Position: Compositional Generative Modeling: A Single Model is Not All You Need

Yilun Du, Leslie Pack Kaelbling
Proceedings of the 41st International Conference on Machine Learning, PMLR 235:11721-11732, 2024.

Abstract

Large monolithic generative models trained on massive amounts of data have become an increasingly dominant approach in AI research. In this paper, we argue that we should instead construct large generative systems by composing smaller generative models together. We show how such a compositional generative approach enables us to learn distributions in a more data-efficient manner, enabling generalization to parts of the data distribution unseen at training time. We further show how this enables us to program and construct new generative models for tasks completely unseen at training. Finally, we show that in many cases, we can discover separate compositional components from data.

Cite this Paper

BibTeX
@InProceedings{pmlr-v235-du24d, title = {Position: Compositional Generative Modeling: A Single Model is Not All You Need}, author = {Du, Yilun and Kaelbling, Leslie Pack}, booktitle = {Proceedings of the 41st International Conference on Machine Learning}, pages = {11721--11732}, year = {2024}, editor = {Salakhutdinov, Ruslan and Kolter, Zico and Heller, Katherine and Weller, Adrian and Oliver, Nuria and Scarlett, Jonathan and Berkenkamp, Felix}, volume = {235}, series = {Proceedings of Machine Learning Research}, month = {21--27 Jul}, publisher = {PMLR}, pdf = {https://raw.githubusercontent.com/mlresearch/v235/main/assets/du24d/du24d.pdf}, url = {https://proceedings.mlr.press/v235/du24d.html}, abstract = {Large monolithic generative models trained on massive amounts of data have become an increasingly dominant approach in AI research. In this paper, we argue that we should instead construct large generative systems by composing smaller generative models together. We show how such a compositional generative approach enables us to learn distributions in a more data-efficient manner, enabling generalization to parts of the data distribution unseen at training time. We further show how this enables us to program and construct new generative models for tasks completely unseen at training. Finally, we show that in many cases, we can discover separate compositional components from data.} }
Endnote
%0 Conference Paper %T Position: Compositional Generative Modeling: A Single Model is Not All You Need %A Yilun Du %A Leslie Pack Kaelbling %B Proceedings of the 41st International Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2024 %E Ruslan Salakhutdinov %E Zico Kolter %E Katherine Heller %E Adrian Weller %E Nuria Oliver %E Jonathan Scarlett %E Felix Berkenkamp %F pmlr-v235-du24d %I PMLR %P 11721--11732 %U https://proceedings.mlr.press/v235/du24d.html %V 235 %X Large monolithic generative models trained on massive amounts of data have become an increasingly dominant approach in AI research. In this paper, we argue that we should instead construct large generative systems by composing smaller generative models together. We show how such a compositional generative approach enables us to learn distributions in a more data-efficient manner, enabling generalization to parts of the data distribution unseen at training time. We further show how this enables us to program and construct new generative models for tasks completely unseen at training. Finally, we show that in many cases, we can discover separate compositional components from data.
APA
Du, Y. & Kaelbling, L.P.. (2024). Position: Compositional Generative Modeling: A Single Model is Not All You Need. Proceedings of the 41st International Conference on Machine Learning, in Proceedings of Machine Learning Research 235:11721-11732 Available from https://proceedings.mlr.press/v235/du24d.html.

Related Material