critical applications such as healthcare where being able to understand, validate, edit, and ultimately trust a model is important. We have developed a learning method that is as accurate as full complexity models such as boosted trees and random forests, but even more intelligible than linear models. This makes it easy to understand what a model has learned and to edit the model when it learns inappropriate things. Making it possible for medical experts to understand and repair a model is critical because most clinical data has unexpected problems. I’ll present several healthcare case studies where these high-accuracy GAMs discover surprising patterns in the data that would have made deploying a black-box model risky, and also allow us to learn important new insights from our healthcare data.
What You'll Learn:
1) The risk of using machine learning in healthcare when you can't understand what the model is learned.
2) How new methods in intelligible machine learning can help mitigate this risk.
3) The amazing things we can learn about healthcare by using this kind of model on medical data.
Talk: Race Data As An Anti-Racism Tool
Abstract:
Race is a concept, a tool, and a structure that defines a set of relationships between people. We will unpack the idea of race as relationships and race as data in its historical and current contexts. We will discuss what it means to build equity into data practices and what dismantling systemic racism can look like in technology (and the pitfalls to avoid).
What You'll Learn:
You will learn about and better understand what systemic racism is, the historical legacy of race data and how to challenge and question data practices for a more equitable society.
Talk: AI. MD in the Time of Corona Virus in China See Abstract
Abstract:
The practice of apply machine learning technology in healthcare, especially to deal with corona virus pandemic.
What You'll Learn:
The challenges, the solutions, the effectiveness, and the remaining issues, including technology progress and institution reform.
Talk: Interpretability vs. Explainability in Machine Learning
Abstract:
With widespread use of machine learning, there have been serious societal consequences from using black box models for high-stakes decisions, including flawed bail and parole decisions in criminal justice. Explanations for black box models are not reliable, and can be misleading. If we use interpretable machine learning models, they come with their own explanations, which are faithful to what the model actually computes.
What You'll Learn:
You will learn that there is a chasm between explaining black box models and using inherently interpretable models. You may also find my experience helpful, which is that we have never needed a black box model for a high stakes decision, because we have always been able to construct an interpretable model that is at the same level of predictive performance as the best black box we could find.
AI Research Scientist, NASA Jet Propulsion Laboratory
Shreyansh Daftry is a Research Scientist at NASA Jet Propulsion Laboratory (JPL) in Pasadena, California, working at the intersection of Artificial Intelligence and Space Technology to help develop the next generation of robots for Earth, Mars and beyond. Shreyansh received his M.S. degree in Robotics from the Robotics Institute, Carnegie Mellon University, USA, and his B.S. degree in Electronics and Communications Engineering from Manipal University, India. His research interests spans computer vision, machine learning and autonomous robotics, with a focus on real-time computation, safety and adaptability. Shreyansh has received numerous awards and honors, including the NASA Space Act Award and the JPL Software of the Year award, for his contributions to the field of space and aeronautics.
Talk: Machine Learning for Space Exploration
Product Manager, Google AI
Gaurav is a Product Manager at Google AI. He focuses on bringing cutting-edge Conversation AI technologies from research to production. He is also a product advisor for AI startups in Google's internal startup incubator, Area120. Prior to that, he led Payments Trust & Safety efforts at Google for the APAC region and co-founded a FinTech startup Novus Minds in India. He holds a degree in Computer Science from IIT Roorkee in India.
Talk: Fine-Grained Emotion Detection for Products & Research
Talk: Machine Learning for Space Exploration
Abstract:
Artificial Intelligence is playing an increasing role in the space industry, where AI related technologies such as machine learning have the potential to revolutionize almost every aspect of space exploration. In this talk, we will discuss the evolution of autonomous robots for space exploration and planetary science. Next we will look at examples of machine learning technologies we are developing for autonomous robotic applications on Earth, Mars and beyond, and describe some of the grand challenges in AI for such safety-critical systems. Finally, we describe lessons learnt from space industry that can be applied to industrial applications here on Earth.
What You'll Learn:
- How AI/ML is being used by NASA to enable the next frontier in robotics space exploration
Talk: Fine-Grained Emotion Detection for Products & Research
Abstract:
Understanding emotion expressed in language has a wide range of applications, from building empathetic chatbots to detecting harmful online behavior. In this talk, we will present our work at Google AI Research towards building GoEmotion, a large-scale dataset containing 58K social media comments labeled with a fine-grained emotion taxonomy, which is adaptable to multiple downstream tasks. We will share results demonstrating generalizability towards existing emotion benchmarks from other domains. Lastly, we will share how organizations could use this dataset to train custom models for their use cases.
What You'll Learn:
1. How emotions can be detected from textual content for business use cases & research purposes
2. Details about data for training own models
Data Scientist, Scotiabank
Navid is an applied research scientist with a master's degree in computer science from the University of Toronto. His ultimate objective is to apply cutting-edge researches and latest breakthroughs to solve real-world issues. He enjoys challenging himself to solve problems that we don’t necessarily know if there’s, yet, a solution for
Talk: Customer Segmentation, Pricing, and Profit Optimization for international Banking
Talk: Customer Segmentation, Pricing, and Profit Optimization for international Banking
Abstract:
Background: Pricing is a famous business issue in many companies and organizations. The approach behind pricing analytics can be formulate as customer segmentation and constrained optimization problem in order to increasing sales and/or revenue.
Aim: Our main objectives is to design a pricing product that can help to:
1) Identify groups of elastic and inelastic customers,
2) Determine the optimal rate for each group of customers,
3) Be agonistic pipeline and can be reusable for other pricing use cases.
Methodology: We propose to use model based recursive partitioning (MOB) which use product characteristics and customer attributes as input and customer willingness to pay as output to segment customers. For each customer segmentation, we found the demand curve function and formulate the nonlinear optimization problem that maximize the sale or revenue using PYOMO and IPOPT.
Results: This pricing product has been used in three different countries, Peru, Coloumbia and Mexico in various products such as mortgage, SPL, and term deposit with great feedback. It helped Scotiabank to capture international banking customer behaviour and their price sensitivity more promptly .
What You'll Learn:
This is about applying cutting edge machine learning domain in the banking domain. As pricing is very critical, mainly companies do not reveal their methodology so google search will not help that much.
Data Scientist, Scotiabank
Cheng is a PhD of economist turned data scientist. With 6+ years of experience in conducting empirical research in applied microeconomics, she is proficient in economic modelling and machine learning techniques
Talk: Customer Segmentation, Pricing, and Profit Optimization for international Banking
Talk: Customer Segmentation, Pricing, and Profit Optimization for international Banking
Abstract:
Background: Pricing is a famous business issue in many companies and organizations. The approach behind pricing analytics can be formulate as customer segmentation and constrained optimization problem in order to increasing sales and/or revenue.
Aim: Our main objectives is to design a pricing product that can help to:
1) Identify groups of elastic and inelastic customers,
2) Determine the optimal rate for each group of customers,
3) Be agonistic pipeline and can be reusable for other pricing use cases.
Methodology: We propose to use model based recursive partitioning (MOB) which use product characteristics and customer attributes as input and customer willingness to pay as output to segment customers. For each customer segmentation, we found the demand curve function and formulate the nonlinear optimization problem that maximize the sale or revenue using PYOMO and IPOPT.
Results: This pricing product has been used in three different countries, Peru, Coloumbia and Mexico in various products such as mortgage, SPL, and term deposit with great feedback. It helped Scotiabank to capture international banking customer behaviour and their price sensitivity more promptly .
What You'll Learn:
This is about applying cutting edge machine learning domain in the banking domain. As pricing is very critical, mainly companies do not reveal their methodology so google search will not help that much.
Software Engineer, Google Research
Dana Movshovitz-Attias is a Staff Software Engineer and Researcher at Google Research, where she leads an NLP research group focused on Conversational AI, Graph ML, and Efficient ML Computation. She has over 15 years of experience in Natural Language Processing and Machine Learning. Her work has been published in top-tier conferences, incl. ACL, NeurIPS, and NAACL, and appeared in the press.
Dana received a Ph.D. in Computer Science from Carnegie Mellon University in 2015 and a M.S. and B.S. in Computer Science and Computational Biology from the Hebrew University of Jerusalem, Israel in 2010 and 2007.
Talk: Fine-Grained Emotion Detection for Products & Research
Talk: Fine-Grained Emotion Detection for Products & Research
Abstract:
Understanding emotion expressed in language has a wide range of applications, from building empathetic chatbots to detecting harmful online behavior. In this talk, we will present our work at Google AI Research towards building GoEmotion, a large-scale dataset containing 58K social media comments labeled with a fine-grained emotion taxonomy, which is adaptable to multiple downstream tasks. We will share results demonstrating generalizability towards existing emotion benchmarks from other domains. Lastly, we will share how organizations could use this dataset to train custom models for their use cases.
