RE•WORK Summits, Montreal 2019

Back

Exhibition Area & Refreshment Breaks

REGISTRATION & LIGHT BREAKFAST

08:00 AM 09:00 AM

Registration will open from 8:00am, please have your registration details to hand on your device. A light breakfast, tea and coffee will be available for you to help yourself!

Back

Deep Learning Summit

Responsible AI Summit

WELCOME

09:00 AM 09:15 AM

The host for the Deep Learning Summit is Hessam Amini, PhD Student at Concordia University. The host for the Responsible AI Summit morning sessions is Saif Mohammad, Senior Research Scientist at the National Research Council Canada.

Speakers

Hessam Amini

Ph.D. Student

Concordia University

Saif Mohammad

Senior Research Scientist

National Research Council Canada

Back

Deep Learning Summit

Towards Combining Statistical Relational Learning and Graph Neural Networks - HEC Montreal

09:15 AM 09:35 AM

Developing statistical machine learning methods for predictions on graphs has been a fundamental problem for many applications such as semi-supervised node classification and link prediction on knowledge graphs. Such problems have been extensively studied by traditional statistical relational learning methods and recent graph neural networks, which are attracting increasing attention. In this talk, I will introduce our recent efforts to combine the advantages of both worlds for prediction and reasoning on graphs. I will introduce our work on combining conditional random fields and graph neural networks for semi-supervised node classification (Graph Markov Neural Networks, ICML'19) and also recent work on combining Markov Logic Networks and knowledge graph embedding (Probabilistic Logic Neural Network, in submission) for reasoning on knowledge graphs. Key Takeaways: (1) Reasoning is important for decision making and missing in existing deep learning systems (2) In this presentation, I will introduce our recent works on combing traditional statistical relational learning and recent graph neural networks for relation reasoning (3) A more general direction for the future is how to combine traditional symbolic logic rule based approaches with deep neural networks for reasoning?

Speakers

Jian Tang

Assistant Professor

HEC Montreal

Back

Responsible AI Summit

AI Ethics: A Deflationary Yet Cautionary Tale - UNIVERSITE LAVAL

09:15 AM 09:40 AM

The field of “AI ethics” is effervescent. While some scholars wonder whether the interests of artificial agents endowed with “general intelligence” will be aligned with human values, others want us to focus on the immediate challenges raised by the weak and narrow AI systems that are currently being developed. Civil society actors, policy makers, elected officials, international organizations and some business leaders reflect on the best ways to regulate AI. In this talk, I will first suggest that what I call “inflationary” thinking about the future development of AI and its likely impacts on human life hinders a sober reflection on the risks created by current AI technologies and the ways to mitigate them. Second, I will attempt to show how a deflationary view can help addressing the “explainability problem” plaguing decision-making or predictive machine learning algorithms.

Speakers

Jocelyn Maclure

Professor

Université Laval

Back

Deep Learning Summit

Few-Shot Learning: Thoughts On Where We Should Be Going - GOOGLE BRAIN

09:35 AM 10:05 AM

Few-shot learning is the problem of learning new tasks from little amounts of labeled data. This is achieved by performing a form of transfer learning, from the data of many other existing tasks. This topic has gained tremendous interest in the past few years, with several new methods being proposed each month. In this talk, I suggest we take a step back, look at what we have achieved and, most importantly, consider where this research should be going next. Key Takeaways: 1) We can achieve surprisingly high accuracy on current few-shot learning benchmarks, without using the labels. 2) The most popular few-shot learning methods aren't robust to training on heterogeneous sets of tasks. 3) None of these few-shot learning methods dominate in all settings (e.g. of number of shots).

Speakers

Hugo Larochelle

Montreal

Google Brain

Back

Responsible AI Summit

Social Inclusion in the AI Pipeline - POLYTECHNIQUE MONTREAL

09:40 AM 10:05 AM

Key Takeaways: 1) risks and biases associated with the development of AI, 2) development of an ethical framework, 3) process reflecting inclusiveness and diversity.

Speakers

Nathalie de Marcellis

Full Professor

Polytechnique Montréal

Back

Deep Learning Summit

Building Knowledge For AI Agents With Reinforcement Learning - DEEPMIND

10:05 AM 10:35 AM

Reinforcement learning allows autonomous agents to learn how to act in a stochastic, unknown environment, with which they can interact. Deep reinforcement learning, in particular, has achieved great success in well-defined application domains, such as Go or chess, in which an agent has to learn how to act and there is a clear success criterion. In this talk, I will focus on the potential role of reinforcement learning as a tool for building knowledge representations in AI agents whose goal is to perform continual learning. I will examine a key concept in reinforcement learning, the value function, and discuss its generalization to support various forms of predictive knowledge. I will also discuss the role of temporally extended actions, and their associated predictive models, in learning procedural knowledge. Finally, I will discuss the challenge of how to evaluate reinforcement learning agents whose goal is not just to control their environment, but also to build knowledge about their world.

Speakers

Doina Precup

DeepMind

Back

Responsible AI Summit

AI for Humanity: How AI Could Benefit Us All- MILA

10:05 AM 10:35 AM

This will be a TedX format: IA could be scary, we would like to tell another story: how ai could benefit our society, exemples of real applications at Mila, call for actions

Speakers

Myriam Côté

Director, AI for Humanity

MILA

Back

Exhibition Area & Refreshment Breaks

COFFEE

10:35 AM 11:20 AM

Help yourself to tea, coffee and refreshments in the foyer and make sure you check out the exhibitors!

Back

Deep Dive Sessions

DEEP DIVE: Introduction to Deep Reinforcement Learning - MILA

10:40 AM 11:15 AM Tchaikovsky Room

In this talk, I will introduce the field of Reinforcement Learning (RL). Furthermore, I will discuss various approaches to solve the RL problem which includes policy gradient methods, value based learning and model based learning. I will describe how deep learning is effective in all these solutions to RL.

Speakers

Mohammad Amini

Graduate Research Assistant

MILA

Back

Deep Dive Sessions

DEEP DIVE: Neural Networks for Lawyers and Other Non-Technical Professionals - MIT

10:40 AM 11:15 AM Vivaldi Room

In this presentation, I will explain typical machine learning workflows and neural networks in technical detail from scratch. We will begin by defining concepts like "artificial intelligence" and "machine learning" before discussing general machine learning concepts, logistic regression, and neural networks. This presentation is suitable for audiences with no technical background.

