It’s that time again!

Abstract:
NLP tutorials and workshops typically start with a labeled/annotated dataset, and discuss different ways of representing text/building models. However, in many real-world scenarios, we don’t have that luxury of already having a labeled dataset. We may often end up in scenarios where we have a problem, and a way to solve it, but no dataset to start working on the solution! In this workshop, I will introduce some ways of approaching this problem, such as looking for existing datasets, data annotation, automatic data labeling, data augmentation, and transfer learning.

What You’ll Learn:
– Different Ways of Collecting Labeled Data
– Automatic Data Labeling
– Quick Manual Annotation of Data
– Data Augmentation in the Context of NLP.

Sowmya Vajjala currently works as a researcher in Digital Technologies at National Research Council, Canada’s largest federal research and development organization. She has worked in the area of Natural Language Processing (NLP) over the past decade in various roles – as a software developer, researcher, educator, and a senior data scientist. She recently co-authored a book: “Practical Natural Language Processing: A Comprehensive Guide to Building Real World NLP Systems”, published by O’Reilly Media (June, 2020), which was also translated into Chinese. Her research interests lie in multilingual computing and the relevance of NLP beyond research both in industry practice as well as in other disciplines, through inter-disciplinary research.

Abstract:
MLOps hurdles don’t end after models are pushed to production. In the ML lifecycle, inference workloads present a critical challenge, where throughput and latency become key measures and teams struggle to meet efficient GPU utilization levels. In this workshop applicable to both Data Scientists and ML Engineers, Guy Salton will give an overview of the challenges in moving ML prototypes to production, and how best-in-class ML teams are successfully overcoming these hurdles.
We’ll discuss using fractional GPU capabilities to improve throughput and reduce latency, and we’ll show how one organization built an inference platform on top of Kubernetes with the NVIDIA A100 MIG to support their scaling AI initiatives. There are very few organizations using the new NVIDIA MIG functionality successfully, so even if you’re not using A100s yet, this is a unique opportunity to see how the MIG works for inference use cases.

What You’ll Learn:
How to use MPS (Multi-Process Services) with fractional GPU to increase throughput
Dynamic MIG functionality – how to get dynamic MIG slices for each new job when using the NVIDIA A100 GPU
Additional capabilities to run Inference on top of Kubernetes

Guy Salton is the Solutions Engineering Lead at Run:AI, specializing in the fields of DevOps, Cloud Computing, Kubernetes, Containers, Virtualization, CI/CD and AI computing. He runs POCs and technical projects for our commercial and enterprise customers, including on-site installations and workshops. Guy speaks at conferences and meetups around the world, writes blog posts and delivers webinars.

Abstract:
It is indeed a wonderful time to build machine learning systems, as we don’t have much to do anymore! Thanks to a growing ecosystem of tools and shared best practices, even small teams can be incredibly productive at “reasonable scale”. In this talk, we present our philosophy for modern, no-nonsense data pipelines, highlighting the advantages of a PaaS approach, and showing (with open, freely available code) how the entire toolchain works on real-world data with realistic constraints. We conclude with some unsolicited advice on the future of ML for “reasonable” companies, based on our experience in small and large organizations.

What You’ll Learn:
How to build a full-fledged, end-to-end MLOps stack using best-of-breed open source tools and PaaS options, to make ML teams 10x more productive.

Educated in several acronyms across the globe (UNISR, SFI, MIT), Jacopo Tagliabue was co-founder and CTO of Tooso, an A.I. company in San Francisco acquired by Coveo in 2019. Jacopo is currently the Lead A.I. Scientist at Coveo, shipping models to hundreds of customers and millions of users. When not busy building products, he is exploring research topics at the intersection of language, reasoning and learning: he is a committee member for international NLP/IR workshops, and his work is often featured in the general press and A.I. venues (e.g. NAACL, SIGIR, RecSys, ACL). In previous lives, he managed to get a Ph.D., do sciency things for a pro basketball team, and simulate a pre-Columbian civilization.

Abstract:
Over the past years, a new stack of mature, open-source tools for MLOps has started emerging: Metaflow was started at Netflix to make it easy for data scientists to develop robust ML workflows which can be used to train models at scale. Seldon is powering tens of thousands of clusters for model deployments. Together, Seldon and Metaflow cover the full ML project lifecycle from prototyping to production deployments.
This is a technical workshop focusing on the full stack of ML infrastructure. Learn from core developers of Seldon and Metaflow how to develop models at scale, and how to deploy them to production-grade infrastructure. We will have an open QA, so come prepared to ask questions about any aspects of ML infrastructure.

What You’ll Learn:
– How to set up infrastraucture using open-source tools that allows data scientists to develop models at scale and deploy them as microservices.
– How to leverage Kubernetes clusters for ML and data science effectively.
– Gotchas and lessons learned from thousands of ML projects.

Ville has been developing infrastructure for machine learning for more than two decades. He has worked as an ML researcher in academia and as an infrastructure leader at a number of companies, including Netflix where he led the ML infrastructure team that created Metaflow, a popular open-source framework for data science infrastructure. He is a co-founder and CEO of Outerbounds, a company that continues the Metaflow journey. He is also the author of a new book, Effective Data Science Infrastructure, which will be published by Manning in 2021.

Clive is CTO of Seldon. Seldon helps enterprises put machine learning into production. Clive developed Seldon’s open source Kubernetes based machine learning deployment platform Seldon Core. He is also a core contributor to the Kubeflow and KFServing projects.

