Softstar: Heuristic-guided probabilistic inference

Mathew Monfort, Brenden Lake, Brian D. Ziebart, Patrick Lucey, Joshua B. Tenenbaum

Research output: Contribution to journalArticle

Abstract

Recent machine learning methods for sequential behavior prediction estimate the motives of behavior rather than the behavior itself. This higher-level abstraction improves generalization in different prediction settings, but computing predictions often becomes intractable in large decision spaces. We propose the Softstar algorithm, a softened heuristic-guided search technique for the maximum entropy inverse optimal control model of sequential behavior. This approach supports probabilistic search with bounded approximation error at a significantly reduced computational cost when compared to sampling based methods. We present the algorithm, analyze approximation guarantees, and compare performance with simulation-based inference on two distinct complex decision tasks.

Original languageEnglish (US)
Pages (from-to)2764-2772
Number of pages9
JournalAdvances in Neural Information Processing Systems
Volume2015-January
StatePublished - 2015

Fingerprint

Approximation algorithms
Learning systems
Entropy
Sampling
Costs

ASJC Scopus subject areas

  • Computer Networks and Communications
  • Information Systems
  • Signal Processing

Cite this

Monfort, M., Lake, B., Ziebart, B. D., Lucey, P., & Tenenbaum, J. B. (2015). Softstar: Heuristic-guided probabilistic inference. Advances in Neural Information Processing Systems, 2015-January, 2764-2772.

Softstar : Heuristic-guided probabilistic inference. / Monfort, Mathew; Lake, Brenden; Ziebart, Brian D.; Lucey, Patrick; Tenenbaum, Joshua B.

In: Advances in Neural Information Processing Systems, Vol. 2015-January, 2015, p. 2764-2772.

Research output: Contribution to journalArticle

Monfort, M, Lake, B, Ziebart, BD, Lucey, P & Tenenbaum, JB 2015, 'Softstar: Heuristic-guided probabilistic inference', Advances in Neural Information Processing Systems, vol. 2015-January, pp. 2764-2772.
Monfort M, Lake B, Ziebart BD, Lucey P, Tenenbaum JB. Softstar: Heuristic-guided probabilistic inference. Advances in Neural Information Processing Systems. 2015;2015-January:2764-2772.
Monfort, Mathew ; Lake, Brenden ; Ziebart, Brian D. ; Lucey, Patrick ; Tenenbaum, Joshua B. / Softstar : Heuristic-guided probabilistic inference. In: Advances in Neural Information Processing Systems. 2015 ; Vol. 2015-January. pp. 2764-2772.
@article{6b80f9eb1b1e4cad9380858bd574d91b,
title = "Softstar: Heuristic-guided probabilistic inference",
abstract = "Recent machine learning methods for sequential behavior prediction estimate the motives of behavior rather than the behavior itself. This higher-level abstraction improves generalization in different prediction settings, but computing predictions often becomes intractable in large decision spaces. We propose the Softstar algorithm, a softened heuristic-guided search technique for the maximum entropy inverse optimal control model of sequential behavior. This approach supports probabilistic search with bounded approximation error at a significantly reduced computational cost when compared to sampling based methods. We present the algorithm, analyze approximation guarantees, and compare performance with simulation-based inference on two distinct complex decision tasks.",
author = "Mathew Monfort and Brenden Lake and Ziebart, {Brian D.} and Patrick Lucey and Tenenbaum, {Joshua B.}",
year = "2015",
language = "English (US)",
volume = "2015-January",
pages = "2764--2772",
journal = "Advances in Neural Information Processing Systems",
issn = "1049-5258",

}

TY - JOUR

T1 - Softstar

T2 - Heuristic-guided probabilistic inference

AU - Monfort, Mathew

AU - Lake, Brenden

AU - Ziebart, Brian D.

AU - Lucey, Patrick

AU - Tenenbaum, Joshua B.

PY - 2015

Y1 - 2015

N2 - Recent machine learning methods for sequential behavior prediction estimate the motives of behavior rather than the behavior itself. This higher-level abstraction improves generalization in different prediction settings, but computing predictions often becomes intractable in large decision spaces. We propose the Softstar algorithm, a softened heuristic-guided search technique for the maximum entropy inverse optimal control model of sequential behavior. This approach supports probabilistic search with bounded approximation error at a significantly reduced computational cost when compared to sampling based methods. We present the algorithm, analyze approximation guarantees, and compare performance with simulation-based inference on two distinct complex decision tasks.

AB - Recent machine learning methods for sequential behavior prediction estimate the motives of behavior rather than the behavior itself. This higher-level abstraction improves generalization in different prediction settings, but computing predictions often becomes intractable in large decision spaces. We propose the Softstar algorithm, a softened heuristic-guided search technique for the maximum entropy inverse optimal control model of sequential behavior. This approach supports probabilistic search with bounded approximation error at a significantly reduced computational cost when compared to sampling based methods. We present the algorithm, analyze approximation guarantees, and compare performance with simulation-based inference on two distinct complex decision tasks.

UR - http://www.scopus.com/inward/record.url?scp=84965159709&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84965159709&partnerID=8YFLogxK

M3 - Article

VL - 2015-January

SP - 2764

EP - 2772

JO - Advances in Neural Information Processing Systems

JF - Advances in Neural Information Processing Systems

SN - 1049-5258

ER -