• 08:00
    Gayathri Radhakrishnan

    WELCOME & OPENING REMARKS - 8am PST | 11am EST | 4pm GMT

    Gayathri Radhakrishnan - Director Venture Capital - AI Fund - Micron Technology

    Down arrow blue

    Gayathri Radhakrishnan is a Director at the investment team at Micron Ventures, investing from the $100M AI fund. She invests in startups leveraging AI/ML to solve critical problems in the areas of Manufacturing, Healthcare, Automotive and AgTech. She brings 20+ years of multi-disciplinary experience across product management, product marketing, corporate strategy, M&A across Fortune 500 companies such as Dell and Corning and in startups. She was also an early stage investor at Earlybird Venture Capital, a premier European venture fund based in Germany. She has a Masters in EE from The Ohio State University and MBA from INSEAD. She is also a Kauffman Fellow.


  • 08:10
    Kenneth Stanley

    From Open-Endedness to AI

    Kenneth Stanley - Research Manager - OpenAI

    Down arrow blue

    From Open-Endedness to AI

    While much of AI is focused on the ability to solve problems, open-ended processes are arguably far more powerful even though they do not solve any specific problem. Instead, an open-ended process continues to produce increasingly complex yet unpredictable new inventions and innovations forever. We know that open-endedness is possible because humans ourselves are the product of such a process –- evolution –- but also because we have exhibited open-endedness ourselves over the creative explosion of millennia of civilization. Thus to truly achieve intelligence at the human level we must ultimately account for open-endedness.

    Takeaways: 1) Open-endedness is a deep and fundamental facet of our intelligence, yet often neglected in the pursuit of AI. 2) Open-endedness is a growing field within AI and machine learning with great potential to leverage the power of modern approaches such as deep learning. 3) Open-endedness reveals the importance of a number of unconventional topics within ML, such as divergence, populations, diversity preservation, stepping stone collection, generating problems and solutions at the same time, and environment design.

    Kenneth O. Stanley is Charles Millican Professor of Computer Science at the University of Central Florida and director there of the Evolutionary Complexity Research Group. He was also a co-founder of Geometric Intelligence Inc., which was acquired by Uber to create Uber AI Labs, where he is now also a senior research science manager and head of Core AI research. He received a B.S.E. from the University of Pennsylvania in 1997 and received a Ph.D. in 2004 from the University of Texas at Austin. He is an inventor of the Neuroevolution of Augmenting Topologies (NEAT), HyperNEAT, and novelty search neuroevolution algorithms for evolving complex artificial neural networks. His main research contributions are in neuroevolution (i.e. evolving neural networks), generative and developmental systems, coevolution, machine learning for video games, interactive evolution, and open-ended evolution. He has won best paper awards for his work on NEAT, NERO, NEAT Drummer, FSMC, HyperNEAT, novelty search, Galactic Arms Race, and POET. His original 2002 paper on NEAT also received the 2017 ISAL Award for Outstanding Paper of the Decade 2002 - 2012 from the International Society for Artificial Life. He is a coauthor of the popular science book, "Why Greatness Cannot Be Planned: The Myth of the Objective" (published by Springer), and has spoken widely on its subject.

    Twitter Linkedin
  • 08:35
    Puneet Dokania

    Continual Learning: Quirks and Assumptions

    Puneet Dokania - Senior Researcher - University of Oxford

    Down arrow blue

    Continual Learning: Quirks and Assumptions

    In this talk, I'll present our very recent work called GDumb (ECCV2020, Oral) and discuss various quirks and assumptions encoded in recently proposed approaches for continual learning (CL). We argue that some oversimplify the problem to an extent that leaves it with very little practical importance, and makes it extremely easy to perform well on. To validate this, we propose GDumb that (1) greedily stores samples in memory as they come and; (2) at test time, trains a model from scratch using samples only in the memory. We show that even though GDumb is not specifically designed for CL problems, it obtains state-of-the-art accuracies (often with large margins) in almost all the experiments when compared to a multitude of recently proposed algorithms. Surprisingly, it outperforms approaches in CL formulations for which they were specifically designed. This, we believe, raises concerns regarding our progress in CL for classification. Overall, we hope our formulation, characterizations and discussions will help in designing realistically useful CL algorithms, and GDumb will serve as a strong contender for the same.

    3 Key Takeaways:

    1. Continual learning (CL) is important and will become extremely crucial for industries such as Google/Facebook training their models on billions of samples using hundreds of GPUs and weeks of training.

    2. Even though CL is important, there does not exist a very practical use case of CL.

    3. Time and space constraints based experiment design and evaluation for CL needs rethinking and better formulations.

    Puneet holds two research positions, one as a senior researcher in machine learning and computer vision at the Torr Vision Group (University of Oxford) and another as a principal researcher at an amazing startup based in Cambridge (U.K.) called Five AI. He obtained his PhD from INRIA and Ecole Centrale Paris in France in 2016 after which he moved to Oxford as a postdoctoral researcher. Puneet's research theme revolves around developing "reliable and efficient algorithms with natural intelligence using deep learning". Primarily, his current focus is on topics like continual learning, robustness, calibration, and parameter quantization.