What You'll Learn:
1. How emotions can be detected from textual content for business use cases & research purposes
2. Details about data for training own models
Applied Research Scientist, Georgian
Azin Asgarian is currently an Applied Research Scientist on Georgian’s R&D team where she works with companies to help adopt applied research techniques to overcome business challenges. Azin holds a Master of Science in Computer Science from University of Toronto and a Bachelor of Computer Science from University of Tehran. Prior to joining Georgian, Azin was a research assistant at the University of Toronto and part of the Computer Vision Group where she was working on the intersection of Machine Learning, Transfer Learning, and Computer Vision. Due to her interest in HealthCare, she has worked on various healthcare projects as a research assistant at University Health Network (UHN).
Talk: Overcoming the Cold Start Problem: How to Make New Tasks Tractable
Talk: Overcoming the Cold Start Problem: How to Make New Tasks Tractable
Abstract:
In recent years, fuelled by the advances in supervised machine learning, we have seen astonishing leaps in the application of deep neural networks. Despite the remarkable results, these models are data-hungry and their performance relies heavily on the quality and size of the training data. In real-world scenarios, this can increase the time to value add significantly for businesses as collecting huge amounts of labeled data is usually very time and cost consuming. This phenomenon—known as the cold start problem—is a pain point for almost any AI company that wants to scale. In this talk, we demonstrate how this problem can be addressed by aggregating data across sources and leveraging previously trained models.
What You'll Learn:
In this talk, you will see real examples of the cold start problem and how it can prevent businesses from effectively and efficiently growing. You will learn about various machine learning methods that can be used to address this problem.
Research Lead, Tractable
Fran Kirschner is a Research Lead at Tractable. She develops Tractable’s deep learning algorithms, and focuses on diversifying and scaling the core AI across domains. Franziska started life as a physicist, and completed her PhD in condensed matter physics at the University of Oxford.
Talk: Overcoming the Cold Start Problem: How to Make New Tasks Tractable
Talk: Overcoming the Cold Start Problem: How to Make New Tasks Tractable
Abstract:
In recent years, fuelled by the advances in supervised machine learning, we have seen astonishing leaps in the application of deep neural networks. Despite the remarkable results, these models are data-hungry and their performance relies heavily on the quality and size of the training data. In real-world scenarios, this can increase the time to value add significantly for businesses as collecting huge amounts of labeled data is usually very time and cost consuming. This phenomenon—known as the cold start problem—is a pain point for almost any AI company that wants to scale. In this talk, we demonstrate how this problem can be addressed by aggregating data across sources and leveraging previously trained models.
What You'll Learn:
In this talk, you will see real examples of the cold start problem and how it can prevent businesses from effectively and efficiently growing. You will learn about various machine learning methods that can be used to address this problem.
Technology Advisor, The United Nations
Danit Gal is Technology Advisor at the United Nations, leading work on AI in the implementation of the United Nations Secretary-General's Roadmap for Digital Cooperation. She is interested in technology ethics, geopolitics, governance, safety, and security. Previously, she was Project Assistant Professor at the Cyber Civilizations Research Center at Keio University in Tokyo, Japan. Danit serves as the former chair and vice chair of the P7009 IEEE standard on the Fail-Safe Design of Autonomous and Semi-Autonomous Systems, and the executive committee of The IEEE Global Initiative on Ethics of Autonomous and Intelligent Systems. She is an Associate Fellow at the Leverhulme Centre for the Future of Intelligence at the University of Cambridge, and Visiting Research Fellow at the S. Rajaratnam School of International Studies at the Nanyang Technological University.
Talk: The State of AI/ML at the United Nations
Talk: The State of AI/ML at the United Nations
Abstract:
This talk introduces relevant work on AI/ML at the United Nations to provide an overview of key developments and use cases and highlight opportunities for collaboration.
What You'll Learn:
You'll learn about exciting applications of AI/ML at the United Nations and learn about opportunities to collaborate.
Professor, Department of IEOR, Columbia University
Xunyu Zhou is the Liu Family Professor of Industrial Engineering and Operations Research at Columbia University in New York. He was the Nomura Professor of Mathematical Finance, the Director of Nomura Center for Mathematical Finance, and the Director of Oxford-Nie Financial Big Data Lab at the University of Oxford during 2007-2016 before joining Columbia.
His research includes reinforcement learning in continuous time and spaces, quantitative behavioral finance models that incorporate human emotions and psychology into financial decision makings, and intelligent wealth management solutions using stochastic control and machine learning techniques.
Professor Zhou is known for his work in indefinite stochastic LQ control theory and application to dynamic mean-variance portfolio selection, in asset allocation and pricing under cumulative prospect theory, and in general time-inconsistent problems. He directs the Nie Center for Intelligent Asset Management, a research center funded by a FinTech company, at Columbia. He has addressed the 2010 International Congress of Mathematicians, and has been awarded the Wolfson Research Award from The Royal Society (UK), the Outstanding Paper Prize from the Society for Industrial and Applied Mathematics, the Humboldt Distinguished Lecturer, the Alexander von Humboldt Research Fellowship, and the Archimedes Lecturer at Columbia. He is both an IEEE Fellow and a SIAM Fellow.
Professor Zhou received his Ph.D. in Operations Research and Control Theory from Fudan University in China in 1989.
Talk: Reinforcement Learning via Stochastic Control
Head of Software & Algorithms, Xanadu Quantum Technologies
Nathan Killoran is the Head of Software & Algorithms at Xanadu, and one of the founding developers of PennyLane, the world’s leading quantum machine learning software library. Nathan steers Xanadu’s open-source quantum software products and leads algorithm research in photonics and quantum machine learning. Nathan holds a PhD in Physics from the University of Waterloo, with expertise in quantum computing, deep learning, and quantum optics.
Talk: Software for Quantum Machine Learning
Talk: Reinforcement Learning via Stochastic Control
Abstract:
While most existing reinforcement learning (RL) research is in the framework of Markov Decision Processes (MDPs), it is important and indeed necessary, both theoretically and practically, to consider RL in continuous time with continuous feature and action spaces, for which stochastic control theory offers a natural underpinning. The related research is still in its infancy, and this talk reports some of the latest developments and suggests several directions for investigation.
What You'll Learn:
Theoretical foundation and interpretation of some of the commonly used heuristics in reinforcement learning such as entropy regularization and Gibbs/Boltzmann/Gaussian exploration.
Talk: Software for Quantum Machine Learning
Abstract:
One of the fundamental goals in the emerging field of quantum machine learning is to build trainable quantum computing algorithms. It turns out that we can, with very minimal changes, port many existing ideas, algorithms, and training strategies from deep learning over to the quantum domain. This allows us to train quantum computers in largely the same way as we do neural networks, even using familiar software tools like TensorFlow and PyTorch. In this talk, I will give a high-level overview of the key ideas that make this possible.
What You'll Learn:
You'll learn how to use software tools like PennyLane, TensorFlow, and PyTorch to train quantum computers!
Project Manager at Applied AI Project, Vector Institute
Sedef Akinli Kocak is the Project Manager at Vector Institute for Artificial Intelligence engaging Vector sponsors on AI projects. Currently, she leads several multi-industrial participant projects. She holds a Ph.D. degree from the Data Science Lab at Ryerson University, Canada, and earned her master’s degrees in both Chemical Engineering and Business of Administration. She worked in data-intensive R&D project development and academic-industry partnerships in the area of AI/ML at SOSCIP, the Southern Ontario Smart Computing for Innovation Platform. She is also an experienced and accomplished researcher in the area of ICT for sustainability and sustainability design in software-intensive systems and a part-time Data Science and Analytics lecturer and supervisor at Ryerson University since 2014. She served as a member of the Compute Ontario Board Advisory Committee and AI program development advisor at the Continuing Education, University of Toronto.
Talk: Harnessing the Power of NLP: A Vector Institute Industry Collaborative Project
Talk: Harnessing the Power of NLP: A Vector Institute Industry Collaborative Project
Abstract:
The Vector Institute’s project, Recreation of Large Scale Pre-Trained Language Models (the NLP Project), is an industry-academia collaboration that explores how state-of-the-art natural language processing (NLP) models could be applied in business and industry settings at scale. Developing and employing NLP models in industry has become progressively more challenging as model complexity increases, data sets grow in size, and computational requirements rise. These hurdles limit the accessibility many organizations have to NLP capabilities, putting the significant benefits advanced NLP can provide out of reach. The NLP Project addressed these challenges by familiarizing industry participants with advanced NLP techniques and the workflows for developing new methods that could achieve high performance while using relatively small data sets and widely accessible computing resources. The project involved 60 participants: 23 Vector researchers and staff with expertise in machine learning and NLP along with 37 industry technical professionals from 16 Vector sponsor companies. The participants established 11 working groups, each of which developed and performed experiments relevant to existing industry needs.