Alejandro is the Director of Machine Learning Engineering at Seldon Technologies, where he leads large scale projects implementing open source and enterprise infrastructure for Machine Learning Orchestration and Explainability. Alejandro is also the Chief Scientist at the Institute for Ethical AI & Machine Learning, where he leads the development of industry standards on machine learning explainability, adversarial robustness and differential privacy. With over 10 years of software development experience, Alejandro has held technical leadership positions across hyper-growth scale-ups and has a strong track record building cross-functional teams of software engineers.
Alejandro’s Linkedin
Alejandro’s Twitter
Alejandro’s Github
Alejandro’s Website

Romain currently works in the Machine Learning Platform team at Netflix. In that role, he helps data-scientists and other users of machine learning to develop infrastructure that enables them to focus more on data-science, less on infrastructure and provide more business value. Romain is a core developer of Metaflow, the open-source project that originated at Netflix to make data-scientists more productive. Prior to working at Netflix, Romain was at Facebook where he worked on Applied Machine Learning focusing on low-level com
To enable screen reader support, press Ctrl+Alt+Z To learn about keyboard shortcuts, press Ctrl+slash

Abstract:
In this AI art initiation workshop, we introduce powerful yet accessible tools to be creative with machine learning. We first start by reviewing the different aspects of an AI art piece and rapidly move on to the creation and manipulation of the dataset we will use. The second, and most important, part of the workshop is dedicated to learning the basics of Cables.gl, a visual programming framework in the browser. This environment allows us to quickly iterate over different designs and opens the door to a large number of uses outside art such as motion graphics or data visualization. While no real programming experience is required, general tech-savviness is mandatory to be comfortable.
*Prerequisites*
To apply to the workshop, you need to be comfortable with manipulating files on your system, browsing the web and be completely autonomous with your computer/laptop. You have a working microphone and high-speed internet connection. Your browser needs to support WebGL (Chrome preferred). Make sure you have the latest version and test it here: get.webgl.org. A physical mouse is also preferred as we will manipulate 3D objects. A large display (or 2) is recommended to work alongside the videoconference window.

What You’ll Learn:
– The role and function of AI Art and how it relates to Data Art
– The artistic process behind the creation of a AI/Data Art piece
– How to get started with Cables.gl
– The next steps on your AI Art journey

Dr Kirell Benzi is a data artist, public speaker and AI researcher. His work revolves around the creation of aesthetic experiences that inspire, educate and empower large audiences using state-of-the-art technology. Through a hypnotic visual semantic, he tries to demonstrate that algorithms have a soul; and that we can create positive emotions from complexity using methods that come straight from scientific research.
Kirell holds a Ph.D. in Data Science from Ecole Polytechnique Fédérale de Lausanne (EPFL) and teaches Data Visualization / Data Art at various universities.

Abstract:
In this session we’ll introduce and demo a new open source paradigm for AI benchmarking, called DynaTask. This novel benchmark suite uses dynamic adversarial data collection to evaluate AI models, and assess how easily an AI can be fooled by humans.

What You’ll Learn:
– Undestanding AI benchmarking
– Introduction to using DynaBench
– Overview of dynamic adversarial data collection
– Demo of the AI benchmarking software
– How to use and contribute to the open source

Sam Lightstone recently joined the technical leadership at Facebook. From 2020-2021 Sam was IBM Chief Technology Officer for AI. From 2017-2020 Sam was the IBM Chief Technology Officer for Data focusing on IBM’s database and big data portfolio. He is cofounder of the IEEE Workgroup on Self-Managing Database Systems. Sam has more than 65 patents issued and pending and has authored 4 books and over 30 papers. Sam’s books have been translated into Chinese, Japanese and Spanish. In his spare time he is an avid guitar player and fencer. His Twitter handle is @samlightstone.

Abstract:
For practical and pragmatic reasons, the world is often designed for the “average” user. As a result, designers and developers across a variety of disciplines, including Artificial Intelligence (AI) and Machine Learning (ML), often make decisions that exclude particular audiences in their services and products.
This workshop was made for AI and ML experts, businesses, and other researchers and practitioners who are looking to move beyond designing for the average.
This workshop is intended for those who are instead interested in narrowing design gaps through integration of inclusive design into the everyday of business.
This workshop will:
1) Challenge participants to identify the biases, systems, and other barriers that lead to narrow or exclusive design;
2) Encourage participants to consider how these barriers present within their own business and how these barriers impact their products and services;
3) Equip participants with the knowledge, skills, and tools to build inclusive design into how they do business; and,
4) Provide concrete success metrics to provide accountability and help keep teams on track.
As research in cognition and neuroscience grows, we better understand the many different ways in which people learn, think, and interact with the world. This improved understanding of the diversity of human cognition provides concrete evidence of the limitations of designing for the average user and points to how this decision can have the consequence of leaving large portions of the population under-represented and under-served.
At a time when we are increasingly acknowledging the value of including a diversity of voices, companies need to look for ways to incorporate inclusive design into their business practices. By making the decision to continuously incorporate diverse user feedback into research, development and implementation cycles, companies can better ensure that products and services empower participation and inclusion.
Inclusive by default may be the desired end-state but it is a goal that can only be realized by continuous dedication, intentionality, and years of inclusive design practice. By participating in this workshop, you will be making a commitment to, and a step forward on, your journey towards increased inclusion.

What You’ll Learn:
1) Overview of inclusive design principles;
2) Discussion of what it means to design for and with cognitive and neurological diversity;
3) Understanding of the role of cognitive bias and othernbarriers to inclusive design;
4) Examination of AI and ML industry case studies of designs that hit the mark (and some that don’t); and,
5) Sharing of applicable practices for building an inclusive design practice of your own.

Dr. Deirdre Kelly has applied her expertise in cognitive science, decision-making and user experience design to help people, teams, and organizations find novel and innovative solutions to complex design challenges. She is currently leading research and development in the design of healthy and effective organizations and organizational culture. Dr. Kelly is an accomplished writer and speaker that has presented at both international and domestic venues. In her free time, Dr. Kelly is dedicated to serving her community through research and advocacy. She is currently working with YouTube activist Jessica McCabe from “”How to ADHD”” to develop resources that empower effective decision-making for people with Attention Deficit Hyperactivity Disorder (ADHD).
Dr. Kelly has a strong commitment to life-work balance. She stays grounded with a regular yoga and meditation practice and by getting out into the woods. She enjoys writing satire, poetry, and satirical poetry as well as reading all of the time and all of the things. Dr. Kelly is grateful to share her life with a supportive partner, a furry companion named Riley, and friends who she thinks are some of the best people out there.

