For the first time, the Deep Learning Summit will feature 10 separate stages where experts will hone in on specific topics in AI to delve into the nitty-gritty of each area. In between the presentations, workshops, panel discussions, interviews and fireside chats will be extensive opportunities for networking, where attendees will all come together in the exhibition area to share their knowledge and make new connections.
With almost 100 speakers sharing their knowledge, you can be forgiven for not quite knowing where to start when you’re planning your agenda, so we’re taking a look at some of the most popular roles of this year's attendees and highlighting some of the sessions that you might find interesting.
Session: Curiosity Driven Learning, a promising strategy in Deep Reinforcement Learning
Expert: Thomas Simonini, Deep Learning Engineer, Deep Reinforcement Learning Course
Time: Thurs Jan 24th, 09:15 - 10:00
Curiosity Driven Learning is one of the most exciting and promising strategy in deep reinforcement learning: we create agents that are able to produce rewards and learn from them. In this workshop, you’ll learn what is curiosity, how it works, and understand the process of how an agent generates this intrinsic reward using a trained agent in a video game environment. By the end of this workshop, you'll be able to understand how curiosity driven learning agents work and the main elements needed to implement them.
Session: Generative Adversarial Networks
Expert: Ian Goodfellow, Staff Research Scientist, Google Brain
Stage: Deep Learning
Time: Thurs Jan 24th, 10:00 - 10:25
Speaking at RE•WORK’s Deep Learning Summit in San Francisco for the 5th time this year, Ian is the creator of GANs and one of the leading global minds in deep learning. Ian is the lead author of the MIT Press textbook Deep Learning. In addition to generative models, he also studies security and privacy for machine learning. He has contributed to open source libraries including TensorFlow, Theano, and Pylearn2. He obtained a PhD from University of Montreal in Yoshua Bengio's lab, and an MSc from Stanford University, where he studied deep learning and computer vision with Andrew Ng.
Session: Computer Vision and Deep Learning for Coral Ecology
Expert: David Kriegman, Professor of Computer Science & Engineering, UCSD
Stage: Environment & Sustainability
Time: Thurs Jan 24th, 13:30 - 13:50
Research from the UCSD Computer Vision for Coral Ecology Project has lead to new cameras, algorithms, software, and services. UCSD have developed a series of methods for automatic annotation of benthic images and created CoralNet as a public, open source, hosted tool for scientists to upload their photo surveys and perform annotation using deep nets trained on their data. Through a comparative study published on PLOS ONE, CoralNet is as accurate at estimating coral coverage as human experts.
Session: Building Scalable Framework and Environment of Reinforcement Learning
Expert: Yuandong Tian, Research Scientist & Manager, Facebook AI Research (FAIR)
Stage: Technical Labs
Time: Fri Jan 25th, 11:30 - 12:15
FAIR will introduce their recent open-sourced ELF platform: efficient, lightweight and flexible frameworks to facilitate DRL research. Yuandong will show the scalability of FAIR’s platforms by reproducing and open sourcing AlphaGoZero/AlphaZero framework using 2000 GPUs and 1.5 weeks, achieving super-human performance of Go AI that beats 4 top-30 professional players with 20-0. The presentation will also show usability of the platform by training agents in real-time strategy games with only a small amount of resource. The trained agent develops interesting tactics and is able to beat rule-based AIs by a large margin. On the environment side, FAIR propose House3D that makes multi-room navigation easy with fast frame rate. With House3D, FAIR show that model-based agent that plans ahead with uncertain information navigate in unseen environments more successfully.
...we recommend the following sessions:
Session: Breaking Through Challenges in Industry: Closing Panel
Experts: Thomas Berg, ML Engineer, Dropbox; Parkhar Mehrotra, Sr Director of ML, Walmart Labs; Ashish Bansal, Sr Engineering Manager, Twitter
Stage: Industry Applications
Time: Thurs 24th, 16:05 - 17:00
Wrapping up the Break Through Challenges in Industry series, following each individuals’ presentations, these experts will come together in a panel discussion to discuss the obstacles that they have each faced in their applications of AI & DL. The audience will be invited to engage in a Q&A during the panel. What are the top challenges facing the industry at the moment - is it explainability, ethics, security, or something else? How can companies and experts from different areas work together to help each other overcome these pain-points?
Session: Adopting a Machine Learning Mindset: How to Discover, Develop, and Deliver Automation Solutions Company-Wide
Expert: Marsal Gavalda, Head of Machine Learning, Square
Time: Fri Jan 25th, 09:10 - 09:40
As Machine Learning becomes a core component of any forward-looking company, how can we engage the entire workforce to help with ML and automation initiatives? This talk will cover how Square have adopted a "machine learning mindset" by 1. providing training to all employees (both technical and non-technical) on what ML is and how it works, including current applications and ethical considerations, 2. conducting structured brainstorming sessions to elicit automation opportunities, where everyone can contribute what ML could mean for their team or their customers, and 3. implementing a subset of those ideas by partnering with infrastructure, operations, and product teams, resulting in improved risk management, more efficient internal operations, and novel customer-facing product features.