In this talk, I will provide an overview of the NLP project and share how industry participants gained practical knowledge through pre-training large scale language models, learned theoretical concepts from leading NLP practitioners, and broadened their professional network through collaborations with participating sponsors. I will share some of the technical challenges that we encountered throughout the project and how we overcome them. Finally, I will offer best practices to guide future industry collaborative projects.
What You'll Learn:
How Vector Institute's industry collaborations help sponsors in adoption of AI advances, specifically in the NLP domain.
Research Associate, National Research Council of Canada
Rebecca Knowles is a Research Associate at the National Research Council of Canada. Her research focuses on machine translation and computer-aided translation. She holds a Ph.D. in computer science from Johns Hopkins University, where she was a National Science Foundation Graduate Research Fellow.
Talk: Indigenous Language Technologies: Neural Machine Translation for Inuktitut
CEO, AI For the People
Talk:
Postdoctoral Researcher, University of Toronto
Alba Cervera Lierta did her doctoral studies in entanglement applications in quantum information at the University of Barcelona. She also worked on quantum algorithms for near-term applications at the Barcelona Supercomputing Center. Her background includes particle physics phenomenology, multipartitie entanglement and quantum information. She is currently a postdoctoral fellow at the Alán Aspuru-Guzik group at the University of Toronto. She is working on variational quantum algorithms and computational tools for quantum simulation.
Talk: The Quest for the Final Quantum Computer
Product Manager, Google Research
Talk: Everything You Need to Know About New Libraries in the Keras Ecosystem
Professor of Chemistry and Computer Science, University of Toronto
Alán Aspuru-Guzik’s research lies at the interface of computer science with chemistry and physics. He works in the integration of robotics, machine learning and high-throughput quantum chemistry for the development of materials acceleration platforms. These “self-driving laboratories¨ promise to accelerate the rate of scientific discovery, with applications to clean energy and optoelectronic materials. Alán also develops quantum computer algorithms for quantum machine learning and has pioneered quantum algorithms for the simulation of matter. He is jointly appointed as a Professor of Chemistry and Computer Science at the University of Toronto. Alán is a faculty member of the Vector Institute for Artificial Intelligence. Previously, Alán was a full professor at Harvard University where he started his career in 2006. Alán is currently the Canada 150 Research Chair in Quantum Chemistry as well as a CIFAR AI Chair at the Vector Institute. Amongst other awards, Alán is a recipient of the Google Focused Award in Quantum Computing, the MIT Technology Review 35 under 35, and the Sloan and Camille and Henry Dreyfus Fellowships. Alán is a fellow of the American Association of the Advancement of Science and the American Physical Society. He is a co-founder of Zapata Computing and Kebotix, two early-stage ventures in quantum computing and self-driving laboratories respectively.
Talk: Artificial Intelligence for Molecular Design and Self Driving Labs
Lead Quantum Applications, Zapata Computing Inc.
Alejandro did his graduate studies, M.A and Ph.D. in Chemical Physics, at Harvard University. Over the past 12+ years, he has explored the computational limits and opportunities of quantum computing for real-world applications. Before joining Zapata Computing as a Senior Quantum Scientist and Quantum Applications Lead, Alejandro was the lead scientist of the Quantum Machine Learning effort at NASA's Quantum Artificial Intelligence Laboratory (NASA QuAIL) where he worked for 5+ years. He was also the Co-Founder of Qubitera LLC, a consulting company acquired by Rigetti Computing where he worked after NASA and before his current appointment with Zapata Computing. His latest research involves the design of hybrid quantum-classical algorithms to solve hard optimization problems and intractable machine learning subroutines.
Talk: Quantum - Assisted Machine Learning with Near-Term Quantum Devices
Researcher, Microsoft Research
Marco Tulio Ribeiro is a Senior Researcher at Microsoft Research. His work is on facilitating the communication between humans and machine learning models, which includes interpretability, trust, debugging, feedback, robustness, testing, etc. He received his PhD from the University of Washington.
Talk: Productionizing Deep Learning Models at Scale
Advisor, Element AI
Richard is an advisor to companies, start-ups, and policy-makers on AI strategy and governance. He has applied this on-the-ground knowledge of how AI is transforming organizations and the economy as an expert participant in many forums investigating the broader social impact of the technology, including the Brookfield Institute, the Federal Economic Strategy Table for Digital Industries, and the Partnership on AI. He holds an M.B.A., two Law degrees, and a BSc in Cognitive Science.
Talk: Algorithmic Decision Making: Exploring Practical Approaches to Liability, Fairness, and Explainability
Data Scientist, TD Bank
Talieh Tabatabaei holds B.Eng and MASc degrees in Electrical and Computer Engineering. She has more than 8 years of working experience in the field of Machine Learning and Artificial Intelligence in high-level academic research, teaching, and industry, with several publications in this field.
Talieh is currently working as a data scientist at TD Bank.
Talk: Algorithmic Decision Making: Exploring Practical Approaches to Liability, Fairness, and Explainability
Principal Scientist, bnh.ai
Patrick Hall is the principal scientist at bnh.ai, a boutique law firm focused on AI and data analytics. He also serves as a visiting assistant professor of decision sciences at the George Washington University School of Business and as an advisor to select machine learning startups. Before co-founding bnh.ai, Patrick led responsible AI efforts at H2O.ai. His work at H2O resulted in one of the world's first commercial solutions for explainable and fair machine learning. Among other academic and technology media writing, Patrick is the primary author of popular e-books on explainable and responsible machine learning.
Patrick has also held global customer-facing and R&D roles at SAS Institute, where he authored multiple patents in automated market segmentation using novel clustering methods and deep learning. During these years, he became the 11th person worldwide to become a Cloudera certified data scientist. Patrick studied computational chemistry at the University of Illinois before graduating from the Institute for Advanced Analytics at North Carolina State University.
Talk: Algorithmic Decision Making: Exploring Practical Approaches to Liability, Fairness, and Explainability
Researcher, MILA
Sasha Luccioni is a researcher working with Yoshua Bengio at the Mila AI institute to fight climate change using Artificial Intelligence. She leads projects at the nexus of AI and climate change, using generative networks to visualize the consequences of climate change and tracking the carbon footprint of AI. In the past, she worked in the finance sector, but decided to follow her heart and leave a top Wall Street company to use her skills in AI to make the world a better place. She is highly involved in community initiatives, serving on the Advisory board of Kids Code Jeunesse and as a chair of the Climate Change AI initiative.
Talk: The Role of ML in Climate Change
Head of Location Intelligence, RBC
Arthur Berrill is the Head of Data Science at the Royal Bank of Canada. His charter is to research, guide and deliver data science capabilities including location intelligence, new data content, artificial intelligence, ontology and climate change studies across all departments of the bank.
Arthur is an RBC Distinguished Technologist.
Arthur has more than 40 years of experience in the architecture, design and development of enterprise spatial systems including WILDMAP (a GIS before the term existed), SYSTEM 9, SpatialWare®, MapInfo products and Location Hub®. He holds numerous key and valuable patents in the location intelligence and spatial systems domain.
Fascinated by the use of new technology to solve business challenges, Arthur is working on future data strategies, inventing new or improved algorithms and methods and helping businesses see technology opportunities as or before they emerge.
Prior to RBC, Arthur led the location intelligence initiative at Scotiabank and before then was president of DMTI Spatial. Arthur was Inventor of the Year for 2008 at Pitney Bowes (MapInfo) and won the TechAmerica 50th Anniversary Innovation Award in 2009.
Arthur is a graduate with Honours from the University of Queensland and did his postgraduate work at the International Institute of Aerial Survey and Earth Sciences in the Netherlands.
Talk: The Role of ML in Climate Change
Senior Vice President, Enterprise Operations and Payments, RBC
Jules Andrew has held leadership financial and operational positions across the globe with IBM and RBC for over 20 years
Jules was appointed Senior Vice President, Strategy, Transformation and Enterprise Operations, Technology and Operations, RBC, in February 2020. In this role, she is responsible for developing and executing T&O’s strategy in support of RBC’s goal to be a digitally-enabled, relationship bank and transforming shared services, including finance, to enhance RBC’s industry-leading position. Jules’ team facilitates money movement for clients, suppliers and employees and ensures branches and ATMs have cash.
