Conceptual Understanding of Deep Learning Workshop
May 17, 2021
0:00 Welcome by Rina Panigrahy 11:40 Workshop's goal 16:23 talk on "How to Augment Supervised Learning with Reasoning" by Leslie Valiant 39:40:00 talk on "Language in the brain and word representations" by Christos Papadimitriou 1:02:15 talk on "What Do Our Models Really Learn?" by Aleksander Madry 1:27:08 talk on "Implicit Symbolic Representation and Reasoning in Deep Networks" by Jacob Andreas 1:48:16 Panel Discussion on "Is there a Mathematical model for the Mind?" 2:47:09 talk on "Deep Reinforcement Learning and Distributional Shift" by Sergey Levine 3:10:53 talk on "Towards a Representation Learning framework for Reinforcement Learning" by Alekh Agarwal 3:32:37 talk on "Principles for Tackling Distribution Shift: Pessimism, Adaptation, and Anticipation" by Chelsea Finn 4:20:27 talk on "Can human brain recordings help us design better AI models?" by Leila Wehbe 4:38:52 talk on "The benefits of unified frameworks for language understanding" by Colin Raffel 4:54:26 talk on "Are Transformers Universal Approximators of sequence-to-sequence Functions?" by Srinadh Bhojanapalli 5:07:05 talk on "Function space view of Multi-Channel Linear Convolutional Networks with Bounded Weight Norm" by Suriya Gunasekhar 5:32:48 talk on "Theoretical Analysis of Contrastive Learning and Self-training with Neural Networks" by Tengyu Ma 5:55:17 talk on "Escaping Global Minima Using Stochastic Gradients" by Jason Lee 6:09:56 talk on "Guarantees for Tuning the Step Size using a Learning-to-Learn Approach" by Rong Ge Goal: How does the Brain/Mind (perhaps even an artificial one) work at an algorithmic level? While deep learning has produced tremendous technological strides in recent decades, there is an unsettling feeling of a lack of “conceptual” understanding of why it works and to what extent it will work in the current form. The goal of the workshop is to bring together theorists and practitioners to develop an understanding of the right algorithmic view of deep learning, characterizing the class of functions that can be learned, coming up with the right learning architecture that may (provably) learn multiple functions, concepts and remember them over time as humans do, theoretical understanding of language, logic, RL, meta learning and lifelong learning. Panel Discussion: There will also be a panel discussion on the fundamental question of “Is there a mathematical model for the Mind?”. We will explore basic questions such as “Is there a provable algorithm that captures the essential capabilities of the mind?”, “How do we remember complex phenomena?”, “How is a knowledge graph created automatically?”, “How do we learn new concepts, function and action hierarchies over time?” and “Why do human decisions seem so interpretable?” Website can be found here: https://sites.google.com/view/conceptualdlworkshop Twitter: #ConceptualDLWorkshop, @rinapy Speakers include: Alekh Agarwal - Microsoft Research Aleksander Madry - Massachusetts Institute of Technology Chelsea Finn - Stanford University, Google Christos Papadimitriou - Columbia University Colin Raffel - University of North Carolina and Google Jacob Andreas - Massachusetts Institute of Technology Jason Lee - Princeton University Leila Wehbe - Carnegie Mellon University Leslie Valiant - School of Engineering and Applied Sciences, Harvard University Rong Ge - Duke University Sergey Levine - UC Berkeley and Google Srinadh Bhojanapalli - Google Suriya Gunasekhar - Microsoft Research Tengyu Ma - Stanford University Panelists: Bin Yu - UC Berkeley Geoffrey Hinton - University of Toronto and Google Jack Gallant - UC Berkeley Lenore Blum -CMU/UC Berkeley Percy Liang - Stanford University Workshop Chair: Rina Panigrahy, Google Thanks to: Pranjal Awasthi, Manzil Zaheer, Kristen Konrad, Hanieh Haddadian
Reactions (0) | Note
📝 No reactions yet
Be the first one to share your thoughts!