    Twitter Linkedin
  • 09:00
    Abha Laddha

    Supercharge Your Data Quality: Automated QA

    Abha Laddha - Senior Customer Success Engineer - Sama

    Down arrow blue

    Supercharge Your Data Quality: Automated QA

    When it comes to AI, your model is only as good as the data it’s trained on. While it’s key to have a human in the loop when creating and verifying training data, automating processes within the workflow improves efficiency while guaranteeing high quality. Sama’s Auto QA functionality assesses tasks using a fixed set of rules configured to check for invalid combinations of labels and metadata. These checks are triggered before a task is submitted on our training data platform, allowing for the detection of errors early in the pipeline to save time and increase quality.

    We discuss:

    1. Introduction to Sama Automated QA
    2. Eliminating 100% of logical fallacies with auto QA
    3. Increase data trainer efficiency by shortening the feedback loop
    4. Applications in automotive and hi-tech use cases

    With a background in Computer Science, Abha leads the Customer Success Engineering team at Sama. The team is responsible for managing technical relationships with customers and prospects to understand their business needs, ideate upon them, and manage the implementation and communication of the solutions developed. Within the company, she wears many hats: from building complex solutions to deliver high quality data for complex workflows, to improving internal processes, and troubleshooting the in-house SamaHub platform. Her mission is to design and architect solutions that help clients achieve their larger ML vision while creating a positive social impact.

    Twitter Linkedin
  • 09:25



  • 09:35
    Stephanie Kirmer

    Scaling Up Pytorch with GPU Cluster Computing

    Stephanie Kirmer - Senior Data Scientist - Saturn Cloud

    Down arrow blue

    Scaling Up Pytorch with GPU Cluster Computing

    In this session, attendees will get a short overview of GPU computing and why it is valuable for deep learning tasks, and then will be walked through an example of using GPU cluster computing with parallelization to conduct a high volume image classification task with Resnet50 in Pytorch. We'll discuss the advantages and drawbacks to using GPU clusters, including speed, cost, and ease of use.

    Stephanie Kirmer is a Senior Data Scientist at Saturn Cloud, a platform enabling easy to use parallelization and scaling for Python with Dask. Previously she worked as a DS Tech Lead at Journera, a travel data startup, and Senior Data Scientist at Uptake, where she developed predictive models for diagnosing and preventing mechanical failure. Before joining Uptake, she worked on data science for social policy research at the University of Chicago and taught sociology and health policy at DePaul University.

    Twitter Linkedin

  • 10:00
    Alireza Fathi

    Object Detection and Segmentation in 3D

    Alireza Fathi - Senior Research Scientist - Google

    Down arrow blue

    Tensorflow 3D

    At Google, we develop flexible state of the art machine learning systems for computer vision that not only can be used to improve our products and services, but also spur progress in the research community. Creating accurate machine learning models capable of localizing and identifying multiple objects in a 3D scene, predicting object shapes, and assigning semantic labels to different components of the scene is a core challenge in computer vision, with applications in robotics and autonomous driving. We invest a significant amount of time training and experimenting with these systems. Today we are happy to make some parts of this system available to the broader research community via the TensorFlow 3D codebase. This codebase is an open-source framework built on top of TensorFlow 2 and Keras that makes it easy to construct, train and deploy 3D semantic segmentation, 3D object detection and 3D instance segmentation models as well as other potential applications like 3d shape prediction and point cloud registration and completion.

    3 Key Takeaways:

    1) GPU and CPU ops for 3D submanifold sparse convolution in Tensorflow.

    2) A configurable 3D sparse voxel U-Net network that is used as the feature extractor in our models.

    3) Training and evaluation code for 3D Semantic Segmentation, 3D Object Detection and 3D Instance Segmentation, with support for distributed training.

    Alireza Fathi is currently a senior research scientist at Google Research Machine Perception team. His main area of focus has been on 3d scene understanding for the last two to three years. Before joining Google, he spent a couple of great years at Apple working on 3d computer vision research. Before that he was a Postdoctoral Fellow in FeiFei Li's lab at the CS Department at Stanford University. He received his Ph.D. degree from Georgia Institute of Technology, and his B.Sc. degree from Sharif University of Technology.

    Twitter Linkedin
  • 10:25
    Roundtable Discussions & Demos with Speakers


    Roundtable Discussions & Demos with Speakers - - AI EXPERTS

    Down arrow blue

    Join a roundtable discussion hosted by AI experts to get your questions answered on a variety of topics.

    You are free to come in and out of all sessions to ask your questions, share your thoughts, and learn more from the speakers and other attendees.