Abstract:
We created an open source template in python for the entire ML journey from concept to production. The workshop offers a 2 part hands-on tutorial. Each part will be for 2 hours.
Part 1 starts with an example use case. It builds the ML components such as data prep, model hyper train, model train, model deploy and online/batch prediction. These components are unit tested in a python notebook.
Part 2 will show how to deploy these components in a Kubeflow pipeline with orchestration for training and prediction. The entire end to end ML pipeline is now ready for deployment.
At the end of this tutorial you will have hands-on experience building a model from concept to a final production-ready ML pipeline. The tutorial will be implemented on the Google Cloud Platform with Vertex AI. Models include xgboost and tensorflow models.

What You’ll Learn:
At the end of this tutorial you will have hands-on experience building a model from concept to a final production-ready ML pipeline. The tutorial will be implemented on the Google Cloud Platform with Vertex AI. Models include xgboost and tensorflow models.

Chanchal Chatterjee, Ph.D, held several leadership roles in machine learning, deep learning and real-time analytics. He is currently leading Machine Learning and Artificial Intelligence at Google Cloud Platform. Previously, he was the Chief Architect of EMC CTO Office where he led end-to-end deep learning and machine learning solutions for data centers, smart buildings and smart manufacturing for leading customers. Chanchal received several awards including Outstanding paper award from IEEE Neural Network Council for adaptive learning algorithms recommended by MIT professor Marvin Minsky. Chanchal founded two tech startups between 2008-2013. Chanchal has 29 granted or pending patents, and over 30 publications. Chanchal received M.S. and Ph.D. degrees in Electrical and Computer Engineering from Purdue University.

Abstract:
Getting Started & Going to Production with Hugging Face and Amazon SageMaker.

What You’ll Learn:
We will walk users through Getting Started with Amazon SageMaker and Hugging Face, where we will train your first NLP Transformer model. We will also upload that model to the Hugging Face Model Hub. After this introduction, we will go one step further by deploying your trained model and run inference to simulate a production environment.

Philipp Schmid is a Machine Learning Engineer and Tech Lead at Hugging Face, where he leads the collaboration with the Amazon SageMaker team. He is passionate about democratizing and productionizing cutting-edge NLP models and improving the ease of use for Deep Learning.

Mark McQuade is an AWS and cloud-based solution specialist, knowledge addict, relationship builder and ML Success and Business Development Lead at Hugging Face. Every day, Mark gets to learn more about what he is passionate about professionally – AI and machine learning – as well as the fascinating world of data. As a technology evangelist, you’ll often find Mark promoting data and AI/ML at talks, webinars, podcasts and industry events.

Abstract:
Modern AI application development is shifting from a model-centric practice to a data-centric approach. Rather than focusing solely on models trained over static datasets, practitioners are thinking more holistically about their pipelines, with a renewed emphasis on the training data.
In this workshop you will gain an in depth understanding of:
– The principles and strategies of Data-Centric AI
– How to building high-quality applications using the Snorkel framework
– The techniques for iteration, adaptation, and collaboration
– How to empower domain experts to accelerate AI application development

What You’ll Learn:
This talk/workshop is designed for ML practitioners interested in maintaining ML pipelines in production. We will not be discussing model development, data analysis, or feature engineering techniques used to build ML prototypes. In this talk/workshop, you will learn about:
– Types of bugs that occur in production ML pipelines
– How to incorporate testing into your ML application codebase in a sustainable way
In this workshop you will gain an in depth understanding of: 1- The principles and strategies of Data-Centric AI 2- How to building high-quality applications using the Snorkel framework 3- The techniques for iteration, adaptation, and collaboration 4- How to empower domain experts to accelerate AI application development

Roshni Malani is an engineering leader at Snorkel AI, building the core programmatic labeling and iterative model development experience. Previously, she managed teams building internal tools for the AI/ML organization at Apple and was one of the founding engineers on Google Photos. She received her Ph.D. in Software Engineering from UC San Diego. She believes that the true potential for machine learning can be unlocked by collaborating on the data.

Priyal Aggarwal is a Machine Learning Engineer at Snorkel AI. She graduated from Columbia University with a Master’s degree in Computer Science specializing in Machine Learning. She has previously worked as a Software Engineer at Microsoft Azure and an intern at Amazon Robotics. She is a vocal supporter of women in tech and served as the Director for Women Who Code Delhi.

Abstract:
Software organizations are increasingly incorporating machine learning (ML) into their product offerings, driving a need for new data management tools. Many of these tools facilitate the initial development and deployment of ML applications, contributing to a crowded landscape of disconnected solutions targeted at different stages, or components, of the ML lifecycle. A lack of end-to-end ML pipeline visibility makes it hard to address any issues that may arise after a production deployment, such as unexpected output values or lower-quality predictions. In this talk, we propose a system that wraps around existing tools in the ML development stack and offers end-to-end observability. We introduce our prototype and our vision for mltrace, a platform-agnostic system that provides observability to ML practitioners by (1) executing predefined tests and monitoring ML-specific metrics at component runtime, (2) tracking end-to-end data flow, and (3) allowing users to ask arbitrary post-hoc questions about pipeline health.

What You’ll Learn:
This talk/workshop is designed for ML practitioners interested in maintaining ML pipelines in production. We will not be discussing model development, data analysis, or feature engineering techniques used to build ML prototypes. In this talk/workshop, you will learn about:
– Types of bugs that occur in production ML pipelines
– How to incorporate testing into your ML application codebase in a sustainable way
– How to enable on-call ML engineers or data scientists to debug broken pipelines with ML-specific logging, monitoring, and querying

Shreya Shankar is a computer scientist living in the Bay Area and building systems to operationalize machine learning (ML) workflows. Her research focuses on end-to-end observability for ML systems, particularly in the context of heterogeneous stacks of tools. Currently, she is taking her Ph.D. in the RISE lab at UC Berkeley. Previously, she was the first ML engineer at Viaduct, did research at Google Brain, and obtained her BS and MS in computer science from Stanford.

Abstract:
More data means more robust and effective machine learning models. Traditional machine learning techniques rely on centralization of data therefore data silos are a blocker to achieving the desired performance. Federated Learning, is a well studied theoretical methodology that breaks the barrier of data silos by bringing the training of machine learning models to the datasets. In the recent years more efforts have been made to pave the way of moving this research concept into product. In this session we would like to help users become familiar with the concept and get an understanding of benefits and challenges in a hands on environment.