Jules is the current chair of the T&O Diversity Leadership Council where she supports T&O employees to be champions, advocates and examples of diversity, inclusion & belonging at RBC. Working with organizations such as Women in Communications and Technology, Canadian Women in Business and Girls Who Code, Jules actively promotes the importance of encouraging women to pursue STEM careers and leadership roles. She is also responsible for several hackathons and innovation challenges encouraging diversity and community involvement for students globally. She holds a Master’s of Science in Management from Purdue University. Jules lives in Toronto and loves to travel – having lived in 5 countries and visited 58.
Talk: The Role of ML in Climate Change
Director Of Engineering, Uber
Talk: Autonomous Vehicles - The Next Step Forward
Founder, X-Matik Inc.
Talk: Autonomous Vehicles - The Next Step Forward
Founder & CEO, AutoX, Inc.
Talk: Autonomous Vehicles - The Next Step Forward
Lead Data Scientist, Sun Life
Jennifer is the Lead Data Scientist at Sun Life Financial’s Analytics Centre of Excellence, helping the company to build intelligent data solutions to better serve their clients. Her past experience in the field includes the Globe and Mail, Scribd and Slyce. She holds a Master’s in Machine Learning from University College London and a B. Math from the University of Waterloo. Jennifer is a strong proponent of gender diversity in her field and partners with the University of Waterloo to support young females pursuing careers in STEM.
Talk: Forget ROC scores, What Metrics Do your Stakeholders Care About?
Leader of Machine Learning Team, Cyclica
Ali leads the machine learning team at Cyclica Inc focusing on improving the company's technology for predicting interaction between drugs and target proteins. As a computational biologist and machine learning specialist, Ali has worked on a series of scientific articles in high impact scientific journals and international conferences covering such fields as transfer learning and unsupervised clustering. He earned his Ph.D from the University of Toronto, and master of mathematics degree from the University of Waterloo.
Talk: Deep Learning Across Label Confidence Distribution via Transfer Learning
Talk: Beyond Accuracy: Behavioral Testing of NLP Models with CheckList
Abstract:
What You'll Learn:
Talk: Quantum - Assisted Machine Learning with Near-Term Quantum Devices
Abstract:
With quantum computing technologies nearing the era of commercialization and quantum advantage, machine learning (ML) has been proposed as one of the promising killer applications. Despite significant effort, there has been a disconnect between most quantum ML proposals, the needs of ML practitioners, and the capabilities of near-term quantum devices towards a conclusive demonstration of a meaningful quantum advantage in the near future. In this talk, we provide concrete examples of intractable ML tasks that could be enhanced with near-term devices. We argue that to reach this target, the focus should be on areas where ML researchers are struggling, such as generative models in unsupervised and semi-supervised learning, instead of the popular and more tractable supervised learning tasks. We will discuss recent experimental implementations of these quantum generative models, in both, superconducting-qubit and ion-trap quantum computers.
What You'll Learn:
Exciting directions and opportunities for assisting machine learning with quantum computers.
Talk: Everything You Need to Know About New Libraries in the Keras Ecosystem
Abstract:
What You'll Learn:
Talk: The Quest for the Final Quantum Computer
Abstract:
In this talk, I will overview the basic concepts of quantum computing and its applications. I will present what are the state-of-the-art quantum algorithms, its advantages and limitations. Finally, I will explain the state of development of experimental quantum computers and future prospects.
What You'll Learn:
The current state of quantum computation;
Talk: Indigenous Language Technologies: Neural Machine Translation for Inuktitut
Abstract:
Recent advances in machine translation have resulted in systems of very high quality, but only for a very limited set of the world’s more than 7000 languages. This talk provides a brief overview of Indigenous language technology projects at the National Research Council of Canada, before focusing on one project in particular: the development of neural machine translation systems to translate between Inuktitut and English. We will discuss challenges, applications of state of the art models, and future use cases.
What You'll Learn:
Neural machine translation, applications of machine learning to Indigenous languages, challenges of domain adaptation in low-resource settings
Talk: Algorithmic Decision Making: Exploring Practical Approaches to Liability, Fairness, and Explainability
Abstract:
What You'll Learn:
Talk: Algorithmic Decision Making: Exploring Practical Approaches to Liability, Fairness, and Explainability
Abstract:
What You'll Learn:
Talk: Algorithmic Decision Making: Exploring Practical Approaches to Liability, Fairness, and Explainability
Abstract:
What You'll Learn:
Talk: The Role of ML in Climate Change
Abstract:
What You'll Learn:
Talk: The Role of ML in Climate Change
Abstract:
What You'll Learn:
Talk: The Role of ML in Climate Change
Abstract:
What You'll Learn:
Talk: Autonomous Vehicles - The Next Step Forward
Abstract:
What You'll Learn:
Talk: Autonomous Vehicles - The Next Step Forward
Abstract:
What You'll Learn:
Talk: Autonomous Vehicles - The Next Step Forward
Abstract:
What You'll Learn:
Talk: Forget ROC scores, What Metrics Do your Stakeholders Care About?
Abstract:
How do you get buy-in from leadership to sponsor your ML project? How do you convince your stakeholders to put your ML models into production? And finally, how do you communicate the ROI of the project once it’s been deployed?
While these are questions universal to any industry, they are particularly challenging to answer in the insurance industry because of its highly regulated and risk-averse nature. As such, more creative thinking is needed to convince stakeholders that your ML solutions can be trusted and bring value.
What You'll Learn:
In this talk, we will share lessons we learned in answering three questions and the metrics stakeholders care about. The talk is designed so that those managing projects (e.g., data science directors/managers) and those executing the work (e.g., data scientists/analysts) can walk away with tips to help their ML projects start and close off successfully.
Talk: Deep Learning Across Label Confidence Distribution via Transfer Learning
Abstract:
Performance of neural network models relies on the availability of large datasets with minimal levels of uncertainty. However, in aggregated data environments, confidence in the individual data points vary in a quantifiable manner by primary data source or measurement type. Differences in label confidence make model building challenging, as the optimization cannot be done while amalgamating all the data points in the training process. In this work, we apply a transfer learning approach to improve predictive power in noisy data systems with large variable confidence datasets. We propose a deep neural network approach called Filtered Transfer Learning (FTL) that defines multiple tiers of data confidence as separate tasks in a transfer learning setting. The deep neural network is fine-tuned in a hierarchical process by iteratively removing (filtering) data points with lower label confidence, and retraining. In this report we use FTL for predicting the interaction of drugs and proteins. We demonstrate that using FTL to learn stepwise, across the label confidence distribution, results in higher performance compared to deep neural network models trained on a single confidence range.
The challenge of mixed confidence training data is not restricted to the domain of protein and drug interaction; in practice, data labeling is done based on either computational algorithms or human experts (or even non-experts), and neither approach is perfect. Other examples with differences in data point label confidence include: radiological or histopathological images or image segment labels, and measured resistance to cancer drugs. We anticipate that FTL will enable the machine learning community to benefit from large datasets with uncertain labels in fields such as biology and medicine.
What You'll Learn:
How to deal with data points with different levels of confidence in deep learning setting
Co-Founder & COO, Knowtions Research
Christina Cai is the Co-founder and COO of Knowtions Research, an applied artificial intelligence company developing a new generation of Pay-How-you-Live Insurance products where everyone can own their health risks and be insured. The company’s patentable Lydia AI risk prediction engine leverages big data to help health insurers protect customer health and grow insurable base. Christina leads at the intersection of operations, finance and culture building to lay a strong foundation of people, investors and industry advisors to propel Knowtion’s growth. Knowtions is venture-backed by Alibaba Entrepreneurs Fund and Information Venture Partners.
Talk: Forget ROC scores, What Metrics Do your Stakeholders Care About?
Talk: Forget ROC scores, What Metrics Do your Stakeholders Care About?
Abstract:
How do you get buy-in from leadership to sponsor your ML project? How do you convince your stakeholders to put your ML models into production? And finally, how do you communicate the ROI of the project once it’s been deployed?
While these are questions universal to any industry, they are particularly challenging to answer in the insurance industry because of its highly regulated and risk-averse nature. As such, more creative thinking is needed to convince stakeholders that your ML solutions can be trusted and bring value.
What You'll Learn:
In this talk, we will share lessons we learned in answering three questions and the metrics stakeholders care about. The talk is designed so that those managing projects (e.g., data science directors/managers) and those executing the work (e.g., data scientists/analysts) can walk away with tips to help their ML projects start and close off successfully.
