Learning to Filter with Predictive State Inference Machines

Wen Sun, Arun Venkatraman, Byron Boots, J.Andrew Bagnell
; Proceedings of The 33rd International Conference on Machine Learning, PMLR 48:1197-1205, 2016.

Abstract

Latent state space models are a fundamental and widely used tool for modeling dynamical systems. However, they are difficult to learn from data and learned models often lack performance guarantees on inference tasks such as filtering and prediction. In this work, we present the PREDICTIVE STATE INFERENCE MACHINE (PSIM), a data-driven method that considers the inference procedure on a dynamical system as a composition of predictors. The key idea is that rather than first learning a latent state space model, and then using the learned model for inference, PSIM directly learns predictors for inference in predictive state space. We provide theoretical guarantees for inference, in both realizable and agnostic settings, and showcase practical performance on a variety of simulated and real world robotics benchmarks.

Cite this Paper


BibTeX
@InProceedings{pmlr-v48-sun16, title = {Learning to Filter with Predictive State Inference Machines}, author = {Wen Sun and Arun Venkatraman and Byron Boots and J.Andrew Bagnell}, pages = {1197--1205}, year = {2016}, editor = {Maria Florina Balcan and Kilian Q. Weinberger}, volume = {48}, series = {Proceedings of Machine Learning Research}, address = {New York, New York, USA}, month = {20--22 Jun}, publisher = {PMLR}, pdf = {http://proceedings.mlr.press/v48/sun16.pdf}, url = {http://proceedings.mlr.press/v48/sun16.html}, abstract = {Latent state space models are a fundamental and widely used tool for modeling dynamical systems. However, they are difficult to learn from data and learned models often lack performance guarantees on inference tasks such as filtering and prediction. In this work, we present the PREDICTIVE STATE INFERENCE MACHINE (PSIM), a data-driven method that considers the inference procedure on a dynamical system as a composition of predictors. The key idea is that rather than first learning a latent state space model, and then using the learned model for inference, PSIM directly learns predictors for inference in predictive state space. We provide theoretical guarantees for inference, in both realizable and agnostic settings, and showcase practical performance on a variety of simulated and real world robotics benchmarks.} }
Endnote
%0 Conference Paper %T Learning to Filter with Predictive State Inference Machines %A Wen Sun %A Arun Venkatraman %A Byron Boots %A J.Andrew Bagnell %B Proceedings of The 33rd International Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2016 %E Maria Florina Balcan %E Kilian Q. Weinberger %F pmlr-v48-sun16 %I PMLR %J Proceedings of Machine Learning Research %P 1197--1205 %U http://proceedings.mlr.press %V 48 %W PMLR %X Latent state space models are a fundamental and widely used tool for modeling dynamical systems. However, they are difficult to learn from data and learned models often lack performance guarantees on inference tasks such as filtering and prediction. In this work, we present the PREDICTIVE STATE INFERENCE MACHINE (PSIM), a data-driven method that considers the inference procedure on a dynamical system as a composition of predictors. The key idea is that rather than first learning a latent state space model, and then using the learned model for inference, PSIM directly learns predictors for inference in predictive state space. We provide theoretical guarantees for inference, in both realizable and agnostic settings, and showcase practical performance on a variety of simulated and real world robotics benchmarks.
RIS
TY - CPAPER TI - Learning to Filter with Predictive State Inference Machines AU - Wen Sun AU - Arun Venkatraman AU - Byron Boots AU - J.Andrew Bagnell BT - Proceedings of The 33rd International Conference on Machine Learning PY - 2016/06/11 DA - 2016/06/11 ED - Maria Florina Balcan ED - Kilian Q. Weinberger ID - pmlr-v48-sun16 PB - PMLR SP - 1197 DP - PMLR EP - 1205 L1 - http://proceedings.mlr.press/v48/sun16.pdf UR - http://proceedings.mlr.press/v48/sun16.html AB - Latent state space models are a fundamental and widely used tool for modeling dynamical systems. However, they are difficult to learn from data and learned models often lack performance guarantees on inference tasks such as filtering and prediction. In this work, we present the PREDICTIVE STATE INFERENCE MACHINE (PSIM), a data-driven method that considers the inference procedure on a dynamical system as a composition of predictors. The key idea is that rather than first learning a latent state space model, and then using the learned model for inference, PSIM directly learns predictors for inference in predictive state space. We provide theoretical guarantees for inference, in both realizable and agnostic settings, and showcase practical performance on a variety of simulated and real world robotics benchmarks. ER -
APA
Sun, W., Venkatraman, A., Boots, B. & Bagnell, J.. (2016). Learning to Filter with Predictive State Inference Machines. Proceedings of The 33rd International Conference on Machine Learning, in PMLR 48:1197-1205

Related Material