Speakers

Jonathan Frankle

Ph.D Candidate

MIT

Back

Deep Learning Summit

The Lottery Ticket Hypothesis: Finding Sparse, Trainable Neural Networks - MIT

11:20 AM 11:40 AM

Neural network pruning techniques can reduce the parameter counts of trained networks by over 90%, decreasing storage requirements and improving computational performance of inference without compromising accuracy. However, until recently, it was believed that the sparse architectures produced by pruning were difficult to train from the start, which would similarly improve training performance. In this talk, I will discuss a series of experiments that contradicted this received wisdom, showing that sparse neural networks are indeed trainable, provided they are given the same initialization they received at or near the start of training. This observation culminates in a new conjecture about opportunities to improve neural network training: the lottery ticket hypothesis. Key Takeaways: 1. The neural networks that we train in practice are much larger than they have to be to learn. 2. There is an unexploited opportunity to substantially improve the performance of practical neural networks. 3. Neural networks are objects worthy of scientific study.

Speakers

Jonathan Frankle

Ph.D Candidate

MIT

Back

Responsible AI Summit

Ethical Issues in NLP - UNIVERSITY OF TORONTO

11:20 AM 11:45 AM

Ethical issues in natural language processing relate both to issues in the applications of NLP and to problems of bias and discrimination within the systems themselves. I will talk about virtuous and evil applications of NLP, and I will suggest some characteristics of the latter. I’ll then describe the problem of bias in learning from linguistic data, especially for minority groups and for languages other than English that have limited resources, and the bias in word-embeddings that results from linguistic data and that affects even second-order uses of the data. Key Takeaways: 1. Natural language processing has great potential for unethical uses. 2. Bias in NLP systems may arise from the structure of the research milieu in which they are developed. 3. NLP systems may learn both direct and latent bias from real-world data.

Speakers

Graeme Hirst

Ethical Issues in NLP

University of Toronto

Back

Deep Dive Sessions

DEEP DIVE: AI, Deep Learning, Optimization: How to Frame and Solve Real World Problems Effectively to Get the Most Business Benefits - ALPSANALYTICS.COM

11:20 AM 12:00 PM Vivaldi Room

AI, Deep Learning, and Machine Learning (ML) have been receiving much attention in recent years due to their big potentials to deliver significant business benefits. In this talk, we will discuss the interactions and relationships of AI, deep learning, ML, optimization models, the utilities of each and their and best uses in different business contexts. We will look into how to frame your business problems properly, ways to select the more effective models, and how to implement efficient solutions that highlight and optimize the most critical business tradeoffs in line with your (sometimes subtle and unspoken) business priorities and strategies. We will include examples from various industries for illustration. This presentation can benefit both data scientists and business executives who are interested in leveraging machine learning to gain business advantages for their companies. Takeaways for the audience include: - better understanding of AI, deep learning, machine learning, optimization modeling concepts, their common roots, and where they can be best applied - a framework that helps guide the definition, framing and formulation of the business problems to get the model solution that can best support your business priorities - industry examples that help demonstrate how it's done

Speakers

Yanqi Xu

VP of Data Science and AI

AlpsAnalytics.com

Back

Deep Learning Summit

Generalizing The Lottery Ticket Hypothesis Across Datasets and Optimizers and Beyond Supervised Image Classification - FACEBOOK AI RESEARCH

11:40 AM 12:00 PM

The success of lottery ticket initializations (Frankle and Carbin, 2019) suggests that small, sparsified networks can be trained so long as the network is initialized appropriately. This phenomenon is intriguing and suggests that initialization strategies for DNNs can be improved substantially, but a number of open questions remain. Do winning tickets contain generic inductive biases for training or are they just overfitted to a particular problem? Is the lottery ticket phenomenon simply an artifact of image classification or is it present in other domains as well? In this talk, I will discuss recent work to address both of these critical questions.

Speakers

Ari Morcos

Research Scientist

Facebook AI Research

Back

Responsible AI Summit

Finding Bias In Your Systems - SHOPIFY

11:45 AM 12:10 PM

In a large organization, it's easy for bias to live in the form of machine learning models, but also from hand-coded rules. This talk will explore the ways of using machine learning from an analytical perspective and detecting where bias can live in different types of systems. Key Takeaways: 1. Bias lives in everything we do. 2. It's extremely hard to determine what is fair. 3. How can you better spot potential bias and its implications.

Speakers

Dexter Fichuk

Data Scientist (NLP)

Shopify

Back

Deep Learning Summit

Outlines, Explanation, and Deflecting Adversarial Examples - GOOGLE BRAIN

12:00 PM 12:20 PM

Adversarial examples have been a topic of interest since they were first discovered. They illustrate an interesting failure mode in neural networks. Many researchers have attempted to solve this problem by creating detection methods. However these mechanisms are inevitable broken shortly after release by a defense aware attack. One approach to getting ahead of this cycle is to create a model that when adversarially attacked, yields inputs that resemble the target class, thereby deflecting the attack. I will talk about recent work done under Geoff Hinton at google brain that takes such an approach. Key Takeaways: 1. Adversarial examples are hard to beat. 2. Deflecting attacks is a good idea. 3. Capsule reconstruction networks do a good job of this.

Speakers

Nicholas Frosst

Research Scientist

Google Brain Toronto

Back

Deep Dive Sessions

DEEP DIVE: How The Pharmaceutical Industry Is Leveraging Deep Learning - NOVARTIS

12:00 PM 12:40 PM Vivaldi Room

A case study of NOVARTIS, from conceptual design to enterprise level implementation.Pharmaceutical industry is very rich in “unstructured”, “semi-structured” data sets, generated throughout the process of drug discovery, clinical trial, manufacturing, testing and commercial launches in the form of pdfs, word document, databases etc.. Business functions are well equipped to manually/semi-manually mine the information, extract insights and use in their business process. However, typical issues around limitation with respect to understanding causal effects, learnings from prior experiments/mistakes, siloes data connections are vast area of opportunity. The talk sites specific use cases touching 1000’s of business users using AI and ML-based model insights. This focuses on data science and software engineering learnings which has made deep learning based initiatives successful. The talk summarizes the gaps in research which can enable the pharmaceutical industry to be more embracing to advanced technologies. Key Takeaways: 1. Pharma data is rich and less is more in AI. 2. Novartis is leading data science in a true sense. 3. How to scale data science products in enterprise.