Director of Machine Learning, Rogers Communications
Shahid Amlani is the Director of Machine Learning and Automation at Rogers Communications. His work focuses on leveraging the power of machine learning to enhance the digital customer experience – solving problems for customers and driving tangible results. Shahid has spent over a decade creating solutions that utilize the potential of technology and data to create real, measurable business outcomes. He holds a degree in computer science from the University of Toronto and an MBA from the Ivey School of Business.
Talk: Predicting Which Customers Will Experience a Technical Issue Tomorrow and Will Call as A Result
Talk: Predicting Which Customers Will Experience a Technical Issue Tomorrow and Will Call as A Result
Abstract:
Large telecom providers (and many other industries) spend tens of millions of dollars each year reacting to customer issues. This generally takes the form of large call center and repair technician workforces that are waiting for an issue to happen, in order to help solve it.
Utilizing machine learning and the power of robotic process automation (RPA), we have set out to determine a way to predict which customers are going to reach out with an issue, before they actually do – empowering us to take immediate action, to correct the issue, before a customer notices and before they have to spend their valuable time contact us.
This talk will focus on our journey to build this model, and how we are able to operationalize the findings quickly using RPA.
Finding a way to predict which customers will experience those issues, and taking and action BEFORE the customer calls has been a long sought after objective in the business. At Rogers, we have used Machine Learning to develop such a model with precision rates of over 90%. We have also combined this prediction with an action layer driven Robotics Automation, which takes the actions required to correct the technical issue "before the Customer notices it".
I would like to share how this ecosystem (ML, Robotics and process engineering) will result in significant benefits for the organization.
What You'll Learn:
Real world applications of ML
How we operationalize model findings quickly in an
Computer Scientist, Derwen, Inc.
Paco Nathan is known as a "player/coach", with core expertise in data science, natural language, machine learning, cloud computing; 38+ years of tech industry experience, ranging from Bell Labs to early-stage start-ups. Advisor for Amplify Partners, IBM Data Science Community, Recognai, KUNGFU.AI, Primer. Lead committer PyTextRank. Formerly: Director, Community Evangelism @ Databricks, and Apache Spark. Cited in 2015 as one of the Top 30 People in Big Data and Analytics by Innovation Enterprise.
Talk: Industry Survey Analysis: The Industry Landscape of Natural Language Use Cases in 2020
Talk: Industry Survey Analysis: The Industry Landscape of Natural Language Use Cases in 2020
Abstract:
We recently conducted an industry survey of firms that have natural language systems in production. This includes an organization that has a history of leveraging NLP systems as well as those which are just beginning to plan their approach. A "dramatic shift" would be an understatement: since 2018, the field of natural language has undergone a sea change. Breakthroughs in the usage of deep learning, as well as the availability of more sophisticated hardware and cloud resources, led to sudden advances in natural language. The results are pervasive across technology subcategories within the field of natural language: parsing, natural language understanding, sentiment detection, entity linking, speech recognition, abstractive summarization, and so on.
While the tech unicorns and their proxies have conducted almost an "arms race" since early 2018, sometimes publishing papers twice monthly to outdo their competitors' most recently published benchmarks -- how are these advances diffusing into practical use cases, and becoming adopted by mainstream businesses for their needs? Our survey results explore both the contours of the evolving landscape as well as the industry adoption and business trends for NLP.
What You'll Learn:
Real world applications of ML
The 2020 industry landscape for NLP use cases in production; the relative "market share" for the popular open-source libraries/frameworks; and analysis of cloud service usage and failure cases; plus industry drivers for accuracy vs. cost in new NLP advances
Senior Data Scientist, Microsoft
Miguel González-Fierro is a Sr. Data Scientist at Microsoft UK, where his job consists of helping customers leverage their processes using Big Data and Machine Learning. Previously, he was CEO and founder of Samsamia Technologies, a company that created a visual search engine for fashion items allowing users to find products using images instead of words, and founder of the Robotics Society of Universidad Carlos III, which developed different projects related to UAVs, mobile robots, small humanoids competitions, and 3D printers. Miguel also worked as a robotics scientist at Universidad Carlos III of Madrid (UC3M) and King’s College London, where his research focused on learning from demonstration, reinforcement learning, computer vision, and dynamic control of humanoid robots. Miguel is an Electrical Engineer by UC3M, PhD in robotics by UC3M in collaboration with King’s College London and graduated by MIT Sloan School of Management.
Workshop: Knowledge Graph Recommendation Systems for COVID-19
Workshop: Knowledge Graph Recommendation Systems for COVID-19
Core Takeaways:
CORD-19, or COVID-19 Open Research Dataset, was launched in March 2020, in response to a request from the White House’s office of Science and Technology Policy. The joint collaboration from AI2, Microsoft, the NLM at the NIH, and other prestigious research institutes aims at empowering the world’s AI researchers with a text and data mining tools to help accelerate COVID-19 related research.
The objective of this tutorial is to give the audience hands-on experience to work through the basics about knowledge graph and recommender technology, and how to use them for building an article recommender for COVID-19 research.
All the code is available and open sourced.
Prequestite Knowledge:
- Knowledge of python
- Knowledge of recommendation systems
Senior Data Scientist, Manager, Scotiabank
Pooja is a Senior Manager, Data Scientist at Scotiabank, Pooja has given several talks as part of analytics community at Scotiaban and was also a TA at last year's workshop at TMLS
Workshop: Leveraging Pretrained Language Models for Natural Language Understanding
Workshop: Leveraging Pretrained Language Models for Natural Language Understanding
Core Takeaways:
Deep learning based language models based on transformer architecture such as BERT and GPT have changed the way we approach Natural language processing tasks. These huge complex models trained on billions of words of text have been made available to researchers and industry to solve real-world problems.
In this workshop, we learn to leverage pre-trained language models(open source) and apply them to various NLP tasks such as sentiment analysis, Document classification, Q&A, and Named Entity Recognition. In practical scenarios, it is often important to learn how to fine-tune these deep transformer models on your domain-specific datasets. The process of fine-tuning involves labeling data with tools such as 'Doccano' and transforming your datasets to standard formats such as CoNLL. The following are the topics that will be covered by this workshop
1. Sentiment analysis, text classification, Named Entity recognition and QA using Bert and Spacy Models
2. Fine-tune Bert and Spacy models on an open-source text data.
3. Data Labeling using 'Doccano'.
4. Extracting embedding from text using pre-trained language models to score sentence similarity.
All computation for this workshop will be performed on google Colab and will be completely in python using Jupyter notebooks.
Prequestite Knowledge:
Python
Data Scientist, Shopify
Jill is a data scientist at Shopify, where she tackles a wide range of fascinating data problems on the international team. Outside of work, Jill spends her time participating in datathons (hackathons for data scientists), running events for PyLadies and PyData Toronto, and playing tennis (when it's warm enough to go outside)
Workshop: Building a MovieLens Recommender System
Workshop: Building a MovieLens Recommender System
Core Takeaways:
Want to know how Spotify, Amazon, and Netflix generate recommendations for their users? In this workshop, we will explore different types of recommendation systems and their implementations. We will build our own recommendation system from scratch using collaborative filtering and content-based filtering techniques in Python.
Prequestite Knowledge:
Attendees must know how to use Pandas
Chief Data Scientist, Domino Data Lab
Josh Poduska is the Chief Data Scientist at Domino Data Lab. He has 18 years of experience in the analytics space. As a practitioner, he has designed and implemented data science solutions across a number of domains including manufacturing and public sector. He has managed teams and led strategic initiatives for multiple analytical software companies. Josh has a Masters in Applied Statistics from Cornell University.
Workshop: Managing Data Science in the Enterprise
Workshop: Managing Data Science in the Enterprise
Core Takeaways:
Participants will leave with an understanding of the pitfalls that prevent breakthroughs and the best practices that lead to market-leading data science programs. Pointed lessons learned and unique insights from leading data science organizations will be shared covering how to effectively manage your people, your process, and your technology.
Prequestite Knowledge:
This workshop is designed for data science and business managers. Experience managing/leading technical teams and/or organizational initiatives are helpful recommended. Knowledge of the basic steps of a data science project's lifecycle is recommended.
Sr. Product Manager Machine Learning, Amazon Web Services
"Wenming Ye is an AI and ML Product Manager at Amazon Web Services, helping researchers and enterprise customers use cloud-based machine learning services to rapidly scale their innovations. Previously, Wenming had a diverse R&D experience at Microsoft Research, SQL engineering team, and successful startups. Rachel Hu is an applied scientist on the AWS AI working on deep learning. She received her master’s degree of statistics from University of California, Berkeley. She is an instructor at Amazon Machine Learning University and frequently presents at external events such as AWS Re:invent, Nvidia GTC, etc. She enjoys empowering everyone who is curious about start-of-the-art deep learning algorithms with easy to understand instructions and innovative new teaching tools. Before joining Amazon, Rachel also worked on natural language processing projects to promote user engagements in multiple industries."