What You’ll Learn:
The applied challenges of cutting edge techniques in dealing with distributed data

Roshan is a startup addict (she has been among the first 10 employees at 3 successful startups) who is passionate about delivering real value to real people through machine learning products. A proud Waterloo graduate, she has spent 10+ years devoted to the art and science of researching, developing and productionizing machine learning products. She has delivered $100s of millions of value to companies using AI in a wide variety of use cases from wearables to warehouse management to marketing. Ro is a proud mom and art enthusiast (she loves to sketch and hang out at indie art exhibitions).

Nasron Cheong leads engineering at integrate.ai, and helps shape engineering practices needed for integrate.ai to develop privacy safe ML products. His experience includes large scale mobile analytics on big data platforms, as well as MLOps experience with various commercial and open source offerings

Abstract:
The focus of the panel will be on cyber security issues and the threat landscape in cyber physical systems and mobile devices. The panel will also discuss key innovations in Edge Computing and AI and Machine Learning that can be applied to cyber physical and mobile systems for misbehaviour and threat detection, threat containment, threat mitigation and recovery.

What You’ll Learn:
– Threat landscape and attacker TTPs for Control Systems, IoT, Automotive, Mobile Devices and commonalities and differences from traditional IT security
– Current mobile malware detection approaches
– Innovations in Edge Computing and scalable data processing on the edge
– Applications of AI and Machine Learning and Edge Analytics to misbehaviour and threat detection and threat containment and recovery in cyber physical systems
– Future research directions in cyber physical systems security and mobile malware detection

Yevgeny Kulakov has been in the computer security for more then a decade. Currently he is a research team lead at Microsoft focusing on protections from cross-platform threats, mainly on Linux and Mac.Previous to Microsoft, he was part of Lookout working on mobile device compromise detection. Apart of security he is an avid FPV drone pilot and enjoys trail hiking from time to time.

Prof. Sandeep K. Shukla is an IEEE fellow, and ACM Distinguished Scientist. He is currently a professor of Computer Science and Engineering department at IIT Kanpur which he headed during 2017-2020. He was the editor in chief of the ACM Transactions on Embedded Computing Systems during 2013-2020. He is currently associate editors of ACM Transactions on Cyber Physical Systems, and Journal of the British Blockchain Association. In the past he has served as associate editors of IEEE Transactions on Computers, IEEE Transactions on Industrial Informatics, IEEE Design and Test, and IEEE Embedded Systems Letters. Before joining IIT Kanpur in 2015, he was a professor at Virginia Tech, USA. He served as ACM Distinguished Speaker, and IEEE Computer Society Distinguished Visitor in the past. He has authored over 200 peer reviewed journal and conference papers, and authored/edited 10 books. He was awarded the Presidential Early Career Award in Science and Engineering (PECASE) in 2004, The Bessel Award by Humboldt Foundation in 2009, a Distinguished Alumnus Award by SUNY Albany in 2007, a Ramanujan Fellowship in 2015. His major research interest is Cyber Security of Critical Infrastructures, Cyber Security of IT/OT systems, and Applications of Blockchain Technology in Security and Privacy.

Abdul is the Senior Principal of Security Analytics at Arctic Wolf Networks where he is leading the strategy and R&D for application of advanced analytics and big data to cybersecurity. He has led research collaboration on cybersecurity analytics with multiple academic and industry bodies. He is the Steering Committee Member of multiple national and international technical communities in the advanced analytics and cybersecurity area. His research interests include cybersecurity, advanced analytics, distributed systems, edge computing, intelligent networks, IoT, CAVs and ICS.

Brandon’s focus in cybersecurity began at Brown University with academic research into automotive security while studying Computer Engineering. Brandon Barry founded Block Harbor in Detroit in 2014 on a mission to provide services to keep vehicles safe, ranging from starting a 24/7/365 security operation center for vehicle systems to building a vehicle cybersecurity lab for penetration testing. After having successfully stood up several automotive cybersecurity projects with automakers and tier-1 suppliers, Brandon and the team continue on at Block Harbor, where we are building great solutions to keep mobility safe. Brandon is also one of the world’s youngest full CISSP holders. Brandon’s team placed first place at DEF CON’s Car Hacking Village competition in 2019. In his spare time, Brandon is the Americas Lead for the Automotive Security Research Group (ASRG), a global nonprofit focused on knowledge, networking, and collaboration.

Ken Birman is the N. Rama Rao Professor of Computer Science at Cornell. An ACM and IEEE Fellow and the winner of the IEEE Tsutomu Kanai Award, Ken has written 3 textbooks and published more than 150 papers in prestigious journals and conferences. Software he developed operated the New York Stock Exchange for more than a decade without trading disruptions, and has been used in the French Air Traffic Control System and the US Navy AEGIS warship since 1996. Other technologies from his group found their way into IBM’s Websphere product, Amazon’s EC2 and S3 systems, Microsoft’s cluster management solutions, and the US Northeast bulk power grid. His newest project, the Cascade intelligent key-value storage system, is intended as a platform for hosting AI and ML code in demanding settings such as autonomous vehicles, the smart power grid, smart highways and homes, and smart farms.

Abstract:
According to projections, 80% of worldwide data will be unstructured by 2025. Financial Services (FS) industry is no different, where most enterprises hold vast array of unstructured data which is largely under-analysed. Typically, unstructured data refers to information that is not organized in a pre-defined manner or does not have a pre-defined data model. Here, the data is more challenging to interpret but can deliver a more comprehensive and holistic understanding for financial services use-cases. In this talk, we focus on unstructured documents in Finance and how Document Intelligence i.e., AI powered automated analysis of documents, allows to tap into the opportunities by analyzing huge amount of information present in such unstructured documents.
In financial services, unstructured documents include financial statements, invoices, bank statements, policies, contracts, marketing creatives etc. Data residing in such documents can be of variety types including images, tables, figures, and text. While there are challenges around processing documents, ability to quickly make decisions by leveraging such data can provide differentiated value propositions and competitive benefits. These benefits include improved operational excellence, automated compliance, or regulatory workflows, discovered insights from mining/ matching disparate data sources and overall enhanced customer experience. However, the very nature of unstructured data prohibits the direct application of AI/ML techniques that can be seamlessly applied on the structured data.
This talk will present the arts and sciences behind developing Document Intelligence solutions covering select use cases involving unstructured documents, show the business opportunities present and describe the technical challenges involved. Subsequently, we provide an outline to develop various Document Intelligence solutions that can aggregate, query, analyse, and accelerate the understanding of such data to unveil deep insights across Financial Services use-cases.