Speakers

Nimit Jain

Director, Data Science

Novartis

Back

Responsible AI Summit

Agnostic Data Debiasing Through a Local Sanitizer Learnt from an Adversarial Network Approach - UNIVERSITE DU QUEBEC A MONTREAL

12:10 PM 12:35 PM

The widespread use of automated decision processes in many areas of our society raises serious ethical issues concerning the fairness of the process and the possible resulting discriminations. In this work, we propose an approach called GANsan whose objective is to prevent the possibility of any discrimination, )direct or indirect) based on a sensitive attribute by removing the attribute itself as well as the existing correlations with the remaining attributes. Our sanitization algorithm GANsan is partially inspired by the framework of generative adversarial networks (in particular the Cycle-GANs), which offers a flexible way to learn a distribution empirically or to translate between two different distributions.
In contrast to prior work, one of the strengths of our approach is that the sanitization is performed in the same space as the original data by only modifying the other attributes as little as possible and thus preserving the interpretability of the sanitized data. As a consequence, once the sanitizer is trained, it can be applied to new data, such as for instance, locally by an individual on his profile before releasing it. We explore the utility preserved by the sanitization by conducting on a dataset various experiments, which we believe represents possible real world use case of the sanitization procedure and the sanitized data. Our observations bring forward new research questions which we will briefly introduce.

Speakers

Rosin Claude Ngueveu

PhD Student

Université du Québec à Montréal

Back

Deep Learning Summit

Learning to Act More Like Humans, and Learning with Less Data - POLYTECHNIQUE MONTREAL

12:20 PM 12:40 PM

Humans recognize objects, read and understand language, observe the world and control their bodies with incredible skill. Furthermore, we learn to do these things with very little training data. In this talk I survey recent work from my group and collaborators on deep learning techniques for these settings. I'll focus on the theme of learning with less labelled data using techniques such as multi-task learning, encoder-decoder architectures and a new technique for visual imitation learning with reinforcement learning (VIRL). I'll show how this new VIRL method permits a humanoid robot living in a software simulation to learn new behaviours by simply watching video of a desired activity.

Speakers

Christopher Pal

Full Professor

Polytechnique Montréal, Mila & Element AI

Back

Exhibition Area & Refreshment Breaks

LUNCH

12:35 PM 01:40 PM

A hot, 3-course, lunch buffet will be served in the foyer area. A great time for networking and to get to know your fellow participants or you can join the Lunch & Learn Session taking place in the Deep Dive Track and take a seat at one of the tables to hear more from the speakers and exhibitors.

Back

Deep Dive Sessions

DEEP DIVE: Lunch and Learn

12:45 PM 01:45 PM Tchaikovsky Room & Vivaldi Room

Join speakers and exhibitors to explore various AI & DL topics over lunch! You can find Huawei, CBC, Facebook AI Research & Axionable in Tchaikovsky and Shopify, Deeplite, Deloitte, Umaneo, & Valohai in Vivaldi.

Speakers

Jason Cornell

Manager, Machine Learning, Recommendation Systems

CBC

Dexter Fichuk

Data Scientist (NLP)

Shopify

Jeff Lui

Director of Artificial Intelligence

Deloitte

Jun Luo

Director of Decision Making and Reasoning

Huawei, Noah's Ark Lab

Ari Morcos

Research Scientist

Facebook AI Research

Back

Responsible AI Summit

PM Host Introduction

01:35 PM 01:40 PM

The host for the Responsible AI Summit afternoon sessions is Valentine Goddard, Founder of the AI Impact Alliance.

Speakers

Valentine Goddard

Founder

AI Impact Alliance

Back

Deep Learning Summit

Productionizing Deep Learning Models - EXPEDIA

01:40 PM 02:00 PM

Lessons learned from productionizing deep learning models. How to go about improving existing production models. How to organize the infra and modeling efforts. Automating continuous training. Online vs offline metrics.

Speakers

Nicholas Léonard

Senior Data Scientist

Expedia Group

Back

Responsible AI Summit

Fairwashing: The Risk of Rationalization - UNIVERSITÉ DE MONTRÉAL

01:40 PM 02:00 PM

Black-box explanation is the problem of explaining how a machine learning model -- whose internal logic is hidden to the auditor and generally complex -- produces its outcomes. Current approaches for solving this problem include model explanation, outcome explanation as well as model inspection. While these techniques can be beneficial by providing interpretability, they can be used in a negative manner to perform fairwashing, which we define as promoting the false perception that a machine learning model respects some ethical values. In particular, we demonstrate that it is possible to systematically rationalize decisions taken by an unfair black-box model using the model explanation as well as the outcome explanation approaches with a given fairness metric. Our solution, LaundryML, is based on a regularized rule list enumeration algorithm whose objective is to search for fair rule lists approximating an unfair black-box model. We empirically evaluate our rationalization technique on black-box models trained on real-world datasets and show that one can obtain rule lists with high fidelity to the black-box model while being considerably less unfair at the same time.

Speakers

Alain Tapp

Full Professor

Université de Montréal

Back

Deep Dive Sessions

DEEP DIVE: Algorithmic Profiling & Illegal Discrimination: A Cross-Industry Analysis - MONTREAL AI ETHICS INSTITUTE

01:45 PM 03:20 PM Vivaldi Room

Workshop and mini meet up. *Laptops are recommended (but not required)*

Speakers

Mirka Snyder Caron

Associate

Montreal AI Ethics Institute

Back

Deep Learning Summit

AI for Self Driving - From Research to Production - UBER ATG

02:00 PM 02:20 PM

At Uber ATG R&D centre, we are working on advanced state-of-the-art AI models for solving a large range of problems in self driving - perception and prediction, motion planning, mapping and localization, sensor simulation, and more. All that work is publicly available through academic conferences and venues. In this talk I will cover some exciting recent advances and also discuss the path to production - how we go from research prototypes to deployed systems on vehicle. Key Takeaways: how self driving works, Uber's R&D Lab - what and why, new tech coming out of Uber ATG R&D.

Speakers

Inmar Givoni

Senior Autonomy Engineering Manager

Uber ATG

Back

Responsible AI Summit

Learning Fair Rule Lists - UNIVERSITE DU QUEBEC A MONTREAL

02:00 PM 02:25 PM

The widespread use of ML models in high stakes decision-making systems raises many ethical issues concerning fairness and interpretability. While the research in these domains is booming, very few works have addressed these two issues simultaneously. To solve this shortcoming, we propose FairCORELS, a supervised learning algorithm whose objective is to learn at the same time fair and interpretable models. FairCORELS is a multi-objective variant of CORELS, a branch-and-bound algorithm, designed to compute accurate and interpretable rule lists. We also made additional contributions regarding search strategies for improving the performances of FairCORELS.