Workshop: Put Deep Learning to Work: Accelerate Deep Learning through AWS SageMaker and ML Services
Workshop: Put Deep Learning to Work: Accelerate Deep Learning through AWS SageMaker and ML Services
Core Takeaways:
Distributed multiGPU BERT fine tuning and GPU Hosting using Amazon SageMaker.
Prequestite Knowledge:
Basic Deep Learning knowledge.
Applied Scientist, Amazon Web Services
Rachel Hu is an applied scientist on the AWS AI working on deep learning. She received her master’s degree of statistics from University of California, Berkeley. She is an instructor at Amazon Machine Learning University and frequently presents at external events such as AWS Re:invent, Nvidia GTC, etc. She enjoys empowering everyone who is curious about start-of-the-art deep learning algorithms with easy to understand instructions and innovative new teaching tools. Before joining Amazon, Rachel also worked on natural language processing projects to promote user engagements in multiple industries.
Workshop: Put Deep Learning to Work: Accelerate Deep Learning through AWS SageMaker and ML Services
Workshop: Put Deep Learning to Work: Accelerate Deep Learning through AWS SageMaker and ML Services
Core Takeaways:
Distributed multiGPU BERT fine tuning and GPU Hosting using Amazon SageMaker.
Prequestite Knowledge:
Basic Deep Learning knowledge.
Sr Manager | RAPIDS, NVIDIA
"Brad Rees is a Senior Manager at NVIDIA and lead of the RAPIDS cuGraph team. Brad has been focusing on Data Analytics, Artificial Intelligence, and HPC for over 30 years. Brad holds a Ph.D. in Computer Corey Nolet is a Sr. Data Scientist & Engineer on the RAPIDS cuML team at NVIDIA, where he focuses on building and scaling machine learning algorithms to support extreme data loads at light-speed. Prior to working at NVIDIA, Corey spent over a decade building massive-scale exploratory data science & real-time analytics platforms for HPC environments in the defense industry. Corey holds Bs. & Ms. degrees in Computer Science. He is also working towards his PhD in the same discipline, focused on scaling and accelerating algorithms for exploratory data analysis. Corey has a passion for using data to make better sense of the world."
Workshop: Reaching Lightspeed Data Science: ETL, ML, and Graph with NVIDIA RAPIDS
Workshop: Reaching Lightspeed Data Science: ETL, ML, and Graph with NVIDIA RAPIDS
Core Takeaways:
Performance gains possible using GPUs and RAPIDS. Easy of migrating to RAPIDS. Data Science possibilities opened by using RAPIDS
Prequestite Knowledge:
Python. Would be nice to know: Pandas, ScikitLearn, NetworkX
Sr Data Scientist & Engineer | RAPIDS, NVIDIA
Corey Nolet is a Sr. Data Scientist & Engineer on the RAPIDS cuML team at NVIDIA, where he focuses on building and scaling machine learning algorithms to support extreme data loads at light-speed. Prior to working at NVIDIA, Corey spent over a decade building massive-scale exploratory data science & real-time analytics platforms for HPC environments in the defense industry. Corey holds Bs. & Ms. degrees in Computer Science. He is also working towards his PhD in the same discipline, focused on scaling and accelerating algorithms for exploratory data analysis. Corey has a passion for using data to make better sense of the world.
Workshop: Reaching Lightspeed Data Science: ETL, ML, and Graph with NVIDIA RAPIDS
Workshop: Reaching Lightspeed Data Science: ETL, ML, and Graph with NVIDIA RAPIDS
Core Takeaways:
Performance gains possible using GPUs and RAPIDS. Easy of migrating to RAPIDS. Data Science possibilities opened by using RAPIDS
Prequestite Knowledge:
Python. Would be nice to know: Pandas, ScikitLearn, NetworkX
Data Scientist/Machine Learning Specialist & Senior ML Specialist, AWS
Stefan Natu is a Sr. Machine Learning (ML) Specialist at Amazon Web Services with a focus on financial services. He spends his time working with enterprise financial services customers from investment banking, asset management and investment research on building secure environments, best practices on model development, model governance and operationalizing ML workflows. He did his PhD in Atomic and Condensed Matter Physics from Cornell, and worked as a research physicist at ExxonMobil building machine learning models for oil and gas exploration. Prior to joining AWS, he worked at Publicis Sapient as a data scientist focused on ML in the retail vertical.
Saeed Aghabozorgi Ph.D. is senior ML Specialist in AWS, with a track record of developing enterprise level solutions that substantially increase customers’ ability to turn their data into actionable knowledge. He is also a researcher in the artificial intelligence and machine learning field.
Workshop: Build, Train, and Deploy Models with Amazon SageMaker
Workshop: Build, Train, and Deploy Models with Amazon SageMaker
Core Takeaways:
Hands-on experience with SageMaker to train and deploy a ML model
Prequestite Knowledge:
Basics of machine learning
ML Specialist Solutions Architect, AWS
Saeed Aghabozorgi Ph.D. is senior ML Specialist in AWS, with a track record of developing enterprise level solutions that substantially increase customers’ ability to turn their data into actionable knowledge. He is also a researcher in the artificial intelligence and machine learning field.
Workshop: Build, Train, and Deploy Models with Amazon SageMaker
Workshop: Build, Train, and Deploy Models with Amazon SageMaker
Core Takeaways:
Hands-on experience with SageMaker to train and deploy a ML model
Prequestite Knowledge:
Basics of machine learning
Data Scientist, The Home Depot
Cody is a data scientist on the online visual intelligence team at the Home Depot. He received his PhD from Georgia Institute of Technology in 2019. He received the 2019 Innovative Applications in Analytics Finalist Award from the Caterpillar Informs Analytics Society with his research project on using machine learning to discover best practice across 737 hospital sites. He was also a recipient of the J. Leland Jackson Fellowship Award for being recognized as the Outstanding Bioinformatics Graduate Student of the year at Georgia Tech in 2018. His work at the Home Depot focuses on developing machine learning and deep learning models for improving visual experience for online customers.
Workshop: Computer Vision in Practice - Building an End-to-End Pipeline for Object Detection and Segmentation
Workshop: Computer Vision in Practice - Building an End-to-End Pipeline for Object Detection and Segmentation
Core Takeaways:
Image segmentation is the task of associating pixels in an image with their respective object class labels. In this tutorial, participants will learn to: 1. perform image segmentation using the state-of-art deep learning approaches and customized datasets; 2. build an end-to-end image segmentation pipeline: dataset customization and transformation, model training, validating, and testing, techniques for post-processing; 3. convert trained models to ios and tensorflow-lite for mobile deployment for building downstream applications.
Prequestite Knowledge:
Basic Python programming skills
Data Scientist, LVMH
Ala is a Data Scientist at LVMH - Moët Hennessy Louis Vuitton in Paris, Kaggle Expert, and speaker with interests in the fields of machine learning, programming, and big data, working on real-life problems and a passion holder for deploying predictive and deep learning models. He had gained expert knowledge of cutting-edge machine learning methods and applications while focusing on implementing many different machine learning approaches, features' engineering, productionize machine learning models and human-in-the-loop Data Science. Transforming software engineers and master students into data scientists and helping them to shape their careers has been the most rewarding thing for him. In the age of big digital data transformation, his mission is to discover algorithms and techniques that make sense and work in real life. Discovery is probably the easiest step but not enough. So, he spends time implementing what he finds interesting.
He also spends a lot of time of his career on both the engineering side but mostly in the data science side, developing active learning models to help descision maker, using scalable machine learning application and most importantly doing the research behind them.
Workshop: Computer Vision in Practice - Building an End-to-End Pipeline for Object Detection and Segmentation
Workshop: Computer Vision in Practice - Building an End-to-End Pipeline for Object Detection and Segmentation
Core Takeaways:
Image segmentation is the task of associating pixels in an image with their respective object class labels. In this tutorial, participants will learn to: 1. perform image segmentation using the state-of-art deep learning approaches and customized datasets; 2. build an end-to-end image segmentation pipeline: dataset customization and transformation, model training, validating, and testing, techniques for post-processing; 3. convert trained models to ios and tensorflow-lite for mobile deployment for building downstream applications.