What You’ll Learn:
The participants are expected to come out after the talk with an enhanced understanding of AI/ML techniques in the Document Intelligence space including discovery of different types of document sources and their inherent potential, how to analyse & interpret unstructured documents such as financial & bank statements, marketing documents, invoices, receipts etc. We will cover specific cutting-edge techniques to perform information extraction on documents, including form-based, verbose, or tabular data, which helps transform documents into a structured format. The talk will introduce multiple real-world applications that are being powered by Document Intelligence such as credit ratings, under writings, mitigating compliance and regulatory risks. The talk will also provide an opportunity to the participants to get familiar with some of the open questions, research challenges, and opportunities in this space. Overall, we expect that our talk will educate the participants and will empower them to find answers to the following questions in this context:
(1) What are the types of unstructured data available & their potential for the financial industry?
(2) What are the challenges faced by the organizations in leveraging AI/ML to tap into the huge amount of information present in unstructured data?
(3) What is the value/impact that can be unlocked by leveraging such unstructured documents?
(4) What are some of the upcoming trends and open challenges?

Rahul Ghosh is a Vice President at American Express AI Labs. He received his MS and PhD from Duke University, NC, USA. In his current role at Amex, Rahul is responsible for the R&D function of the lab cutting across several areas of NLP, document AI, ML algorithms and optimization. His research interests include AI systems, Cloud AI and in general at the intersection of distributed systems and AI. Prior to joining in Amex, Rahul worked at Xerox Research India, at IBM Product Group in NC and at IBM T. J.Watson Research Center in NY. He is a co-author of 35+ peer reviewed papers and co-inventor of 50 US patents (granted/pending).

Himanshu Sharad Bhatt is currently a Research Director at American Express AI Labs where he is actively involved in developing Document AI-based solutions for unstructured data analytics. Prior to joining Amex in 2017, Himanshu has worked with Xerox Research, India towards building “unstructured data analytics”” capabilities for contact centers and services division. Himanshu holds a PhD degree in Computer Science & Engineering where his thesis was acknowledged with the “best thesis award”” by INAE and IUPRAI in 2014. Over the years, his work has led to 30+ publications in reputed conferences and journals and 5 US patent to his credit.

Abstract:
Extracting knowledge from text data has always been one of the most researched topics in machine learning, but only recently have we witnessed breakthroughs that put NLP in the spotlight. Much information is stored in unstructured data, like text, which is extremely important in many different fields, from finance to social media and e-commerce.
In this course, we will go through Natural Language Processing fundamentals, such as pre-processing techniques, embedding, and more. It will be followed by practical coding examples, in python, to teach how to apply the theory to real use cases.
The goal of this workshop is to provide the attendees all the basic tools and knowledge they need to solve real problems and understand the most recent and advanced NLP topics.

What You’ll Learn:
– Create an NLP pipeline to preprocess the data using python
– Perform topic modelling
– Use python libraries for NLP tasks, in particular, NLTK, Gensim and Glove
– Leverage transfer learning and text embeddings to perform NLP classification
– Use cutting edge embeddings in your project

Leonardo De Marchi holds a Master in Artificial intelligence and has worked as a Data Scientist in the sports world, with clients such as the New York Knicks and Manchester United, and with large social networks, like Justgiving.
He provides consultancy and training for small and large companies. His previous experience includes Head of Data Science and Analytics in Bumble, the largest dating site with over 500 million users, heading the team through acquisition and an IPO. He is also the lead instructor at ideai.io, a company specialized in Reinforcement Learning, Deep Learning and Machine Learning training. More details on the workshops can be found here.
He is also a contractor for several companies and for the European Commission, as an expert in AI and Machine Learning. As an author he wrote “Hands On Deep Learning” and he authored an online training course for O’Reilly, Introduction to Reinforcement Learning.
In the academic world, he also helped set up the PhD centre on Interactive Artificial Intelligence and will take part in the Inner Assessment Board to assign funding to Irish research in AI.

Working with data can be challenging: it often doesn’t come in the best format for analysis, and understanding it well enough to extract insights requires both time and the skills to filter, aggregate, reshape, and visualize it. This session will equip you with the knowledge you need to effectively use pandas – a powerful library for data analysis in Python – to make this process easier.
Pandas makes it possible to work with tabular data and perform all parts of the analysis from collection and manipulation through aggregation and visualization. While most of this session focuses on pandas, during our discussion of visualization, we will also introduce at a high level matplotlib (the library that pandas uses for its visualization features, which when used directly makes it possible to create custom layouts, add annotations, etc.) and seaborn (another plotting library, which features additional plot types and the ability to visualize long-format data).

What You’ll Learn:
Session Outline
Section 1: Getting Started with Pandas
We will begin by introducing the Series, DataFrame, and Index classes, which are the basic building blocks of the pandas library, and showing how to work with them. By the end of this section, you will be able to create DataFrames and perform operations on them to inspect and filter the data.
Section 2: Data Wrangling
To prepare our data for analysis, we need to perform data wrangling. In this section, we will learn how to clean and reformat data (e.g. renaming columns, fixing data type mismatches), restructure/reshape it, and enrich it (e.g. discretizing columns, calculating aggregations, combining data sources).
Section 3: Data Visualization
The human brain excels at finding patterns in visual representations of the data; so in this section, we will learn how to visualize data using pandas along with the matplotlib and seaborn libraries for additional features. We will create a variety of visualizations that will help us better understand our data.
Section 4: Hands-On Data Analysis Lab
We will practice all that you’ve learned in a hands-on lab. This section features a set of analysis tasks that provide opportunities to apply the material from the previous sections.

