Workshop
Self-Supervised Learning -- Theory and Practice
Pengtao Xie · Shanghang Zhang · Pulkit Agrawal · Ishan Misra · Cynthia Rudin · Abdelrahman Mohamed · Wenzhen Yuan · Barret Zoph · Laurens van der Maaten · Xingyi Yang · Eric Xing
Sat 12 Dec, 8:50 a.m. PST
Self-supervised learning (SSL) is an unsupervised approach for representation learning without relying on human-provided labels. It creates auxiliary tasks on unlabeled input data and learns representations by solving these tasks. SSL has demonstrated great success on images (e.g., MoCo, PIRL, SimCLR) and texts (e.g., BERT) and has shown promising results in other data modalities, including graphs, time-series, audio, etc. On a wide variety of tasks, SSL without using human-provided labels achieves performance that is close to fully supervised approaches.
The existing SSL research mostly focuses on improving the empirical performance without a theoretical foundation. While the proposed SSL approaches are empirically effective, theoretically why they perform well is not clear. For example, why certain auxiliary tasks in SSL perform better than others? How many unlabeled data examples are needed by SSL to learn a good representation? How is the performance of SSL affected by neural architectures?
In this workshop, we aim to bridge this gap between theory and practice. We bring together SSL-interested researchers from various domains to discuss the theoretical foundations of empirically well-performing SSL approaches and how the theoretical insights can further improve SSL’s empirical performance. Different from previous SSL-related workshops which focus on empirical effectiveness of SSL approaches without considering their theoretical foundations, our workshop focuses on establishing the theoretical foundation of SSL and providing theoretical insights for developing new SSL approaches.
We invite submissions of both theoretical works and empirical works, and the intersection of the two. The topics include but are not limited to:
Theoretical foundations of SSL
Sample complexity of SSL methods
Theory-driven design of auxiliary tasks in SSL
Comparative analysis of different auxiliary tasks
Comparative analysis of SSL and supervised approaches
Information theory and SSL
SSL for computer vision, natural language processing, robotics, speech processing, time-series analysis, graph analytics, etc.
SSL for healthcare, social media, neuroscience, biology, social science, etc.
Cognitive foundations of SSL
In addition to invited talks by leading researchers from diverse backgrounds including CV, NLP, robotics, theoretical ML, etc., the workshop will feature poster sessions and panel discussion to share perspectives on establishing foundational understanding of existing SSL approaches and theoretically-principled ways of developing new SSL methods. We accept submissions of short papers (up to 4 pages excluding references in NeurIPS format), which will be peer-reviewed by at least two reviewers. The accepted papers are allowed to be submitted to other conference venues.
Schedule
Sat 8:50 a.m. - 9:00 a.m.
|
Opening remarks
(
Opening remarks
)
>
|
🔗 |
Sat 9:00 a.m. - 9:23 a.m.
|
Invited Talk: Oriol Vinyals
(
Invited Talk
)
>
SlidesLive Video |
Oriol Vinyals 🔗 |
Sat 9:23 a.m. - 9:25 a.m.
|
QA: Oriol Vinyals
(
QA
)
>
|
Oriol Vinyals 🔗 |
Sat 9:25 a.m. - 9:48 a.m.
|
Invited Talk: Ruslan Salakhutdinov
(
Invited Talk
)
>
|
Ruslan Salakhutdinov 🔗 |
Sat 9:48 a.m. - 9:50 a.m.
|
QA: Ruslan Salakhutdinov
(
QA
)
>
|
Ruslan Salakhutdinov 🔗 |
Sat 9:50 a.m. - 10:13 a.m.
|
Invited Talk: Yejin Choi
(
Invited Talk
)
>
SlidesLive Video |
Yejin Choi 🔗 |
Sat 10:13 a.m. - 10:15 a.m.
|
QA: Yejin Choi
(
QA
)
>
|
Yejin Choi 🔗 |
Sat 10:15 a.m. - 11:15 a.m.
|
Poster Session I ( Poster Session ) > link | 🔗 |
Sat 11:15 a.m. - 11:38 a.m.
|
Invited Talk: Jitendra Malik
(
Invited Talk
)
>
|
Jitendra Malik 🔗 |
Sat 11:38 a.m. - 11:40 a.m.