Speakers

Ulrich Aivodji

Postdoctoral Researcher

Université du Québec à Montréal

Back

Deep Learning Summit

New Optimization Perspectives on Generative Adversarial Networks - SAIT AI LAB

02:20 PM 02:40 PM

Generative adversarial networks (GANs) form a generative modeling approach known for producing appealing samples, but they are notably difficult to train. One common way to tackle this issue has been to propose new formulations of the GAN objective. Yet, surprisingly few studies have looked at optimization methods designed for this adversarial training. In this work, we cast GAN optimization problems in the general variational inequality framework and investigate the effect of the noise in this context. We thereby propose to use a variance reduced version of the extragradient method, which shows very promising results for stabilizing the training of GANs. Key Takeaways: 1) variance is a bigger problem in solving games (such as GANs) than standard minimization in machine learning; 2) Our method, SVRE, combines extragradient and variance reduction to solve games (like GANs) with more stability; 3) It's a new exciting direction of investigation for GANs work!

Speakers

Simon Lacoste-Julien

VP Lab Director/Associate Professor

SAIT AI Lab/Université de Montréal

Back

Responsible AI Summit

The Evolving Relationship Between Ethics and Safety in AI - FUTURE OF LIFE INSTITUTE

02:25 PM 02:55 PM

As AI systems support more distant training supervision, wider action space ranges, more powerful optimization, more general multi-task algorithms, and more meta specification, the dependencies between AI ethics and AI safety become more pronounced. We discuss why architectural foresight for both safety and ethical concerns is both possible and warranted, and how technical architecture, professional ethics, safety engineering, and stakeholder interests are increasingly inextricably bound together. Through the paradigm of technical value alignment we explore ways to predict new classes of misalignment bugs and ways to prevent or mitigate them as early as possible.

Speakers

Richard Mallah

Director of AI Projects

Future of Life Institure

Back

Deep Dive Sessions

DEEP DIVE: Consumer Airfare Prediction and other Big Data AI Challenges at Hopper- HOPPER

02:25 PM 02:55 PM Tchaikovsky Room

At Hopper, we help consumers make smarter decisions about booking their air & hotel at the right time for the right price. This session is a deep dive into airfare pricing, including data visualization, how we enable millions of personalized conversations with our users focused on travel flexibility and alternative suggestions, building trust with consumers around data, our price prediction algorithms, tackling big data AI challenges in a low-frequency industry and other problems that keep us up at night.

Speakers

Patrick Surry

Chief Data Scientist

Hopper

Back

Deep Learning Summit

PANEL: How Can We Overcome Challenges To Fully Leverage the Oportunities of GANs?- MILA, UNIVERSITY OF MASSACHUSETTS, SAIT AI LAB/UNIVERSITE DE MONTREAL, CONCORDIA UNIVERSITY

02:40 PM 03:20 PM

The purpose of this panel discussion is to consider what challenges and obstacles are currently in play which are halting the progression, development and more widespread application of GANs. Particular topics include the challenges/solutions relating to the type of data, computing resources, software/packages.

Speakers

Alexia Jolicoeur-Martineau

PhD Student

MILA

Simon Lacoste-Julien

VP Lab Director/Associate Professor

SAIT AI Lab/Université de Montréal

Bharatendra Rai

Professor

University of Massachusetts

Prashant Raina

Ph.D Candidate

Concordia University

Back

Responsible AI Summit

Adversarial Machine Learning: Ensuring Security of ML Models and Sensitive Data - VECTOR INSTITUTE

02:55 PM 03:20 PM

As machine learning (ML) has seen dramatic growth in industrial applications, so have we begun to question what trust and security mean in the context of ML. I will give an overview of adversarial ML as a research area and explore some of the attack and defense strategies that have been developed in recent literature. In particular, I will showcase some of the use cases and implementations of differential privacy and how it can be used to protect sensitive data used for training ML models.

Speakers

Christopher Choquette Choo

Vector Institute

Back

Exhibition Area & Refreshment Breaks

COFFEE

03:20 PM 04:00 PM

Help yourself to tea, coffee and refreshments in the foyer and make sure you check out the exhibitors!

Back

Deep Dive Sessions

DEEP DIVE: Talent and Talk

03:25 PM 04:00 PM Tchaikovsky Room

Find out who's hiring!

Back

Deep Learning Summit

Responsible AI Summit

Deep Learning and Cognition - UNIVERSITÉ DE MONTRÉAL

04:00 PM 05:00 PM

Neural networks and deep learning have been inspired by brains, neuroscience and cognition, from the very beginning, starting with distributed representations, neural computation, and the hierarchy of learned features. More recently, it has been for example with the use of rectifying non-linearities (ReLU) - which enables training deeper networks - as well as the use of soft content-based attention - which allow neural nets to go beyond vectors and to process a variety of data structures and led to a breakthrough in machine translation. Ongoing research is now suggesting that brains may use a process similar to backpropagation for estimating gradients and new inspiration from cognition suggests how to learn deep representations which disentangle the underlying factors of variation, by allowing agents to intervene in their environment and explore how to control some of its elements. Key Takeaways: Lots of interesting challenges toward human level AI.

Speakers

Yoshua Bengio

Université de Montréal

Full Professor

Back

Exhibition Area & Refreshment Breaks

CONVERSATION & DRINKS

05:00 PM 06:00 PM

Join us in the foyer area and grab a drink to celebrate the end of Day 1 and continue to network with attendees.

Back

Exhibition Area & Refreshment Breaks

END OF SUMMIT

06:00 PM 06:05 PM

Thank you for attending! Doors open at 8am tomorrow morning. See you tomorrow for more discussions, presentations and networking! And please remember to bring your badge! :)

Back

Exhibition Area & Refreshment Breaks

REGISTRATION & LIGHT BREAKFAST

08:00 AM 09:00 AM

Registration will open from 8:00am, please have your registration details to hand on your device. A light breakfast, tea and coffee will be available for you to help yourself!

Back

Deep Learning Summit

Responsible AI Summit

WELCOME

09:00 AM 09:15 AM

The host for the Deep Learning Summit is Hessam Amini, PhD Student at Concordia University. The host for the Responsible AI Summit is Valentine Goddard, Founder of the AI Impact Alliance.

Speakers

Hessam Amini

Ph.D. Student

Concordia University

Valentine Goddard

Founder

AI Impact Alliance

Back

Deep Learning Summit

Visual Perception for Better Understanding of Sports Games - SPORTLOGIQ

09:15 AM 09:35 AM

This talk will be about some of the main challenges in developing and deploying deep learning algorithms at scale, i.e. processing more than 60,000 sports videos which are coming from different sources, and touches the issues from proper problem definition to generalization and robustness in the deep learning models.