Session: Why Aren't Our AI Assistants Smarter?
Expert: Cathy Pearl, Head of Conversation Design Outreach, Google
Stage: AI Assistant
Time: Thurs Jan 24th, 10:00 - 10:25
Cathy has been designing and creating Voice User Interfaces (VUIs) for nearly 20 years and is passionate about helping everyone make the best conversational experiences possible. Previously, Cathy was VP of User Experience at Sensely, whose virtual nurse avatar helps people stay healthy. She has worked on everything from programming NASA helicopter pilot simulators to designing a conversational iPad app in which Esquire magazine's style columnist tells users what they should wear on a first date. She has an MS in Computer Science from Indiana University and BS in Cognitive Science from UC San Diego.
Session: Rising Star Feature: Hierarchical Reinforcement Learning
Expert: Bonnie Li, Research Intern & ML Developer, The Knowledge Society (TKS)
Stage: Education & AI
Time: Thurs Jan 24th, 09:45 - 10:05
Bonnie Li is a Machine Learning researcher who is passionate about pushing the current boundaries of the field. At just 17 years old, Bonnie is working on fundamental research in Reinforcement Learning at Mila under Yoshua Bengio. Bonnie holds a Deep Reinforcement Learning nanodegree from Udacity and was mentored by Microsoft. She is currently working on meta learning for efficient exploration, through which she hopes will lead us closer to artificial general intelligence.
...we recommend the following sessions:
Session: Panel: Industry or Academia - Helping Students Decide What Next
Experts: Karol Hausamn, Research Scientist, Google Brain; Vivek Thakral Director, GE; Lubomir Bourdev, Co-Founder & CEO, WaveOne
Time: Thurs Jan 24th, 14:00 - 14:30
PhD student? Undergrad? Keen to know more about the next steps in your AI career? Join experts who have a wealth of experience from both sides of the table to find out what’s right for you. Whilst there is often more money, opportunities and experiences in industry, without academia, these companies cannot survive. The panel will explore incentives for experts to join both academia and AI and weigh up the incentives and career options from both sides. The session will draw on challenges that may be found in industry as a whole, and will explore how and academia will work together.
Session: Understanding the limitations of AI: When Algorithms Fail
Expert: Timnit Gebru, Research Scientist, Ethical AI Team at Google
Stage: Ethics & Social Responsibility
Time: Thurs Jan 24th, 13:00 - 13:30
Automated decision making tools are currently used in high stakes scenarios. From NLP tools used to automatically determine one’s suitability for a job, to health diagnostic systems trained to determine a patient’s outcome, machine learning models are used to make decisions that can have serious consequences on people’s lives. In spite of the consequential nature of these use cases, vendors of such models are not required to perform specific tests showing the suitability of their models for a given task. Nor are they required to provide documentation describing the characteristics of their models, or disclose the results of algorithmic audits to ensure that certain groups are not unfairly treated. Timnit will show some examples to examine the dire consequences of basing decisions entirely on ML based systems, and discuss recent work on auditing and exposing the gender and skin tone bias found in commercial gender classification systems.
Session: Investing in Startups Panel – Part 1 : Current landscape and the ‘Do’s & Don’ts’
Experts: Ashu Garg, General Partner, Foundation Capital; Robert Neivert, Venture Partner, 500 Startups; Albert Wang, Director, Qualcomm Ventures
Stage: Investors & Startups
Time: Fri Jan 25th, 13:20 - 13:50
As a startup looking for your first round of funding it can be overwhelming. Standing our from the crowd is of course key, but it’s equally important to ensure you don’t make any mistakes. These panelists have seen hundreds of pitches and worked with countless startups, so know exactly what makes a good company to invest in. Hear inside information on the current companies thriving, ask any burning questions you have about raising money, and find out what makes a good startup from the investor perspective.
Session: Applying ML & NLP in Google Ads
Experts: Sugato Basu, Sr Staff Research Scientist/Tech Lead of AdsAI, Google
Stage: Deep Learning
Time: Thurs Jan 24th, 13:20 - 13:40
Building and deploying machine-learning (ML) models at Google comes with interesting challenges. For example, some models have to handle massive amounts of training data, while some supervised tasks have insufficient amount of training labels. Or, even when the model quality is good enough for a product requirement, it may not meet other requirements (e.g., serving latency, memory footprint). In this talk we will discuss some of these challenges and share our experiences from deploying ML models for quality improvements in Search Ads products via some case studies. One particular case study I will discuss in detail is a recent paper where we use deep neural networks to understand ad performance and attribute it to particular parts of ad text. This is an interesting research problem in Natural Language Processing (NLP) -- we will outline our key results related to this problem, and discuss interesting areas of future research.