Stefanie Molin is a data scientist and software engineer at Bloomberg in New York City, where she tackles tough problems in information security, particularly those revolving around anomaly detection, building tools for gathering data, and knowledge sharing. She is also the author of “Hands-On Data Analysis with Pandas,” which is currently on in its second edition. She holds a bachelor’s of science degree in operations research from Columbia University’s Fu Foundation School of Engineering and Applied Science. She is currently pursuing a master’s degree in computer science, with a specialization in machine learning, from Georgia Tech. In her free time, she enjoys traveling the world, inventing new recipes, and learning new languages spoken among both people and computers.

Abstract:
Imagine a steady stream of insights to fuel business decisions, 360-degree customer views to boost relevance and revenue for faster, smarter recommendations to accelerate innovation. In reality, AI models that come fragmented, undocumented, and ungoverned can increase the risk of exposing your business to operational, reputational, and legal hazards. This is why reimagining how you govern AI projects and models is imperative to ensure transparency, trust, and control to scale your business at speed with AI.
In this session, Sophie Dionnet, General Manager, Business Solutions at Dataiku will show you how to scale at speed while taking control of AI.

What You’ll Learn:
​ – Tackle the growing complexity of AI models and projects
​ – Use AI Governance to enable, not disable, your AI projects
​ – Supporting AI Governance capability development to scale with AI

​Sophie is the General Manager of Dataiku’s Business Solutions team. She is focused on the development of industry-specific offerings as well as offerings related to AI governance. Sophie joined Dataiku in 2019 after 14 years in the financial industry, where she was Chief Operating Officer of a multi-asset management unit at a leading asset management firm. Sophie draws on extensive experience in strategic business leadership, management of complex transformation projects (IT, regulatory, risk management, operational) and development of new businesses, particularly in responsible investment.

Abstract:
This highly-interactive workshop gives participants a deep dive into AI, machine learning, and emerging technology, through the lens of human-centered design. This approach provides individuals & teams a repeatable and scalable toolkit that can be applied to building better AI-enabled solutions.
The program, delivered by the author of the upcoming AI Product Design Handbook, is intended for those involved with bringing disruptive digital products & services to market. Participants will walk away with a set of practical tools, methods & frameworks focused on applying these technologies to customer and organizational problems.

What You’ll Learn:
Some of the topics & methods introduced in the workshop include:
– Empathy maps and customer journey maps when working with AI
– Ideation tools & strategies for intelligent systems
– Prototyping methods for data-driven applications
– Fundamentals of design thinking, human-centred design, and service design as they apply to AI and machine learning
– Considerations for responsible use of data and avoiding risks of data bias
– Gaining leadership/executive support for a design thinking approach to working with these technologies
(Not part of the description)
We could look to introduce the topic through a short talk as well as offering a much more hands-on workshop. As you know, we’ve done many full-day workshops in the past and they are very hands-on, with great feedback. I’ve evolved the content A LOT over the last 2 years and been able to make it fit better into a half-day ish program.

Ramy is the founder of 1000 Days Out and author of the upcoming AI Product Design Handbook. As the former Managing Director of Design & Strategy for Architech and Head of Innovation for Mattel, he leads diverse teams in the creation of disruptive new digital products, services & platforms.
Ramy and his team at 1000 Days Out work across a wide range of industries with clients including Cadillac Fairview, Apple, Air Canada, Facebook, New Balance, Telus and CIBC.
Ramy teaches Design Thinking at McMaster University and in the Master’s of Engineering, Innovation & Entrepreneurship program at Ryerson University. Ramy is a regular speaker at international events including World Usability Congress, IxDA, FITC, AI Everything, Machine Learning Exchange, AI Business Summit and World Mobile Congress.

Abstract:
Enabling responsible development of artificial intelligent technologies is one of the major challenges we face as the field moves from research to practice. Researchers and practitioners from different disciplines have highlighted the ethical and legal challenges posed by the use of machine learning in many current and future real-world applications.  Now there are calls from across the industry (academia, government, and industry leaders) for technology creators to ensure that AI is used only in ways that benefit people and “to engineer responsibility into the very fabric of the technology.”   Overcoming these challenges and enabling responsible development is essential to ensure a future where AI and machine learning can be widely used. In this talk, we are discussing the latest practical approaches to responsible AI and demonstrate how our latest open source and cloud integrated responsible ML capabilities empower data scientists and developers to understand and improve ML models better.

What You’ll Learn:
In this talk, we will showcase the new release of the Responsible AI Toolbox, which integrates together several Responsible AI tools like Error Analysis and the Interpretability dashboard as well as brand new model assessment and decision-making functionalities such as counterfactual analysis and causal inference. The toolbox is built with two intentions in mind:
– Accelerating the development lifecycle for Machine Learning in a way that implements and applies Responsible AI principles in practice.
– Serving as a collaboration framework for research in the Responsible AI field within Microsoft and beyond.
While different open-source tools have been proposed for assessing fairness, explainability or errors of a machine learning model, these properties are not independent and Machine Learning practitioners may need several of these functionalities to fully identify, diagnose, mitigate issues, and take actions in the real world. These functionalities interact in non-trivial ways and the best results are obtained through interactively analyzing workflows that integrate them in combination, allowing for flexible deep dives. Therefore, we showcase the new, integrated release of the Responsible AI Toolbox.

Minsoo is a Technical Program Manager in the Responsible AI tooling team at Microsoft focusing on building out offerings for Microsoft’s open-sourced Responsible AI Toolbox and its integration into Azure Machine Learning platform. She has contributed to the UX/UI for tooling such as InterpretML, Interpret-text, DiCE (Diverse Counterfactual Examples), Error Analysis and EconML. She has B.A. in Applied Math and B.F.A. in Painting from Brown University and Rhode Island School of Design (RISD). Coming from an interdisciplinary background with experience in building models and applications, analyzing data, and designing UX, she loves to work in the intersection of AI/ML, design, and social sciences to empower ML practitioners to work ethically and responsibly end-to-end.

Rachel is a Technical Program Manager on Microsoft’s Responsible AI Tooling team in Azure Machine Learning. In that role, she is working to productionalize tools for data scientists that enable them to identify, diagnose, and mitigate model errors and take a pro-active approach to fair and responsible modeling. Prior to joining Microsoft, Rachel attended Texas A&M University and received and B.S. in Computer Science, focusing in Business and Cyber Security. Other passions of hers include using AI and technology to act as a catalyst towards solving humanitarian-centered problems for non-profits around the world.