Speakers

Bahar Pourbabaee

Machine Learning Team Lead

Sportlogiq

Back

Responsible AI Summit

AI and The Evolution of Work - KINDRED.AI

09:15 AM 09:35 AM

The replacement of human workers with automata has been a topic of debate since the time of Aristotle. Ancient Rome had been known to ban labour-saving inventions in fear it would prevent the poor from earning their bread. It has motivated radical factions of textile workers in the 19th century to destroy machinery as a form of protest. Today there is a growing belief that AI will soon make skilled workers obsolete; some even calling it the “AI catastrophe”. Are there reasons to believe this? What tendencies are baked-in to the technological systems we are building that help us understand their direction? How can we make better sense of the evolution of work?

Speakers

Julian Villella

SW/ML Manager, Robotic Grasping

Kindred.ai

Back

Deep Learning Summit

Version Control for Jupyter Notebooks - VALOHAI

09:35 AM 09:55 AM

While Jupyter Notebook is an excellent tool for experimenting with machine learning models on early stage, there is still a big leap to enable company-wide, transparent, and production-scale ML development.During Toni’s presentation, you’ll learn how to support machine learning exploration, experimentation, and deployment in Jupyter Notebooks using Jupyhai. Jupyhai is Valohai’s Jupyter Notebook extension letting data scientists run their experiments asynchronously on the cloud while having automatic versioning for code, data, hyperparameters, and more. Time to step up your Jupyter game to the next level!

Speakers

Toni Perämäki

COO

Valohai

Back

Responsible AI Summit

Artificial Intelligence & Ethical Deployment - DELOITTE

09:35 AM 09:55 AM

Artificial Intelligence (AI) is poised to permeate and potentially disrupt many, if not all, industries. With new technology comes a new set of challenges and considerations from an ethical perspective. Join Jeff Lui in this session, where he will discuss the impact of AI and the questions that must be considered when developing any responsible AI implementation and the ethical concerns through the lens of his AI Ethics Funnel. He will also discuss his career journey through the field of AI and how he continues to be involved in the wider industry. Key Takeaways: 1. Cognitive Diversity critical for AI & Ethics, 2. Multidisciplinary Team Composition recommended for AI deployments, 3. AI Ethics must be a shared conversation with all stakeholders.

Speakers

Jeff Lui

Director of Artificial Intelligence

Deloitte

Back

Deep Learning Summit

Pushing the Energy Efficiency Frontier with Deep Learning - BRAINBOX AI

09:55 AM 10:15 AM

Deep learning will be imperative in optimizing our use of energy and ultimately supporting our global fight against climate change. More specifically, deep learning can be used to predict the energy consumption in real time of one of the world’s biggest energy consumers, buildings. Equipped with these predictions, it becomes possible to shift energy loads by creating thermal batteries within buildings, resulting in a more efficient use of energy resources and less waste. This talk will also introduce the concept of swarm intelligence applied to buildings and the ways in which it can benefit grid operators and ultimately the planet as a whole.

Speakers

Jean-Simon Venne

CTO

BrainBox AI

Back

Responsible AI Summit

Fireside Chat: Responsible AI at Target - TARGET & AI IMPACT ALLIANCE

09:55 AM 10:15 AM

Discover more about how Frankie Cancino’s team at Target are designing models with Responsible AI topics in mind!

Speakers

Frankie Cancino

Senior Engineer & Data Scientist

Target

Valentine Goddard

Founder

AI Impact Alliance

Back

Exhibition Area & Refreshment Breaks

COFFEE

10:15 AM 11:00 AM

Help yourself to tea, coffee and refreshments in the foyer and make sure you check out the exhibitors!

Back

Deep Dive Sessions

DEEP DIVE: Investing in AI Startups: Challenges & Opportunities

10:20 AM 10:55 AM Tchaikovsky Room

Speakers

Martin Duchaîne

Director

Défi Montréal

Sergio Escobar

Partner

BCF Ventures

Kris Lachance

Partner Associate

Real Ventures

Back

Deep Learning Summit

Deep Learning for Conversational AI - BOREALIS.AI

11:00 AM 11:25 AM

In this talk, I will review recent progress in building conversational AI to complete tasks and hold general conversations. Today's dialogue systems are the fruit of decades of progress in linguistics, compute power, and machine learning. In particular, modern machine learning techniques like deep learning hold the promise of accelerating the development of dialogue systems and achieving more complex interactions but they also entail many challenges including controllability, evaluation, and data efficiency. I will describe these challenges as well as some of the promising solutions that will bring the next generation of dialogue systems. Key Takeaways: 1) Conversational agents can now combine casual chatting capabilities and task-achieving capabilities (e.g., playing songs, informing about the weather,...), 2) Big language models such as BERT and GPT-2 are giving a great boost in performance and the gap towards good conversational agents seems to be shrinking, 3) conversational AI is making a foray not only in customer service but also in education, retail, and finance.

Speakers

Layla El Asri

Research Team Lead

Borealis AI

Back

Responsible AI Summit

AI Governance in Banking: The Role and Challenges of Deploying AI in Banking - NATIONAL BANK OF CANADA

11:00 AM 11:25 AM

In the last ten years, we have witnessed the evolution of the concept of Artificial Intelligence into a notion that has captured the collective imagination with its promises of transformation. Deep learning and its flagship applications such as Alpha Go, self-driving cars, image recognition, automatique translation, etc., have created a lot of expectation in terms of its potential to bring fundamental transformation to various economic sectors. Historically, the banking sector has always been a large ship that is hard to steer into new directions. In terms on innovation and technology, they have been less agile than Fintech startups. Now that banks are beginning to catch up, alongside technological challenges, there come structural and cultural shifts that bring about a different type of challenges. In this presentation, we discuss the potential of transformation in banking made possible by AI technologies but also the cultural and structural challenges that this transformation represents. We will discuss these issues around a few exemples of applications of machine learning in banking.

