Learning how not to act in text-based games

Matan Haroush; Tom Zahavy; Daniel J. Mankowitz; Shie Mannor

Learning how not to act in text-based games

Matan Haroush, Tom Zahavy, Daniel J. Mankowitz, Shie Mannor

Technion - Israel Institute of Technology

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

Abstract

Large actions spaces impede an agent’s ability to learn, especially when many of the actions are redundant or irrelevant. This is especially prevalent in text-based domains. We present the action-elimination architecture which combines the generalization power of Deep Reinforcement Learning and the natural language capabilities of NLP architectures to eliminate unnecessary actions and solves quests in the text-based game of Zork, significantly outperforming the baseline agents.

Original language	English
Title of host publication	6th International Conference on Learning Representations, ICLR 2018
State	Published - 2018
Event	6th International Conference on Learning Representations, ICLR 2018 - Vancouver, Canada Duration: 30 Apr 2018 → 3 May 2018

Conference

Conference	6th International Conference on Learning Representations, ICLR 2018
Country/Territory	Canada
City	Vancouver
Period	30/04/18 → 3/05/18

All Science Journal Classification (ASJC) codes

Education
Computer Science Applications
Linguistics and Language
Language and Linguistics

Access to Document

https://openreview.net/pdf?id=B1-tVX1Pz

Cite this

@inproceedings{e3a4127cb95c4a56bd1f900b0ec29861,

title = "Learning how not to act in text-based games",

abstract = "Large actions spaces impede an agent{\textquoteright}s ability to learn, especially when many of the actions are redundant or irrelevant. This is especially prevalent in text-based domains. We present the action-elimination architecture which combines the generalization power of Deep Reinforcement Learning and the natural language capabilities of NLP architectures to eliminate unnecessary actions and solves quests in the text-based game of Zork, significantly outperforming the baseline agents.",

author = "Matan Haroush and Tom Zahavy and Mankowitz, {Daniel J.} and Shie Mannor",

note = "Publisher Copyright: {\textcopyright} 6th International Conference on Learning Representations, ICLR 2018 - Workshop Track Proceedings. All rights reserved.; 6th International Conference on Learning Representations, ICLR 2018 ; Conference date: 30-04-2018 Through 03-05-2018",

year = "2018",

language = "الإنجليزيّة",

booktitle = "6th International Conference on Learning Representations, ICLR 2018",

}

TY - GEN

T1 - Learning how not to act in text-based games

AU - Haroush, Matan

AU - Zahavy, Tom

AU - Mankowitz, Daniel J.

AU - Mannor, Shie

PY - 2018

Y1 - 2018

N2 - Large actions spaces impede an agent’s ability to learn, especially when many of the actions are redundant or irrelevant. This is especially prevalent in text-based domains. We present the action-elimination architecture which combines the generalization power of Deep Reinforcement Learning and the natural language capabilities of NLP architectures to eliminate unnecessary actions and solves quests in the text-based game of Zork, significantly outperforming the baseline agents.

AB - Large actions spaces impede an agent’s ability to learn, especially when many of the actions are redundant or irrelevant. This is especially prevalent in text-based domains. We present the action-elimination architecture which combines the generalization power of Deep Reinforcement Learning and the natural language capabilities of NLP architectures to eliminate unnecessary actions and solves quests in the text-based game of Zork, significantly outperforming the baseline agents.

UR - http://www.scopus.com/inward/record.url?scp=85083951396&partnerID=8YFLogxK

M3 - منشور من مؤتمر

BT - 6th International Conference on Learning Representations, ICLR 2018

T2 - 6th International Conference on Learning Representations, ICLR 2018

Y2 - 30 April 2018 through 3 May 2018

ER -

Learning how not to act in text-based games

Abstract

Conference

All Science Journal Classification (ASJC) codes

Access to Document

Other files and links

Fingerprint

Cite this