Unified Deep Discrete Representation Learning Framework
Keywords: Discrete representation learning
Abstract: Recent years have seen significant success of deep discrete representation learning across a broad range of domains. Existing frameworks typically operate on flat latents, overlooking the hierarchical structure within discrete latent spaces, which if effectively exploited could yield richer and more expressive representations. This work contributes a novel hierarchical discrete representation learning framework that flexibly generalizes to a variety of tasks and demonstrates effectiveness across diverse applications. We provide a theoretical analysis on sample complexity and additionally study the effect of codebook utilization on task performance. We offer practical insights into how these factors interplay in different learning scenarios.
Primary Area: unsupervised, self-supervised, semi-supervised, and supervised representation learning
Code Of Ethics: I acknowledge that I and all co-authors of this work have read and commit to adhering to the ICLR Code of Ethics.
Submission Guidelines: I certify that this submission complies with the submission instructions as described on https://iclr.cc/Conferences/2025/AuthorGuide.
Reciprocal Reviewing: I understand the reciprocal reviewing requirement as described on https://iclr.cc/Conferences/2025/CallForPapers. If none of the authors are registered as a reviewer, it may result in a desk rejection at the discretion of the program chairs. To request an exception, please complete this form at https://forms.gle/Huojr6VjkFxiQsUp6.
Anonymous Url: I certify that there is no URL (e.g., github page) that could be used to find authors’ identity.
No Acknowledgement Section: I certify that there is no acknowledgement section in this submission for double blind review.
Submission Number: 5935
Loading