@inproceedings{gupta-etal-2019-seeded,
title = "Seeded self-play for language learning",
author = "Gupta, Abhinav and
Lowe, Ryan and
Foerster, Jakob and
Kiela, Douwe and
Pineau, Joelle",
editor = "Mogadala, Aditya and
Klakow, Dietrich and
Pezzelle, Sandro and
Moens, Marie-Francine",
booktitle = "Proceedings of the Beyond Vision and LANguage: inTEgrating Real-world kNowledge (LANTERN)",
month = nov,
year = "2019",
address = "Hong Kong, China",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/D19-6409",
doi = "10.18653/v1/D19-6409",
pages = "62--66",
abstract = "How can we teach artificial agents to use human language flexibly to solve problems in real-world environments? We have an example of this in nature: human babies eventually learn to use human language to solve problems, and they are taught with an adult human-in-the-loop. Unfortunately, current machine learning methods (e.g. from deep reinforcement learning) are too data inefficient to learn language in this way. An outstanding goal is finding an algorithm with a suitable {`}language learning prior{'} that allows it to learn human language, while minimizing the number of on-policy human interactions. In this paper, we propose to learn such a prior in simulation using an approach we call, Learning to Learn to Communicate (L2C). Specifically, in L2C we train a meta-learning agent in simulation to interact with populations of pre-trained agents, each with their own distinct communication protocol. Once the meta-learning agent is able to quickly adapt to each population of agents, it can be deployed in new populations, including populations speaking human language. Our key insight is that such populations can be obtained via self-play, after pre-training agents with imitation learning on a small amount of off-policy human language data. We call this latter technique Seeded Self-Play (S2P). Our preliminary experiments show that agents trained with L2C and S2P need fewer on-policy samples to learn a compositional language in a Lewis signaling game.",
}
<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="gupta-etal-2019-seeded">
<titleInfo>
<title>Seeded self-play for language learning</title>
</titleInfo>
<name type="personal">
<namePart type="given">Abhinav</namePart>
<namePart type="family">Gupta</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Ryan</namePart>
<namePart type="family">Lowe</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Jakob</namePart>
<namePart type="family">Foerster</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Douwe</namePart>
<namePart type="family">Kiela</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Joelle</namePart>
<namePart type="family">Pineau</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<originInfo>
<dateIssued>2019-11</dateIssued>
</originInfo>
<typeOfResource>text</typeOfResource>
<relatedItem type="host">
<titleInfo>
<title>Proceedings of the Beyond Vision and LANguage: inTEgrating Real-world kNowledge (LANTERN)</title>
</titleInfo>
<name type="personal">
<namePart type="given">Aditya</namePart>
<namePart type="family">Mogadala</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Dietrich</namePart>
<namePart type="family">Klakow</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Sandro</namePart>
<namePart type="family">Pezzelle</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Marie-Francine</namePart>
<namePart type="family">Moens</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<originInfo>
<publisher>Association for Computational Linguistics</publisher>
<place>
<placeTerm type="text">Hong Kong, China</placeTerm>
</place>
</originInfo>
<genre authority="marcgt">conference publication</genre>
</relatedItem>
<abstract>How can we teach artificial agents to use human language flexibly to solve problems in real-world environments? We have an example of this in nature: human babies eventually learn to use human language to solve problems, and they are taught with an adult human-in-the-loop. Unfortunately, current machine learning methods (e.g. from deep reinforcement learning) are too data inefficient to learn language in this way. An outstanding goal is finding an algorithm with a suitable ‘language learning prior’ that allows it to learn human language, while minimizing the number of on-policy human interactions. In this paper, we propose to learn such a prior in simulation using an approach we call, Learning to Learn to Communicate (L2C). Specifically, in L2C we train a meta-learning agent in simulation to interact with populations of pre-trained agents, each with their own distinct communication protocol. Once the meta-learning agent is able to quickly adapt to each population of agents, it can be deployed in new populations, including populations speaking human language. Our key insight is that such populations can be obtained via self-play, after pre-training agents with imitation learning on a small amount of off-policy human language data. We call this latter technique Seeded Self-Play (S2P). Our preliminary experiments show that agents trained with L2C and S2P need fewer on-policy samples to learn a compositional language in a Lewis signaling game.</abstract>
<identifier type="citekey">gupta-etal-2019-seeded</identifier>
<identifier type="doi">10.18653/v1/D19-6409</identifier>
<location>
<url>https://aclanthology.org/D19-6409</url>
</location>
<part>
<date>2019-11</date>
<extent unit="page">
<start>62</start>
<end>66</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T Seeded self-play for language learning
%A Gupta, Abhinav
%A Lowe, Ryan
%A Foerster, Jakob
%A Kiela, Douwe
%A Pineau, Joelle
%Y Mogadala, Aditya
%Y Klakow, Dietrich
%Y Pezzelle, Sandro
%Y Moens, Marie-Francine
%S Proceedings of the Beyond Vision and LANguage: inTEgrating Real-world kNowledge (LANTERN)
%D 2019
%8 November
%I Association for Computational Linguistics
%C Hong Kong, China
%F gupta-etal-2019-seeded
%X How can we teach artificial agents to use human language flexibly to solve problems in real-world environments? We have an example of this in nature: human babies eventually learn to use human language to solve problems, and they are taught with an adult human-in-the-loop. Unfortunately, current machine learning methods (e.g. from deep reinforcement learning) are too data inefficient to learn language in this way. An outstanding goal is finding an algorithm with a suitable ‘language learning prior’ that allows it to learn human language, while minimizing the number of on-policy human interactions. In this paper, we propose to learn such a prior in simulation using an approach we call, Learning to Learn to Communicate (L2C). Specifically, in L2C we train a meta-learning agent in simulation to interact with populations of pre-trained agents, each with their own distinct communication protocol. Once the meta-learning agent is able to quickly adapt to each population of agents, it can be deployed in new populations, including populations speaking human language. Our key insight is that such populations can be obtained via self-play, after pre-training agents with imitation learning on a small amount of off-policy human language data. We call this latter technique Seeded Self-Play (S2P). Our preliminary experiments show that agents trained with L2C and S2P need fewer on-policy samples to learn a compositional language in a Lewis signaling game.
%R 10.18653/v1/D19-6409
%U https://aclanthology.org/D19-6409
%U https://doi.org/10.18653/v1/D19-6409
%P 62-66
Markdown (Informal)
[Seeded self-play for language learning](https://aclanthology.org/D19-6409) (Gupta et al., 2019)
ACL
- Abhinav Gupta, Ryan Lowe, Jakob Foerster, Douwe Kiela, and Joelle Pineau. 2019. Seeded self-play for language learning. In Proceedings of the Beyond Vision and LANguage: inTEgrating Real-world kNowledge (LANTERN), pages 62–66, Hong Kong, China. Association for Computational Linguistics.