Prequestite Knowledge:
Basic Python programming skills
Lead Architect - Cybersecurity Analytics, Telus
Abdul is leading the cybersecurity analytics roadmap at Telus. In this role he is leading the Telus home grown cybersecurity analytics data lake which uses AI, Machine Learning, Deep Learning, statistical modelling and rule based approaches for enterprise IT security and intrusion detection at scale on billions of events from heterogenous IT systems. Abdul is also leading discussions with academia in cybersecurity analytics on behalf of Telus to establish strategic partnerships for pushing state-of-the-art in security analytics and extending the Telus cybersecurity analytics data lake capabilities beyond IT security. He is also a steering committee member and “IoT and Edge Analytics” stream owner at AISC, a research based community in Toronto, where he invites leading researchers and industry practitioners for presentations and panel talks to discuss the latest research and state-of-the-art in IoT and edge security analytics.
Workshop: Intrusion Detection Systems - An Overview
Workshop: Intrusion Detection Systems - An Overview
Core Takeaways:
The audience will get an overview of techniques used by intrusion detection systems, both AI/Machine Learning and non machine learning, for identifying intrusions and malicious behaviours in systems. The audience will also get insights into how edge computing, edge analytics and fog computing can be leveraged by Intrusion Detection systems for security analytics at the edge for IoT. The talk will also discuss how big data and cloud is used for security analytics at scale both for IoT and enterprise security.
Prequestite Knowledge:
Some machine learning knowledge will be good to have but not required
Developer Relations Manager, Docker
Peter McKee is a Developer Advocate for Docker and maintainer of the open source project Ronin.js. Originally from Pittsburgh, PA but currently residing in Austin, TX, Peter built his career developing full-stack applications for over 25 years. He has held multiple roles but enjoys teaching and mentoring the most. When he’s not slapping away at the keyboard, you can find him practicing Spanish and hang out with his wife and seven children.
Workshop: Docker Based Workflow for Deploying a Machine Learning Model
Workshop: Docker Based Workflow for Deploying a Machine Learning Model
Core Takeaways:
After attending this workshop, students will learn how Docker and containers fit into the ML development lifecycle. We will start with the basics of containers and work our way up to a full example. We will cover topics such as: reproducibility, portability, and ease of deployment.
Program Manager, Microsoft
Aniththa Umamahesan is a Program Manager on Microsoft Azure Machine Learning team. Her latest mission is accelerating and democratizing Artificial Intelligence via Automated Machine Learning.
Workshop: Machine Learning Simplified: From Ideation to Deployment in Minutes with Automated Machine Learning
Workshop: Machine Learning Simplified: From Ideation to Deployment in Minutes with Automated Machine Learning
Core Takeaways:
Artificial intelligence (AI) has become the hottest topic in tech. Executives, analysts, engineers, and developers all want to leverage the power of AI to gain better insights and make better predictions. But machine learning requires advanced data science skills that are hard to come by. Automated ML is an emerging field that helps developers and new data scientists build ML models without understanding the complexity of algorithm selection and hyper parameter tuning. This session shows you how to train a high quality model with Azure Machine Learning automated ML by supplying only a dataset and a few configuration parameters.
Prequestite Knowledge:
Tutorial will be for all levels.
Data Scientist, Iterative.ai
Elle is a data scientist at Iterative, a startup building open source software tools for machine learning. She completed her PhD at the University of Washington where she conducted research on speech and hearing using mathematical models. Elle is broadly interested in developing methods, standards, and educational resources for anyone who works with data.
Workshop: How to Automate Machine Learning with GitHub Actions
Workshop: How to Automate Machine Learning with GitHub Actions
Core Takeaways:
- How to write your own GitHub Actions to automatically train, test, and report machine learning models
- Why automation is key to speeding up development cycles and getting models into production faster
Prequestite Knowledge:
- Basic familiarity with machine learning (framework agnostic)
- Some exposure to Git/GitHub is helpful, but not necessary. Git beginners are welcome.
AI Instructor, WeCloudData
Rhys Williams is a senior data scientist at Beam Data. He’s an avid lover of machine learning and AI. In his spare time, he builds his own robots using Raspberry Pi and trains deep neural nets to teach the bots to recognize objects and understand human language. Rhys is also a part-time AI instructor at WeCloudData.
Workshop: Deploying Deep Learning Models to The Edge
Workshop: Introduction to Tensorflow (Hands-on Workshop)
Workshop: Deploying Deep Learning Models to The Edge
Core Takeaways:
Learn how Tensorflow models can be deployed to mini-robots
Prequestite Knowledge:
Python, Deep Learning, Computer Vision basics, IoT basics
Workshop: Introduction to Tensorflow (Hands-on Workshop)
Core Takeaways:
Complete an end to end Tensorflow tutorial
Prequestite Knowledge:
Python, ML basics
CTO and founder, Iguazio
Yaron Haviv is a serial entrepreneur who has been applying his deep technological experience in data, cloud, AI and networking to leading startups and enterprise companies since the late 1990s. As the co-founder and CTO of Iguazio, Yaron drives the strategy for the company’s data science platform and leads the shift towards real-time AI. He also initiated and built Nuclio, a leading open source serverless platform with over 3,400 Github stars and MLRun, Iguazio’s open source MLOps orchestration framework.
Prior to co-founding Iguazio in 2014, Yaron was the Vice President of Datacenter Solutions at Mellanox (now NVIDIA), where he led technology innovation, software development and solution integrations. He was also the CTO and Vice President of R&D at Voltaire, a high-performance computing, IO and networking company which floated on the NYSE in 2007. Yaron is an active contributor to the CNCF Working Group and was one of the foundation’s first members. He presents at major industry events and writes tech content for leading publications including TheNewStack, Hackernoon, DZone, Towards Data Science and more.
Workshop: MLOps & Automation Workshop: Bringing ML to Production in a Few Easy Steps
Workshop: MLOps & Automation Workshop: Bringing ML to Production in a Few Easy Steps
Core Takeaways:
How to deploy ML/AI faster to production using various MLOps technologies
Prequestite Knowledge:
Familiarity with Jupyter Notebooks, Pandas, and common ML tools
Partner & Co-Founder, INQ Data Law
Talk: Autonomous Vehicles - The Next Step Forward
Talk: Autonomous Vehicles - The Next Step Forward
Abstract:
What You'll Learn:
Sr. Manager, Applied Data Science, Dell Technologies
Randi manages a team of data scientists at Dell Technologies within Support and Deployment Services who deliver data science solutions using telemetry data to proactively prevent customer issues and resolve them more quickly when they do happen. As a data scientist, she brought data science solutions to business problems involving tech support, warranties, and repairs on Dell products. She continues to focus on raising visibility for data science at the executive level and connecting global Dell data scientists into a networked community that can collaborate and learn from one another. Additionally, she is a co-organizer of Women in Data Science ATX and promotes diversity and fostering a welcoming space for newcomers to the field. Before venturing into industry, Randi completed a PhD in Astrophysics at UT Austin, including research on both active galactic nuclei and how students learn astronomy, which gave her experience with varied statistical data-mining techniques and many kinds of data sets.
Workshop: Managing Data Science in the Enterprise
Workshop: Managing Data Science in the Enterprise
Core Takeaways:
Participants will leave with an understanding of the pitfalls that prevent breakthroughs and the best practices that lead to market-leading data science programs. Pointed lessons learned and unique insights from leading data science organizations will be shared covering how to effectively manage your people, your process, and your technology.
Prequestite Knowledge:
This workshop is designed for data science and business managers. Experience managing/leading technical teams and/or organizational initiatives are helpful recommended. Knowledge of the basic steps of a data science project's lifecycle is recommended.
VP of AI Engineering and CIO, Vector
Ron is the VP of AI Engineering and CIO at Vector Institute and is the Engineering Lead at the Schwartz Reisman Institute for Technology and Society. Ron is responsible for leading engineering teams that apply Vector’s leading AI research to industry and health care problems for Canada, establishing and supporting world class scientific computing infrastructure to scale the adoption of beneficial AI, and ensuring that all Vector users, sponsor participants and partners are upskilled to use it effectively.
Talk: Responsible AI
CTO and Co-founder, Evidently AI
Emeli Dral is a Co-founder and Chief Technology Officer at Evidently AI, a startup developing tools to analyze and monitor the performance of machine learning models. Prior to that, she co-founded a startup focused on the application of machine learning in the industrial sector and served as the Chief Data Scientist at Yandex Data Factory. She led a team of accomplished data scientists and oversaw the development of machine learning solutions for various industries - from banking to manufacturing. Emeli is a lecturer at the Graduate School of Management of St. Petersburg State University and Harbour.Space University, where she teaches courses on machine learning and data analysis tools. In addition, she is a co-author of the Machine Learning and Data Analysis curriculum at Coursera with over 100,000 students. In 2017, she also co-founded Data Mining in Action, the largest open data science course in Russia.