Abstract:
In this workshop, Aparna will present a talk on how to operationalize responsible AI using machine learning observability techniques: notably with explainability – she will look at how to use statistical distance checks to monitor features and model output in production, how to analyze the effects of the changes on models and how to use explainability techniques to determine if issues are model or data related. The first step in better understanding how to manage responsible AI is model transparency. Understand and explain how models arrive at specific outcomes for any cohort of predictions and implement ML Observability into your ML workflow.
Secondly, this workshop will include a panel discussion. Engage with unique perspectives on how to approach responsible AI from an operational standpoint. Join major thought leaders championing responsible, fair, and ethical AI. Armed with a diverse array of educational and professional backgrounds, this discussion will uncover actionable insights into how to approach responsible AI from the ground up.

What You’ll Learn:
– Learn about ML Observability: how to troubleshoot, monitor, and explain models in production
– A deep dive into explainability and common explainability techniques at the global, cohort, and local level
– An understanding of common explainability use cases and how explainability can help detect bias issues in models
– A chat about ML fairness, new research, and how to make sure your models aren’t acting in discriminatory ways – operationalizng AI Ethics

Aparna Dhinakaran is Chief Product Officer at Arize AI; a startup focused on ML Observability. She was previously an ML engineer at Uber, Apple, and Tubemogul (acquired by Adobe). During her time at Uber, she built several core ML Infrastructure platforms, including Michaelangelo. She has a bachelor’s from Berkeley’s Electrical Engineering and Computer Science program, where she published research with Berkeley’s AI Research group. She is on a leave of absence from the Computer Vision Ph.D. program at Cornell University.

Amber Roberts is an astrophysicist and machine learning engineer, who was previously the Head of AI at Insight Data Science. Since then she has been at Splunk in their ML Product Org to build out ML feature solutions as an ML Product Manager. She now joins us at Arize as an ML Sales Engineer looking to help teams across industries build ML Observability into their productionalized AI environments.

Abigail Hing Wen is a New York Times Best Selling Author, rare woman-in-tech leader specializing in artificial intelligence, a filmmaker as well as a wife and mother of two. She writes and speaks about tech, AI ethics, women’s leadership, implicit bias, equity, and transforming culture.
Abigail penned the New York Times best selling novel, Loveboat, Taipei. The companion novel, Loveboat Reunion, follows a girl navigating her interests in artificial intelligence (HarperCollins Jan 25, 2022). Abigail is executive producing the Loveboat, Taipei film adaptation with the creators of the Netflix franchise, To All the Boys I’ve Loved Before. She and her work have been profiled in Entertainment Weekly, Forbes, Fortune, Cosmopolitan, NBCNews, Bloomberg, Google Talk and the World Journal, among others.
Abigail holds a BA from Harvard, where she took coursework in film, ethnic studies and government. She also holds a JD from Columbia and MFA in writing from Vermont College of Fine Arts. In her career in tech, she has negotiated multibillion dollar deals on Wall Street and in Silicon Valley, worked in venture capital and hosted Intel’s Artificial Intelligence podcast featuring leading industry experts including Andrew Ng, Facebook’s Chief AI Scientist Yann LeCun and US Congresswoman Robin Kelly. She also serves with the Partnership on AI.
“One of the most respected voices in fairness and AI.” Forbes.
Abigail lives with her husband and two sons in the San Francisco Bay Area. Her third novel (2023) explores cognitive differences in Silicon Valley and she is writing her fourth novel and a feature film script based in Silicon Valley, as well as producing a girls-in-tech animated series.
For more information: www.abigailhingwen.com
Follow IG: @abigailhingwen and Twitter: @abigailhingwen

Reid Blackman, Ph.D., is the Founder and CEO of Virtue, where he works with companies to integrate ethical risk mitigation into the development, procurement, and deployment of emerging technology products. He is the author of the book Ethical Machines, forthcoming from Harvard Business Review Press, he sat on Ernst & Young’s Artificial Intelligence Advisory Board, is a member of IEEE’s Ethically Aligned Design Initiative, and is the Chief Ethics Officer of the Government Blockchain Association.
Reid’s work has been profiled in The Wall Street Journal and Dell Perspectives and he has contributed multiple pieces to The Harvard Business Review, TechCrunch, and VentureBeat. He is regularly quoted in the news media and he delivers keynote addresses to organizations in industry, non-profit, and academia. Recent venues include The World Economic Forum, SAP, Cannes Lions, AIG, Forbes, NYU Stern School of Business, and Columbia University.
Prior to founding Virtue, Reid was a professor of philosophy at Colgate University and the University of North Carolina, Chapel Hill. His research appeared in numerous prestigious professional journals including the European Journal of Philosophy, The Canadian Journal of Philosophy, and Erkenntnis. He also founded a fireworks wholesaling company and was even a flying trapeze instructor. He received his B.A. from Cornell University, his M.A. from Northwestern University, and his Ph.D. from The University of Texas, Austin.

Abstract:
This session discuses Algorithmic Bias in Human Resources including AI and ML projects. It provides a convincing case that the bias and fairness issues that have been identified are grounded in business strategy decisions. Importantly, the session takes a practitioner perspective, viewing Human Resource technology project implementation from a Project Managers and middle managers point of view. The session starts by identifying the symptoms of the problem that are discussed in public press articles, and then completes a deep dive into the root causes of the issues, by unpacking layers of business and strategy decisions that have led to the issues that have been discovered after system implementation. In this session we also identify strategies and tactics and that can be used by Project Managers and middle management in developing responsible Human Resource algorithms, including AI and ML projects.

What You’ll Learn:
Root causes of bias and fairness issues in Human Resource AI and ML projects

​Ushnish Sengupta is an award-winning instructor at the Schulich School of Business at York University, and a PhD candidate at OISE at the University of Toronto. He has an Industrial Engineering and MBA education, worked in various private sector and public sector organizations as an Information Technology Project Manager and Product Manager. Ushnish’s research interests include Entrepreneurship, Blockchain, Artificial Intelligence, Open Data, Diversity, and the Social and Environmental impact of technology projects.