Speakers

Manuel Morales

Chief AI Scientist

National Bank of Canada

Dominique Payette

Lawyer, Legal Affairs/ Collaborator

National Bank of Canada/ FinML

Back

Deep Dive Sessions

DEEP DIVE: Deep Learning in Transport & Cities: The Human Urban Mobility - RYERSON UNIVERSITY

11:00 AM 11:45 AM Vivaldi Room

As autonomous vehicles are finding their way on the streets of the future, they have the potential to fundamentally alter the dynamics of urban areas. Interaction between pedestrians and autonomous vehicles is one of the lesser discussed topics in the field. In this context, a virtual reality-based framework is presented to collect and analyze a large naturalistic dataset of pedestrians’ road crossing behaviour. To analyze the data, an interpretable deep survival model, along with a deep LSTM with auxiliary information is used to capture the inherently complex behavioural characteristics of pedestrians, including their crossing intention, waiting time and trajectory. Key Takeaways: - How to utilize deep learning to manage interactions of pedestrians and autonomous vehicles. - Interpretability of Deep learning Models: a case study on pedestrian behaviour. - What cities can do to prepare a pedestrian-friendly urban area in the age of autonomous vehicles.

Speakers

Arash Kalatian

PhD Candidate

Ryerson University

Back

Deep Learning Summit

Understanding the Behavior of Time Series Data Using the Matrix Profile and Deep Learning - TARGET

11:25 AM 11:45 AM

Target is a large retail company with over 1,800 stores in the U.S. Because of this scale, it can be difficult to find anomalous behavior in data or pinpoint what metrics could potentially be correlated. In order to understand the behavior of this data at scale, Target open-sourced the Python library matrixprofile-ts. Using this library, we can layer models on top of the Matrix Profile to find when anomalous behavior occurs or when different metrics in different areas of the company affect each other. This talk will briefly go over the matrixprofile-ts library and examples of where deep learning models can be applied to complement it. Key Takeaways: Target implements research at scale, the difficulty of having hundreds of thousands of data streams, how deep learning can be applied at a large company.

Speakers

Frankie Cancino

Senior Engineer & Data Scientist

Target

Back

Responsible AI Summit

Informing, Delighting and Entertaining ALL Canadians through Machine Learning - CBC

11:25 AM 11:45 AM

The CBC's mandate is to Inform, Delight and Entertain all Canadians, which the CBC does not take lightly. As part of the fulfilment of this mandate, the CBC has embarked on the creation of a personalization team to ensure we're reaching all Canadians. At the core of this objective we're ensuring that we uphold the privacy of our audience, that we focus on Fairness, Accountability and Trust in machine learning, and that we ensure that our personalization approaches are accessible to everyone. Key Takeaways: An understanding Privacy by design, FATRec and key approaches to make it all happen.

Speakers

Jason Cornell

Manager, Machine Learning, Recommendation Systems

CBC

Back

Deep Learning Summit

Recent Advances in Real-Time Road Traffic Analytics - UNIVERSITE DE SHERBROOKE

11:45 AM 12:10 PM

In this presentation, I will present our latest advances on traffic surveillance. In the wake of the CVPR 2017 MIO-TCD challenge, we developed various models to analyze traffic based on ultra-low frame rate videos. This includes applications such as vehicle recognition, orientation estimation, and motion detection. We also explored solutions to compress deep conv nets on cameras whose hardware does not accommodate with large state-of-the-art convolutional neural network. The proposed compression algorithm is a budgeted dropout sparsity learning approach. We also studied ways of measuring the complexity of certain datasets following a novel "complexity measure" which allows to assess the fundamental complexity of a give image classification problem. Key Takeaways: Real-life concerns in video traffic analysis, network compression, vehicle localization and recognition.

Speakers

Pierre-Marc Jodoin

Full Professor

Université de Sherbrooke

Back

Responsible AI Summit

Collaboratively Building Trust in AI - AI GLOBAL

11:45 AM 12:10 PM

There are several good reasons why there's increasing wariness and a limited degree of trust in the use and adoption of AI, especially in the public sector. We've all read the news stories about predictive policing, the use of affective computing in the education systems, and inaccurate health diagnosis made by automated systems. While these technologies have yet to be perfected, they are not going away. This is why it's imperative to work together not to protest these technologies, but influence how they are being designed, developed, and deployed. This presentation will highlight work underway to build the practical tools needed to support the responsible implementation of AI in the public sector and beyond.

Speakers

Ashley Casovan

Executive Director

AI Global

Back

Deep Dive Sessions

DEEP DIVE: Deep Learning in the 3D World - CONCORDIA UNIVERSITY

11:45 AM 12:30 PM Vivaldi Room

Applying deep learning to 3D geometry is a challenging area of research because we typically represent 3D shapes as non-Euclidean data. This makes it difficult to apply convolutional neural networks (CNNs), because CNNs are designed to operate on Euclidean domains. Nevertheless, the deep learning community has devised many ingenious techniques to use deep learning for processing and generating 3D shapes. This deep dive session will cover the most exciting developments in the field of deep learning for 3D geometry. The following broad topics will be covered:1) Using grid structures to create Euclidean domains for convolution. 2) Applying deep learning directly to unstructured 3D point clouds. 3) Generating 3D shapes using deep neural networks. Along the way, we will encounter popular 3D deep learning architectures such as PointNet and AtlasNet. Key Takeaways: 1) 3D geometry can be made compatible with CNNs using grid structures. 2) Point clouds can be used directly with deep neural networks. 3) Deep neural networks can be used to generate 3D geometry.

Speakers

Prashant Raina

Ph.D Candidate

Concordia University

Back

Deep Learning Summit

Deep Learning for Program Repair - GOOGLE

12:10 PM 12:30 PM

The software development process can be frustrating, painful and costly; rife with bugs, project delays and unexpected outages. If machine learning were to help with software engineering it would make for the stuff of dreams. ML4SE (Machine Learning for Software Engineering) is an active research area in this space. In this talk we describe progress we've made at Google on training deep learning models to fixing build errors encountered by software developers.

Speakers

Subhodeep Moitra

Research Software Engineer

Google

Back

Responsible AI Summit

Trustworthy Decision Making: The Case of Autonomous Driving - HUAWEI

12:10 PM 12:30 PM

How to make AI-based decision making trustworthy is a core research question at Noah’s Ark Lab at Huawei. We take the position that trustworthy decision making must be responsive, reliable, reasonable, and resilient. We review some of the specific challenges from and efforts in areas such as communication network, data centre management, supply chain optimization, and autonomous driving.

Speakers

Jun Luo

Director of Decision Making and Reasoning

Huawei, Noah's Ark Lab

Back

Exhibition Area & Refreshment Breaks

LUNCH

12:30 PM 01:30 PM

A hot, 3-course, lunch buffet will be served in the foyer area. A great time for networking and to get to know your fellow participants or you can join the Lunch & Learn Session taking place in the Deep Dive Track and take a seat at one of the tables to hear more from the speakers and exhibitors.