Talk: How Your ML Model Will Fail - And How to Prepare for It?
Talk: Responsible AI
Abstract:
Advances in AI promise tremendous benefits to society but also pose significant challenges. The Vector Institute is at the forefront of AI research and operationalizing AI, and collaborates closely with the Schwartz Reisman Institute for Technology and Society to advance the state of Responsible AI. As the field continues to advance, responsibility is becoming increasingly important to meet expectations of all stakeholders. Learn about challenges such as unintended user and societal harm, unfair bias, surveillance, adversarial attacks.
In this talk we look at challenges and emerging research and practices to address:
• Unintended consequences
• Model interpretability
• Fairness
• Engineering objectives (loss functions)
Talk: How Your ML Model Will Fail - And How to Prepare for It?
Abstract:
The data scientist’s job does not finish when the model is shipped. Models degrade and break in production. The failure modes of machine learning systems are also different from those of traditional software applications. They require purpose-built monitoring and debugging. However, this aspect is often overlooked in practice. In this talk, we will explore:
- How and why the machine learning models break;
- How to analyze production model performance, data drift and monitor data quality;
- How to set up your monitoring strategy in a pragmatic way.
What You'll Learn:
How to set up your model monitoring from scratch, and how to prioritise different metrics
Data Science Lead, Statflo Inc.
Doug Creighton is the Data Science Lead at Statflo. Statflo’s TextKit is a 1:1 business text messaging platform that works seamlessly with your existing systems.
Before working and understanding retail data, Doug worked as an analyst in the energy industry finding low-cost ways to get energy from A to B. Doug has spent his career solving innovative data problems to impact decision making internally or for his customers. He often spends a lot of his free time solving personal data problems or collaborating with others. His latest AI project at hopupon.com.
Talk: A Machine Learning based Privacy-Preserving Framework for Generating Synthetic Data from Aggregated Sources
Talk: A Machine Learning based Privacy-Preserving Framework for Generating Synthetic Data from Aggregated Sources
Abstract:
A synthetic dataset is a data object that is generated programmatically, and it is often necessary for situations where data privacy is a concern, or when collecting data is difficult or costly. Although it is a fundamental step for many data science tasks, an efficient and standard framework is absent. In this presentation, we study a specific synthetic data generation task called downscaling, a procedure to infer high-resolution information (e.g., individual level records) from low-resolution variables (e.g., average of many individual records), and propose a multi-stage framework. Specifically, we discuss 1) how synthetic data is generated from aggregated sources like census, 2) why is this important from a application perspectives, and 3) two real world use cases demonstrating why using synthetic data generation can significantly improve model performances.
What You'll Learn:
I will present a novel method for generating synthetic datasets (which has not yet been published) as well as 2 real world case studies of Arima's partners on how synthetic data has improved their model performances.
Professorial Lecturer, American University
Selika is an innovative strategist, transportation executive, and motivational speaker. Currently researching and educating on electric and autonomous vehicle policy, impacts on governments, OEMs and regulatory stakeholder communities at American University in Washington, DC.
Selika formerly served as the co-chair for the FMCSA (Federal Motor Carrier Safety Administration) Automated Working Group and served on the AAMVA (American Association of Motor Vehicle Administrators) Automated Vehicle Work Group. From electric/autonomous vehicles to politics she has written and spoken about public policy regarding transportation, innovation and technology as well as provided strategic assistance for transportation associations and companies.
Selika was formerly the Appointed Deputy Administrator for the State of New Jersey Motor Vehicle Commission in charge of all Operations for state services, the Director of Field Operations for the FMCSA (Federal Motor Carrier Safety Administration) and has over 18 years’ experience as an attorney litigator and was most recently the Senior Advisor to the Administrator of the FMCSA.
She has spoken nationally and internationally providing unique strategies for companies and governments in a variety of policy areas. Selika currently serves on the PAVE (Partners for Autonomous Vehicle Education) Academic Advisory Counsel and has authored several articles for Forbes.com. Selika is a sought after subject matter expert including having been recently interviewed by Axios and Automotive News.
Talk: Political Economy of Future Transportation and Equity
Talk: Political Economy of Future Transportation and Equity
Abstract:
The concept of the innovation and transportation and real world issues of public engagement, political policy and equity
What You'll Learn:
Policy considerations when designing for future mobility
Data Scientist, Loblaw Digital
Lan is a data scientist in the growth and marketing team at Loblaw digital. She holds Master's degrees from the University of Toronto and Imperial College London. Her work at Loblaw Digital includes developing a feature hub to serve multiple models, building customer models for campaign targeting, and implementing an auto model selection platform.
Talk: From Data Warehouse To Feature Warehouse: How DBT & AirFlow DAGs Orchestrate Building A Reliable & Reusable Feature Repository To Accelerate Model Development
Talk: From Data Warehouse To Feature Warehouse: How DBT & AirFlow DAGs Orchestrate Building A Reliable & Reusable Feature Repository To Accelerate Model Development
Abstract:
At Loblaw Digital, we have abundant data resources. Yet, it takes a lot of processing before we can build predictive models or perform analysis on them. Data engineers, data analysts, and data scientists have to conduct time-consuming and repetitive tasks to understand the business logic within and across data components to get the desired features and datasets. Data discovery and data generation became the most challenging piece before putting ML solutions in production. To conquer these difficulties, we enrich millions of transactions from a variety of sources using data build tool (DBT) while ensuring quality checks. The pipelines are scheduled using AirFlow DAGs and they output in a single, scalable, consolidated repository. These features enable our teams to have a quicker turnaround time on our solutions' development.
What You'll Learn:
Senior Technology Leader & Platform Head-Data Engineer, Data Science & BI, Rogers Communications
Phoenix Majumder is an accomplished Data and Analytics leader, who has delivered cutting-edge analytics solutions that have provided a sound basis for informed business decisions. Adept in the execution of a wide spectrum of analytics capabilities that spreads across Analytics Strategy formulation to production-grade implementations. Phoenix thrives in delivering dependable data solutions and explaining complex analytical outcomes to a diverse set of audiences. He has lead implementation of Data and Analytics enablement technologies across a wide array of technology platforms. Excellent communicator, proven history of delivering impactful projects that have transformed analytical outcomes to enterprise strategies. Data Analytics Product Management, Machine Learning/ AI Operations, and Digital Transformation are few key areas of his recent focus. An Electrical Engineer by technical training, Phoenix holds graduate degrees in Business Administration and Analytics.
Talk: Lessons Learned Transitioning to the Cloud
Talk: Lessons Learned Transitioning to the Cloud
Abstract:
The discussion will cover a broad-spectrum of considerations on moving Analytics journey to cloud. The design aspects of the technology and how the technology can accelerate the Data Science and Machine Learning use community.
What You'll Learn:
Technology Strategy
Darwin AI
Workshop: Founder's Circle
Workshop: Founder's Circle
Core Takeaways:
Prequestite Knowledge:
What are the most imp lessons that AI practitioners and founders can learn from your experiences
What is the biggest blocker that’s blocking ML research and the adoption of AI/ML in the business world?
Is the technology side or business side harder in building an AI company?
Working with orgs who are new to AI and how to manage their expectations
What is the state of the art in ML and how can our audience benefit from it?
Role of the cloud provider and how it is evolving?
How does AI for good and sustainability play into your work?
How have VCs, accelerators and incubators helped you grow your startup?
Are folks doing ML the right way? What are ways we will see major improvement over the next 5-10 years?
Principal Researcher, Microsoft Research
Rich Caruana is a Senior Principal Researcher at Microsoft. His focus is on intelligible/transparent modeling, machine learning for medical decision making, deep learning, and computational ecology. Before joining Microsoft, Rich was on the faculty in Computer Science at Cornell, at UCLA's Medical School, and at CMU's Center for Learning and Discovery. Rich's Ph.D. is from CMU. His work on Multitask Learning helped create interest in a subfield of machine learning called Transfer Learning. Rich received an NSF CAREER Award in 2004 (for Meta Clustering), best paper awards in 2005 (with Alex Niculescu-Mizil), 2007 (with Daria Sorokina), and 2014 (with Todd Kulesza, Saleema Amershi, Danyel Fisher, and Denis Charles), and co-chaired KDD in 2007 with Xindong Wu.
Workshop: Black-Box and Glass-Box Explanation in Machine
Workshop: Black-Box and Glass-Box Explanation in Machine
Abstract:
The overall importance of intelligibility and explanation in machine learning.
New methods for providing explanations.
Glass-Box vs. Black-Box ML and explanation methods.
What You'll Learn:
Participants should be familiar with Supervised Machine Learning
Phenomic AI
Workshop: Founder's Circle