Abstract:
In this session we will explore the differences between ModelOps and MLOps, and what this means to your ability to manage model risk. As more and more models are deployed in the business environment, of both traditional and AI/ML models, the risk to your business increases. Whether the risk is direct financial impact to your business, or additional scrutiny by governing agencies, we will explore how a robust ModelOp solution can help you reduce that risk and enable your business to be successful.

What You’ll Learn:
Shows real-world examples of how to de-risk models, with live talk and Q&A.

At ModelOp, Jon works with F500 companies to understand their challenges with AI and decision-making model operations (ModelOps) and demonstrate how ModelOp Center helps solve their ModelOp technical, operational, and business challenges. Jon has 15 years of experience as a technologist and sales engineer. He has worked with AI technology products for the last four years, complementing his many years of experience with digital business automation products.

Abstract:
Deep AutoViML is a powerful new deep learning library with a very simple design goal: Make it as easy as possible for novices and experts alike to experiment with and build tensorflow.keras preprocessing pipelines and models in as few lines of code as possible.
Deep AutoViML will enable data scientists, ML engineers and data engineers to fast prototype tensorflow models and data pipelines for MLOps workflows using the latest TF 2.4+ and keras preprocessing layers. You can now upload your saved model to any Cloud provider and make predictions out of the box since all the data preprocessing layers are attached to the model itself!
In this webinar, we will discuss the problems that Deep AutoViML can solve, its architecture design and demo how to build powerful TF.Keras models on structured data, NLP and Image data domains.

What You’ll Learn:
​As an ML instructor on Coursera and current ML consultant for Google, you will learn how to apply deep learning models to structured data, NLP and image problems with Ram Seshadri using his latest deep learning AutoML library called Deep AutoViML

​As a popular ML instructor on Coursera and currently a Program Manager for Google, Ram Seshadri has worked in Machine Learning for more than 5 years. Previously, Ram served as a data scientist at Morgan Stanley and as an Instructor at General Assembly. He is the author of the popular Python AutoML libraries, AutoViz and AutoViML.

Abstract:
In recent years, we have seen astonishing leaps in the application of machine learning in various industries. However, as the complexity of machine learning models and the size of the datasets increase, experimentation with these models and productionizing them also become more complex and time-consuming! To overcome these challenges and facilitate the adoption of these models in the industry, various solutions are proposed by the ML community over the last few years. In this workshop, we walk you through some of these solutions and show you useful practices to overcome the aforementioned challenges! More specifically, we show you how you can supercharge your machine learning experimentation pipeline with tools like PyTorch Lightning and DVC and make your path to production smoother and faster using Kubeflow and its add-ons.

What You’ll Learn:
In this workshop, you will learn about useful tools and practices which can help you to supercharge the whole machine learning cycle from R&D experimentation to production. More especially, our workshop is split into two parts:
1) In the first half of the workshop, we will focus on supercharging the R&D experimentation cycle. Here we will show you how you can make your experimentation pipeline faster, more memory efficient and reproducible using tools like PyTorch Lightning and DVC
2) In the second half of the workshop, we will focus on the productionization side and explain how tools such as KubeFlow Pipelines, Kserve, SeldonIO, aws-cdk and many more that can be used to make the path to production easier, faster, and smoother! We hope that by learning about these practices and tools, you can unleash the untapped potential of complex ML models for your own problems!

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. 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.

Chris is a machine learning engineer at Georgian’s R&D team where he builds scalable machine learning pipelines and cloud infrastructure. Prior to this, Chris was an applied research scientist intern at Georgian focusing on AutoML and domain adaptation for NLP. He contributed to open-sourced toolkits based on these applied research areas.

Kyryl has over seven years of experience in the field of Machine Learning. For the bulk of this career, he has helped build machine learning startups, from inception to a product. He has also developed expertise in choosing and implementing state-of-the-art deep learning architectures and large-scale solutions based on them.

Abstract:
Machine learning models are conventionally trained under the premise that the training and the real-world (i.e., both source and target) data are sampled from the same distribution. This assumption may potentially lead to predictive problems in dynamic environments where the distribution of data changes over time. This is known as dataset shift.
In most real-world situations, machine learning models have to cope with dataset shift after deployment. The shift in the distribution could be dramatic for unexpected reasons, e.g., the breakout of COVID-19 pandemic or cyber attacks.
This tutorial will present:
– The principles behind data shift.
– Strategies for detecting dataset shift.
– Adaptation techniques.
– Advanced topics in data shift and hands-on practice

What You’ll Learn:
This tutorial aims to provide a comprehensive understanding of dataset shift and explore potential remedies in the face of distribution shift. The learning outcomes for the participants are:
– Understand characteristics of dataset shift by real-world examples and relationship between input features and output changes
– Learn different types of dataset shifts and the terminologies
– Learn approaches to detect dataset shifts
– Learn how to adapt an existing ML model to shifts using transfer learning and active learning rather than retraining on the target task from scratch
– Get acquainted with computational techniques and libraries through hands-on practices​

Sedef Akinli Kocak (co-presenter) is an academic industry R&D partnership and project manager in the area of AI/ML and is an accomplished researcher in the area of ICT for Sustainability and Advance Analytics. She has a Ph.D. in Environmental Applied Science and Management from Data Science Lab at Ryerson University. She is currently with the Vector Institute as an AI Project Manager. She is also a part-time lecturer and supervisor in the Data Science and Analytics Program at Ryerson University.

Ali Pesaranghader (co-presenter) is an AI Research Scientist at LG Toronto AI Lab, and a former Sr. Research Scientist at the Canadian Imperial Bank of Commerce (CIBC) with primary research interests in adaptive learning, data stream mining, natural language processing, and transfer learning. Ali obtained his Ph.D. in Computer Science with a focus on Adaptive Machine Learning at the University of Ottawa in 2018.

Mehdi Ataei (co-presenter) is a research affiliate at the Vector Institute with a Ph.D. in Computational Physics from University of Toronto. He is currently a researcher at Autodesk’s Simulation, Optimization, and Systems group. His current research is focused on computational physics, applied mathematics, topology optimization, and machine learning.