"Another Brick in the AI Wall: Building Practical Solutions from Theoretical Foundations": Workshop at CVPR'25

News

Overview

The shift towards foundation models has overshadowed the unique insights of deep learning theory, resulting in a loss of valuable knowledge and resources for the community. As machine learning and computer vision extend into new domains, such as biology, a deeper understanding of vision tasks becomes increasingly important. This workshop will provide a crucial platform for discussing the systematic challenges of integrating theory and practice.

Some key questions that we are interested in are the following:

• Generalization gap: Modern deep neural networks show that current theories can't accurately predict how large and small models perform differently. Can we close this gap, and if so, how?
• Understanding emergent behaviors: Large networks show behaviors like in-context learning and making multiple predictions at once. How can we explain this theoretically? How can it influence model design and the study of these behaviors?
• Algorithmic advances: For a long time, the field has relied on variations of stochastic gradient descent. As machine learning tasks become more varied, are these algorithms still enough for everything we need? How do modern algorithms affect various vision tasks?

Note that the submissions can be broadly related to theoretical contributions and are not limited to the questions above.

Important Dates

Submission link: https://openreview.net/group?id=thecvf.com/CVPR/2025/Workshop/BASE

Submission format: CVPR template, up to 4 pages for regular submissions (non-archival)

Submission deadline: 15th March 2025, AOE

Author notification: 26th March 2025, AOE