|
QA: Jitendra Malik
(
QA
)
>
|
Jitendra Malik 🔗 |
Sat 11:40 a.m. - 12:03 p.m.
|
Invited Talk: Jia Deng
(
Invited Talk
)
>
|
Jia Deng 🔗 |
Sat 12:03 p.m. - 12:05 p.m.
|
QA: Jia Deng
(
QA
)
>
|
Jia Deng 🔗 |
Sat 12:05 p.m. - 12:28 p.m.
|
Invited Talk: Alexei Efros
(
Invited Talk
)
>
|
Alexei Efros 🔗 |
Sat 12:28 p.m. - 12:30 p.m.
|
QA: Alexei Efros
(
QA
)
>
|
Alexei Efros 🔗 |
Sat 12:30 p.m. - 1:30 p.m.
|
Break
|
🔗 |
Sat 1:30 p.m. - 1:53 p.m.
|
Invited Talk: Yann LeCun
(
Invited Talk
)
>
|
Yann LeCun 🔗 |
Sat 1:53 p.m. - 1:55 p.m.
|
QA: Yann LeCun
(
QA
)
>
|
Yann LeCun 🔗 |
Sat 1:55 p.m. - 2:18 p.m.
|
Invited Talk: Kristen Grauman
(
Invited Talk
)
>
|
Kristen Grauman 🔗 |
Sat 2:18 p.m. - 2:20 p.m.
|
QA: Kristen Grauman
(
QA
)
>
|
Kristen Grauman 🔗 |
Sat 2:20 p.m. - 2:43 p.m.
|
Invited Talk: Katerina Fragkiadaki
(
Invited Talk
)
>
|
Katerina Fragkiadaki 🔗 |
Sat 2:43 p.m. - 2:45 p.m.
|
QA: Katerina Fragkiadaki
(
QA
)
>
|
Katerina Fragkiadaki 🔗 |
Sat 2:45 p.m. - 3:08 p.m.
|
Invited Talk: Abhinav Gupta
(
Invited Talk
)
>
|
Abhinav Gupta 🔗 |
Sat 3:08 p.m. - 3:10 p.m.
|
QA: Abhinav Gupta
(
QA
)
>
|
Abhinav Gupta 🔗 |
Sat 3:10 p.m. - 4:10 p.m.
|
Poster Session II ( Poster Session ) > link | 🔗 |
Sat 4:10 p.m. - 4:33 p.m.
|
Invited Talk: Leonidas J. Guibas
(
Invited Talk
)
>
SlidesLive Video |
Leonidas Guibas 🔗 |
Sat 4:33 p.m. - 4:35 p.m.
|
QA: Leonidas J. Guibas
(
QA
)
>
|
Leonidas Guibas 🔗 |
Sat 4:35 p.m. - 4:58 p.m.
|
Invited Talk: Quoc V. Le
(
Invited Talk
)
>
|
Quoc V. Le 🔗 |
Sat 4:58 p.m. - 5:00 p.m.
|
QA: Quoc V. Le
(
QA
)
>
|
Quoc V. Le 🔗 |
Sat 5:00 p.m. - 5:23 p.m.
|
Invited Talk: Chelsea Finn
(
Invited Talk
)
>
SlidesLive Video |
Chelsea Finn 🔗 |
Sat 5:23 p.m. - 5:25 p.m.
|
QA: Chelsea Finn
(
QA
)
>
|
Chelsea Finn 🔗 |
Sat 5:25 p.m. - 5:39 p.m.
|
Contributed Talk: Yuandong Tian
(
Contributed talk
)
>
SlidesLive Video |
Yuandong Tian 🔗 |
Sat 5:39 p.m. - 5:40 p.m.
|
QA: Yuandong Tian
(
QA
)
>
|
Yuandong Tian 🔗 |
Sat 5:40 p.m. - 6:40 p.m.
|
Panel Discussion & Closing
(
Panel Discussion & Closing
)
>
|
Yejin Choi · Alexei Efros · Chelsea Finn · Kristen Grauman · Quoc V Le · Yann LeCun · Ruslan Salakhutdinov · Eric Xing 🔗 |