Latent Diffusion Energy-Based Model for Interpretable Text Modelling

Peiyu Yu, Sirui Xie, Xiaojian Ma, Baoxiong Jia, Bo Pang, Ruiqi Gao, Yixin Zhu, Song-Chun Zhu, Ying Nian Wu
Proceedings of the 39th International Conference on Machine Learning, PMLR 162:25702-25720, 2022.

Abstract

Latent space Energy-Based Models (EBMs), also known as energy-based priors, have drawn growing interests in generative modeling. Fueled by its flexibility in the formulation and strong modeling power of the latent space, recent works built upon it have made interesting attempts aiming at the interpretability of text modeling. However, latent space EBMs also inherit some flaws from EBMs in data space; the degenerate MCMC sampling quality in practice can lead to poor generation quality and instability in training, especially on data with complex latent structures. Inspired by the recent efforts that leverage diffusion recovery likelihood learning as a cure for the sampling issue, we introduce a novel symbiosis between the diffusion models and latent space EBMs in a variational learning framework, coined as the latent diffusion energy-based model. We develop a geometric clustering-based regularization jointly with the information bottleneck to further improve the quality of the learned latent space. Experiments on several challenging tasks demonstrate the superior performance of our model on interpretable text modeling over strong counterparts.

Cite this Paper

BibTeX
@InProceedings{pmlr-v162-yu22h, title = {Latent Diffusion Energy-Based Model for Interpretable Text Modelling}, author = {Yu, Peiyu and Xie, Sirui and Ma, Xiaojian and Jia, Baoxiong and Pang, Bo and Gao, Ruiqi and Zhu, Yixin and Zhu, Song-Chun and Wu, Ying Nian}, booktitle = {Proceedings of the 39th International Conference on Machine Learning}, pages = {25702--25720}, year = {2022}, editor = {Chaudhuri, Kamalika and Jegelka, Stefanie and Song, Le and Szepesvari, Csaba and Niu, Gang and Sabato, Sivan}, volume = {162}, series = {Proceedings of Machine Learning Research}, month = {17--23 Jul}, publisher = {PMLR}, pdf = {https://proceedings.mlr.press/v162/yu22h/yu22h.pdf}, url = {https://proceedings.mlr.press/v162/yu22h.html}, abstract = {Latent space Energy-Based Models (EBMs), also known as energy-based priors, have drawn growing interests in generative modeling. Fueled by its flexibility in the formulation and strong modeling power of the latent space, recent works built upon it have made interesting attempts aiming at the interpretability of text modeling. However, latent space EBMs also inherit some flaws from EBMs in data space; the degenerate MCMC sampling quality in practice can lead to poor generation quality and instability in training, especially on data with complex latent structures. Inspired by the recent efforts that leverage diffusion recovery likelihood learning as a cure for the sampling issue, we introduce a novel symbiosis between the diffusion models and latent space EBMs in a variational learning framework, coined as the latent diffusion energy-based model. We develop a geometric clustering-based regularization jointly with the information bottleneck to further improve the quality of the learned latent space. Experiments on several challenging tasks demonstrate the superior performance of our model on interpretable text modeling over strong counterparts.} }
Endnote
%0 Conference Paper %T Latent Diffusion Energy-Based Model for Interpretable Text Modelling %A Peiyu Yu %A Sirui Xie %A Xiaojian Ma %A Baoxiong Jia %A Bo Pang %A Ruiqi Gao %A Yixin Zhu %A Song-Chun Zhu %A Ying Nian Wu %B Proceedings of the 39th International Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2022 %E Kamalika Chaudhuri %E Stefanie Jegelka %E Le Song %E Csaba Szepesvari %E Gang Niu %E Sivan Sabato %F pmlr-v162-yu22h %I PMLR %P 25702--25720 %U https://proceedings.mlr.press/v162/yu22h.html %V 162 %X Latent space Energy-Based Models (EBMs), also known as energy-based priors, have drawn growing interests in generative modeling. Fueled by its flexibility in the formulation and strong modeling power of the latent space, recent works built upon it have made interesting attempts aiming at the interpretability of text modeling. However, latent space EBMs also inherit some flaws from EBMs in data space; the degenerate MCMC sampling quality in practice can lead to poor generation quality and instability in training, especially on data with complex latent structures. Inspired by the recent efforts that leverage diffusion recovery likelihood learning as a cure for the sampling issue, we introduce a novel symbiosis between the diffusion models and latent space EBMs in a variational learning framework, coined as the latent diffusion energy-based model. We develop a geometric clustering-based regularization jointly with the information bottleneck to further improve the quality of the learned latent space. Experiments on several challenging tasks demonstrate the superior performance of our model on interpretable text modeling over strong counterparts.
APA
Yu, P., Xie, S., Ma, X., Jia, B., Pang, B., Gao, R., Zhu, Y., Zhu, S. & Wu, Y.N.. (2022). Latent Diffusion Energy-Based Model for Interpretable Text Modelling. Proceedings of the 39th International Conference on Machine Learning, in Proceedings of Machine Learning Research 162:25702-25720 Available from https://proceedings.mlr.press/v162/yu22h.html.

Related Material