End-to-end Differentiable Clustering with Associative Memories

Bishwajit Saha, Dmitry Krotov, Mohammed J Zaki, Parikshit Ram
Proceedings of the 40th International Conference on Machine Learning, PMLR 202:29649-29670, 2023.

Abstract

Clustering is a widely used unsupervised learning technique involving an intensive discrete optimization problem. Associative Memory models or AMs are differentiable neural networks defining a recursive dynamical system, which have been integrated with various deep learning architectures. We uncover a novel connection between the AM dynamics and the inherent discrete assignment necessary in clustering to propose a novel unconstrained continuous relaxation of the discrete clustering problem, enabling end-to-end differentiable clustering with AM, dubbed ClAM. Leveraging the pattern completion ability of AMs, we further develop a novel self-supervised clustering loss. Our evaluations on varied datasets demonstrate that ClAM benefits from the self-supervision, and significantly improves upon both the traditional Lloyd’s k-means algorithm, and more recent continuous clustering relaxations (by upto 60% in terms of the Silhouette Coefficient).

Cite this Paper

BibTeX
@InProceedings{pmlr-v202-saha23a, title = {End-to-end Differentiable Clustering with Associative Memories}, author = {Saha, Bishwajit and Krotov, Dmitry and Zaki, Mohammed J and Ram, Parikshit}, booktitle = {Proceedings of the 40th International Conference on Machine Learning}, pages = {29649--29670}, year = {2023}, editor = {Krause, Andreas and Brunskill, Emma and Cho, Kyunghyun and Engelhardt, Barbara and Sabato, Sivan and Scarlett, Jonathan}, volume = {202}, series = {Proceedings of Machine Learning Research}, month = {23--29 Jul}, publisher = {PMLR}, pdf = {https://proceedings.mlr.press/v202/saha23a/saha23a.pdf}, url = {https://proceedings.mlr.press/v202/saha23a.html}, abstract = {Clustering is a widely used unsupervised learning technique involving an intensive discrete optimization problem. Associative Memory models or AMs are differentiable neural networks defining a recursive dynamical system, which have been integrated with various deep learning architectures. We uncover a novel connection between the AM dynamics and the inherent discrete assignment necessary in clustering to propose a novel unconstrained continuous relaxation of the discrete clustering problem, enabling end-to-end differentiable clustering with AM, dubbed ClAM. Leveraging the pattern completion ability of AMs, we further develop a novel self-supervised clustering loss. Our evaluations on varied datasets demonstrate that ClAM benefits from the self-supervision, and significantly improves upon both the traditional Lloyd’s k-means algorithm, and more recent continuous clustering relaxations (by upto 60% in terms of the Silhouette Coefficient).} }
Endnote
%0 Conference Paper %T End-to-end Differentiable Clustering with Associative Memories %A Bishwajit Saha %A Dmitry Krotov %A Mohammed J Zaki %A Parikshit Ram %B Proceedings of the 40th International Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2023 %E Andreas Krause %E Emma Brunskill %E Kyunghyun Cho %E Barbara Engelhardt %E Sivan Sabato %E Jonathan Scarlett %F pmlr-v202-saha23a %I PMLR %P 29649--29670 %U https://proceedings.mlr.press/v202/saha23a.html %V 202 %X Clustering is a widely used unsupervised learning technique involving an intensive discrete optimization problem. Associative Memory models or AMs are differentiable neural networks defining a recursive dynamical system, which have been integrated with various deep learning architectures. We uncover a novel connection between the AM dynamics and the inherent discrete assignment necessary in clustering to propose a novel unconstrained continuous relaxation of the discrete clustering problem, enabling end-to-end differentiable clustering with AM, dubbed ClAM. Leveraging the pattern completion ability of AMs, we further develop a novel self-supervised clustering loss. Our evaluations on varied datasets demonstrate that ClAM benefits from the self-supervision, and significantly improves upon both the traditional Lloyd’s k-means algorithm, and more recent continuous clustering relaxations (by upto 60% in terms of the Silhouette Coefficient).
APA
Saha, B., Krotov, D., Zaki, M.J. & Ram, P.. (2023). End-to-end Differentiable Clustering with Associative Memories. Proceedings of the 40th International Conference on Machine Learning, in Proceedings of Machine Learning Research 202:29649-29670 Available from https://proceedings.mlr.press/v202/saha23a.html.

Related Material