Back

Deep Dive Sessions

DEEP DIVE: Rising Stars

12:35 PM 01:25 PM Tchaikovsky Room

Grab lunch and hear quick-fire presentations from the next generation of AI pioneers.

12:40 - 12:55: Shalev Lifshitz - "Humanity & AI: Preparing for an Intelligent Future"
12:55 - 13:10: Hannah Le - "Evolving a Soft Robot to Explore Space"
13:10 - 13:25: Seyone Chithrananda - "How Deep Learning Will Help Us Eradicate Disease and Disrupt Personalized Medicine"

Speakers

Seyone Chithrananda

The Knowledge Society

Innovator

Hannah Le

Genetics Student Researcher/Innovator

TKS

Shalev Lifshitz

AI Researcher

University of Waterloo & SickKids Hospital

Lucie Luneau

AI Project Manager

Kids Code Jeunesse

Back

Deep Learning Summit

Can You Trust Your Model’s Uncertainty? Evaluating Predictive Uncertainty Under Dataset Shift - GOOGLE BRAIN

01:30 PM 01:50 PM

Modern machine learning methods including deep learning have achieved great success in predictive accuracy for supervised learning tasks, but may still fall short in giving useful estimates of their predictive uncertainty. Quantifying uncertainty is especially critical in real-world settings, which often involve input distributions that are shifted from the training distribution due to a variety of factors including sample bias and non-stationarity. In such settings, well calibrated uncertainty estimates convey information about when a model's output should (or should not) be trusted. Many probabilistic deep learning methods, including Bayesian-and non-Bayesian methods, have been proposed in the literature for quantifying predictive uncertainty, but to our knowledge there has not previously been a rigorous large-scale empirical comparison of these methods under dataset shift. We present a large-scale benchmark of existing state-of-the-art methods on classification problems and investigate the effect of dataset shift on accuracy and calibration. We find that traditional post-hoc calibration does indeed fall short, as do several other previous methods. However, some methods that marginalize over models give surprisingly strong results across a broad spectrum of tasks.

Speakers

Jasper Snoek

Research Scientist

Google Brain

Back

Responsible AI Summit

Why Public Competence in AI Ethics is Essential to the Future of AI? - MONTREAL AI ETHICS INSTITUTE

01:30 PM 01:50 PM

Given all the work that has been coming out of the field of responsible and ethical AI, there has been a push for finding universal solutions, often mediated and prepared by a group of experts. But, relying on a small group of experts and hunting for universal solutions is an exercise in futility. What we really need is cultural and contextual sensitivity. This can only be achieved by engaging at the grassroots with the public and tapping into their implicit knowledge of the local culture and context. But this needs to be nurtured and built over time through a public engagement process which is what this session will dive into. Key Takeaways: 1) How to do effective public consultations in AI ethics? 2) Learn from the case study in Montreal 3) Key pitfalls that any public effort such as this can suffer from.

Speakers

Mirka Snyder Caron

Associate

Montreal AI Ethics Institute

Back

Deep Learning Summit

How Do You Know What A Deep Network Has Learned? - CARNEGIE MELLON UNIVERSITY/ARGO AI

01:50 PM 02:10 PM

Modern deep learning algorithms are able to learn on training sets such that they achieve almost zero train error. What is all the more amazing, is that this performance tends to generalize well to unseen data - especially for visual detection and classification tasks. Increasingly, deep methods are being utilized in vision tasks such as object tracking and visual SLAM. These tasks differ fundamentally to traditional vision tasks where deep learning has been effective (e.g. object detection and classification) tasks as they are attempting to model the relative relationship between image frames. Although receiving state of the art performance on many benchmarks, it is easy to demonstrate empirically that deep methods are not always learning what we want them to learn for a given visual task - limiting their practical usage in real-world applications. In this talk we shall discuss recent advances my group has made towards making better guarantees over the generalization of deep learning methods for visual tasks where the relative relationship between images is important - most notably object tracking and VSLAM. In particular we shall discuss a new paradigm for efficient and generalizable object tracking which we refer to as Deep-LK and its extension to 3D PointNet-LK. We shall also, discuss how these insights can be utilized in recent applications of deep learning to VSLAM. Finally, we will show some initial results on how geometric constraints can be elegantly combined with deep learning to further improve generalization performance. Key Takeaways: (i) Integrating classical methods with deep learning is a good way to ensure generalization. (ii) The application of deep learning to problems in 3D geometry is still in its infancy (but moving fast). (iii) Unsupervised 3D learning has a lot of untapped potential.

Speakers

Simon Lucey

Associate Research Professor/Principal Research Scientist

Carnegie Mellon University/Argo AI

Back

Responsible AI Summit

PANEL: Discussing Ethical Approaches to AI: Reviewing Regulation, Risk Management & Policy- SIDEWALK LABS, FUTURE OF LIFE INSTITUTE, MORGAN STANLEY, QUEEN'S UNIVERSITY

01:50 PM 02:30 PM

The focus of this panel discussion is assessing organization wide ethical approaches to AI and what must be considered in this process with regards to regulation, policy as well as managing and mitigating risk. Questions include: How can organizations/enterprises best harness the capabilities of AI whilst mitigating the risks from the unethical use of data? Are there differences in approaches across different industries and/or enterprises of varying sizes? What are the key aspects of managing data and digital rights in an ethical manner for businesses? How do you expect this to change in the future? What process/people/teams need to be involved in these processes internally and externally to ensure compliance & proper governance of data?

Speakers

Chelsey Colbert

Legal Associate, Privacy and Data Governance

Sidewalk Labs/Fasken

Tina Dacin

Full Professor

Queen's University

Oana Magdalena Frunza

NLP & Machine Learning Scientist

Morgan Stanley

Richard Mallah

Director of AI Projects

Future of Life Institure

Back

Deep Learning Summit

The Difference is Night and Day: Appearance Modelling for Robust Robot Visual Navigation - UNIVERSITY OF TORONTO

02:10 PM 02:30 PM

Long-term localization and mapping are essential capabilities for autonomous mobile robots that must operate independently in dynamic environments. In this talk, I will describe two deep neural network systems that have been developed in my laboratory group to improve vision-based navigation. The first system, Sun-BCNN, uses the sun as an absolute orientation reference to reduce drift in visual motion estimates. Crucially, Sun-BCNN learns to recognize how solar illumination affects images, and does not require the sun to be visible in the image stream to operate. The second system relies on image-to-image translation and learned colour constancy models to transform incoming images in appearance space. This remapping enables a robot to localize against a visual map created under substantially different lighting conditions (for example, between dawn and dusk).