    Roundtable Discussions 28th January: • 'From Open-Endedness to AI' hosted by Kenneth Stanley, Research Manager, OpenAI • ‘Cost Optimize Your Machine Learning with Multi-Cloud’ hosted by Leon Kuperman, Co-Founder & CTO, CAST AI

    • ‘Creating Data Products with Visual Intelligence’ hosted by Daniel Gifford, Senior Data Scientist, Getty Images

    • ‘Swapping Autoencoder for Deep Image Manipulation’ hosted by Richard Zhang, Research Scientist, Adobe

    • 'Conversational AI: Human-Like or All Too Human’ hosted by Mark Jancola, CTO & VP of Engineering, Conversica

    • 'Supercharge Your Data Quality: Automated QA' hosted by Aurelie Drouet, Product Marketing Manager & Shaashwat Saraf, Customer Success Engineer, Sama

    Roundtable Discussions 29th January: • ‘Curriculum Generation for Reinforcement Learning’ hosted by Natasha Jaques, Research Scientist, Google Brain

    • ‘The AI Economist’ hosted by Stephan Zhang, Lead Research Scientist, Salesforce Research

    • ‘A Win-Win in Precision Ag’ hosted by Jennifer Hobbs, Director of Machine Learning, Intelin Air

    • ‘Delivering responsible AI: via the carrot or the stick?’ hosted by Myrna MacGregor, BBC Lead, Responsible AI+ML, BBC

  • 10:45



  • 10:55
    Danqi Chen

    Open-Domain Question Answering: State of the Art and Future Perspectives

    Danqi Chen - Assistant Professor - Princeton University

    Down arrow blue

    Open-domain Question Answering: State of the Art and Future Perspectives

    Question answering (QA) is one of the earliest and core topics in natural language processing and has played a central role in many real-world applications such as search engines and personal assistants. The problem of open-domain QA, which aims to automatically answer questions posed by humans based on a large collection of unstructured documents, has (re-)gained a lot of popularity in the last couple of years. This talk will review some of the exciting advances in the field, including some of my earlier and recent experiences in building neural QA systems. In particular, I will discuss the role of pre-training in question answering, learning dense representations for retrieval, and the trade-off between accuracy, storage, and runtime efficiency. I will conclude with current limitations and future directions.

    3 Key Takeaways:

    1. Today, we can build a single end-to-end neural open-domain QA system that can answer 50\%-70\% of the questions accurately based on the full English Wikipedia.
    2. The progress is largely driven by the development of pre-trained language representations and effective methods for learning dense retrieval.
    3. Representing text source as a collection of dense vectors opens up a new possibility for building next-generation knowledge bases.

    Danqi Chen is an Assistant Professor of Computer Science at Princeton University and co-leads the Princeton NLP Group. Danqi’s research focuses on deep learning for natural language processing, with an emphasis on the intersection be- tween text understanding and knowledge representation/reasoning and applications such as question answering and infor- mation extraction. Before joining Princeton, Danqi worked as a visiting scientist at Facebook AI Research in Seattle. She received her Ph.D. from Stanford University (2018) and B.E. from Tsinghua University (2012), both in Computer Science.

  • 11:20
    Yonatan Geifman

    What if AI Could Craft the Next Generation of AI?

    Yonatan Geifman - Co-Founder & CEO - Deci AI

    Down arrow blue

    What If AI Could Craft the Next Generation of AI

    Taking an AI model from the lab to production is extremely challenging. In fact, recent reports and surveys estimate that only 20%-30% of the deep neural modelling attempts find their way to productive deployment. One of the major bottlenecks in the path from the lab to production is the poor latency or throughput performance of these neural models, which immediately translates to excessively high cost-to-serve. In this talk, we present an innovative solution to this problem, driven by Deci AI’s deep learning platform.

    3 Key Takeaways:

    1. The challenges, solutions, and opportunities associated with building and maintaining AI models that are ready for  production.
    2. The “Algorithmic Complexity” and how to overcome it for faster deep learning inference at scale.
    3. The fundamental principles for leveraging AI to build better deep learning models - a paradigm shift reinvented by Deci.

    Yonatan Geifman is co-founder and CEO of Deci, the deep learning company dedicated to transforming the AI lifecycle. He co-founded Deci after completing his PhD in computer science at the Technion-Israel Institute of Technology. His research focused on making Deep Neural Networks (DNNs) more applicable for mission-critical tasks.

    During his studies, Yonatan was also a member of Google AI’s MorphNet team. His research has been published and presented at leading conferences across the globe, including the Conference on Neural Information Processing Systems (NeurIPS), the International Conference on Machine Learning (ICML), and the International Conference on Learning Representations (ICLR).

    Twitter Linkedin
  • 11:45

    MAKE CONNECTIONS: Meet with Attendees Virtually for 1:1 Conversations and Group Discussions over Similar Topics and Interests

  • 12:00



This website uses cookies to ensure you get the best experience. Learn more