Marino, Joseph and Chen, Lei and He, Jiawei and Mandt, Stephan (2022) Improving sequential latent variable models with autoregressive flows. Machine Learning, 111 (4). pp. 1597-1620. ISSN 0885-6125. doi:10.1007/s10994-021-06092-6. https://resolver.caltech.edu/CaltechAUTHORS:20220104-767137600
![]() |
PDF
- Submitted Version
See Usage Policy. 7MB |
Use this Persistent URL to link to this item: https://resolver.caltech.edu/CaltechAUTHORS:20220104-767137600
Abstract
We propose an approach for improving sequence modeling based on autoregressive normalizing flows. Each autoregressive transform, acting across time, serves as a moving frame of reference, removing temporal correlations and simplifying the modeling of higher-level dynamics. This technique provides a simple, general-purpose method for improving sequence modeling, with connections to existing and classical techniques. We demonstrate the proposed approach both with standalone flow-based models and as a component within sequential latent variable models. Results are presented on three benchmark video datasets and three other time series datasets, where autoregressive flow-based dynamics improve log-likelihood performance over baseline models. Finally, we illustrate the decorrelation and improved generalization properties of using flow-based dynamics.
Item Type: | Article | |||||||||||||||
---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
Related URLs: |
| |||||||||||||||
ORCID: |
| |||||||||||||||
Additional Information: | © The Author(s), under exclusive licence to Springer Science+Business Media LLC, part of Springer Nature 2021. Received 15 November 2020; Revised 02 August 2021; Accepted 30 September 2021; Published 18 November 2021. Author Contributions: Joseph Marino: forming idea, running simulation and writing paper; Lei Chen: running simulation; Jiawei He: writing paper, and Stephan Mandt: forming idea, writing paper and providing feedback. All authors worked toegther for most of the ta sks involved. Conflict of interest: California Institute of Technology (caltech.edu); Simon Fraser University (sfu.ca); University of California Irvine (uci.edu); Disney Research (disneyresearch.com); Borealis AI (borealisai.com); DeepMind (deepmind.com, google.com). Code availability: Source code is available at https://anonymous.4open.science/r/f02199f7-86d2-45ee-ad23-3f13f769ee10/. | |||||||||||||||
Subject Keywords: | Autoregressive flows; Latent variable models; Sequence modeling | |||||||||||||||
Issue or Number: | 4 | |||||||||||||||
DOI: | 10.1007/s10994-021-06092-6 | |||||||||||||||
Record Number: | CaltechAUTHORS:20220104-767137600 | |||||||||||||||
Persistent URL: | https://resolver.caltech.edu/CaltechAUTHORS:20220104-767137600 | |||||||||||||||
Official Citation: | Marino, J., Chen, L., He, J. et al. Improving sequential latent variable models with autoregressive flows. Mach Learn 111, 1597–1620 (2022). https://doi.org/10.1007/s10994-021-06092-6 | |||||||||||||||
Usage Policy: | No commercial reproduction, distribution, display or performance rights in this work are provided. | |||||||||||||||
ID Code: | 112663 | |||||||||||||||
Collection: | CaltechAUTHORS | |||||||||||||||
Deposited By: | Tony Diaz | |||||||||||||||
Deposited On: | 04 Jan 2022 19:50 | |||||||||||||||
Last Modified: | 01 Jun 2022 17:51 |
Repository Staff Only: item control page