Speakers

Jonathan Kelly

Assistant Professor

University of Toronto

Back

Exhibition Area & Refreshment Breaks

NETWORKING MIXER

02:30 PM 03:30 PM

Join us in the foyer area and grab a drink to celebrate the end of the summit and continue to network with attendees.

Back

Exhibition Area & Refreshment Breaks

END OF SUMMIT

03:30 PM 03:35 PM

Thank you for attending! Feel free to continue discussions via the app with your new connections. We look forward to seeing you again soon!

Schedule

REGISTRATION & LIGHT BREAKFAST

WELCOME

Towards Combining Statistical Relational Learning and Graph Neural Networks - HEC Montreal

AI Ethics: A Deflationary Yet Cautionary Tale - UNIVERSITE LAVAL

Few-Shot Learning: Thoughts On Where We Should Be Going - GOOGLE BRAIN

Social Inclusion in the AI Pipeline - POLYTECHNIQUE MONTREAL

Building Knowledge For AI Agents With Reinforcement Learning - DEEPMIND

AI for Humanity: How AI Could Benefit Us All- MILA

COFFEE

DEEP DIVE: Introduction to Deep Reinforcement Learning - MILA

DEEP DIVE: Neural Networks for Lawyers and Other Non-Technical Professionals - MIT

The Lottery Ticket Hypothesis: Finding Sparse, Trainable Neural Networks - MIT

Ethical Issues in NLP - UNIVERSITY OF TORONTO

DEEP DIVE: AI, Deep Learning, Optimization: How to Frame and Solve Real World Problems Effectively to Get the Most Business Benefits - ALPSANALYTICS.COM

Generalizing The Lottery Ticket Hypothesis Across Datasets and Optimizers and Beyond Supervised Image Classification - FACEBOOK AI RESEARCH

Finding Bias In Your Systems - SHOPIFY

Outlines, Explanation, and Deflecting Adversarial Examples - GOOGLE BRAIN

DEEP DIVE: How The Pharmaceutical Industry Is Leveraging Deep Learning - NOVARTIS

Agnostic Data Debiasing Through a Local Sanitizer Learnt from an Adversarial Network Approach - UNIVERSITE DU QUEBEC A MONTREAL

Learning to Act More Like Humans, and Learning with Less Data - POLYTECHNIQUE MONTREAL

LUNCH

DEEP DIVE: Lunch and Learn

PM Host Introduction

Productionizing Deep Learning Models - EXPEDIA

Fairwashing: The Risk of Rationalization - UNIVERSITÉ DE MONTRÉAL

DEEP DIVE: Algorithmic Profiling & Illegal Discrimination: A Cross-Industry Analysis - MONTREAL AI ETHICS INSTITUTE

AI for Self Driving - From Research to Production - UBER ATG

Learning Fair Rule Lists - UNIVERSITE DU QUEBEC A MONTREAL

New Optimization Perspectives on Generative Adversarial Networks - SAIT AI LAB

The Evolving Relationship Between Ethics and Safety in AI - FUTURE OF LIFE INSTITUTE

DEEP DIVE: Consumer Airfare Prediction and other Big Data AI Challenges at Hopper- HOPPER

PANEL: How Can We Overcome Challenges To Fully Leverage the Oportunities of GANs?- MILA, UNIVERSITY OF MASSACHUSETTS, SAIT AI LAB/UNIVERSITE DE MONTREAL, CONCORDIA UNIVERSITY

Adversarial Machine Learning: Ensuring Security of ML Models and Sensitive Data - VECTOR INSTITUTE

COFFEE

DEEP DIVE: Talent and Talk

Deep Learning and Cognition - UNIVERSITÉ DE MONTRÉAL

CONVERSATION & DRINKS

END OF SUMMIT

REGISTRATION & LIGHT BREAKFAST

WELCOME

Visual Perception for Better Understanding of Sports Games - SPORTLOGIQ

AI and The Evolution of Work - KINDRED.AI

Version Control for Jupyter Notebooks - VALOHAI

Artificial Intelligence & Ethical Deployment - DELOITTE

Pushing the Energy Efficiency Frontier with Deep Learning - BRAINBOX AI

Fireside Chat: Responsible AI at Target - TARGET & AI IMPACT ALLIANCE

COFFEE

DEEP DIVE: Investing in AI Startups: Challenges & Opportunities

Deep Learning for Conversational AI - BOREALIS.AI

AI Governance in Banking: The Role and Challenges of Deploying AI in Banking - NATIONAL BANK OF CANADA

DEEP DIVE: Deep Learning in Transport & Cities: The Human Urban Mobility - RYERSON UNIVERSITY

Understanding the Behavior of Time Series Data Using the Matrix Profile and Deep Learning - TARGET

Informing, Delighting and Entertaining ALL Canadians through Machine Learning - CBC

Recent Advances in Real-Time Road Traffic Analytics - UNIVERSITE DE SHERBROOKE

Collaboratively Building Trust in AI - AI GLOBAL

DEEP DIVE: Deep Learning in the 3D World - CONCORDIA UNIVERSITY

Deep Learning for Program Repair - GOOGLE

Trustworthy Decision Making: The Case of Autonomous Driving - HUAWEI

LUNCH

DEEP DIVE: Rising Stars

Can You Trust Your Model’s Uncertainty? Evaluating Predictive Uncertainty Under Dataset Shift - GOOGLE BRAIN

Why Public Competence in AI Ethics is Essential to the Future of AI? - MONTREAL AI ETHICS INSTITUTE

How Do You Know What A Deep Network Has Learned? - CARNEGIE MELLON UNIVERSITY/ARGO AI

PANEL: Discussing Ethical Approaches to AI: Reviewing Regulation, Risk Management & Policy- SIDEWALK LABS, FUTURE OF LIFE INSTITUTE, MORGAN STANLEY, QUEEN'S UNIVERSITY

The Difference is Night and Day: Appearance Modelling for Robust Robot Visual Navigation - UNIVERSITY OF TORONTO

NETWORKING MIXER

END OF SUMMIT

Speakers

Sponsors

Job Board

Exhibitors

Yoshua Bengio

24 October

Deep Learning and Cognition - UNIVERSITÉ DE MONTRÉAL

Hugo Larochelle

24 October

Few-Shot Learning: Thoughts On Where We Should Be Going - GOOGLE BRAIN

Simon Lucey

25 October