Workshop starts in

The workshop will be hosted in TTIC, Chicago, US (6045 S Kenwood Ave, Chicago, IL 60637). Remote attendance is available via Zoom (Link).

Hotel Information

August 4th

  • Opening and Registration
    8:30 am - 9:00 am
  • Unveiling Transformers with LEGO
    by Sebastien Bubeck
    9:00 am - 09:45 am
  • The unreasonable effectiveness of mathematics in large scale deep learning
    by Greg Yang
    09:45 am - 10:30 am
  • Break
    10:30 am - 10:45 am
  • Towards Better Understanding of Contrastive Learning
    by Yuandong Tian
    10:45 am - 11:30 pm
  • Lunch Break
    11:30pm - 1:00 pm
  • What constitutes a good representation for offline and online RL?
    by Sham Kakade
    1:00 pm - 1:45 pm
  • Aligning representations with humans
    by Anca Dragan
    1:45 pm - 2:30 pm
  • Break
    2:30 pm - 3:00 pm
  • Active Multi-Task Representation Learning
    by Simon S. Du
    3:00 pm - 3:45 pm
  • Invited Talk by Sanjeev Arora
    3:45 pm - 4:30 pm
  • Panel Discussion 1
    4:30pm: What are the major theoretical challenges in Understanding Modern Representation Learning?

August 5th

  • Towards Understanding Training Dynamics for Mildly Overparametrized Models
    by Rong Ge
    9:00 am - 09:45 am
  • Invited Talk by Aleksander Madry
    09:45 am - 10:30 am
  • Break
    10:30 am - 10:45 am
  • Training dynamics, inductive bias and generalization
    by Stefanie Jegelka
    10:45 am - 11:30 pm
  • Lunch Break
    11:30pm - 1:00 pm
  • On the mechanism of prediction head in non-contrastive learning
    by Zixin Wen
    1:00 pm - 1:45 pm
  • Invited Talk by Andrej Risteski
    1:45 pm - 2:30 pm
  • Break
    2:30 pm - 3:00 pm
  • Towards Understanding the Mixture-of-Experts Layer in Deep Learning
    by Zixiang Chen
    3:00 pm - 3:45 pm
  • Understanding Gradient Descent on Edge of Stability in Deep Learning
    by Zhiyuan Li
    3:45 pm - 4:30 pm
  • Panel Discussion 2
    4:30pm: How to bridge the gap between theory and practice in (deep) representation learning?