Reinforcement of Semantic Representations in Pragmatic Agents Leads to the Emergence of a Mutual Exclusivity Bias
Autor(en): | Ohmer, X. König, P. Franke, M. |
Stichwörter: | Communication games; gradient-based learning; Learning systems; Literals; Mutual exclusivities; mutual exclusivity; Online inferences; Rational Speech Act model; reinforcement learning; Reinforcement learnings; Semantic representation; Semantics, Artificial agents; Speech act modeling, Reinforcement learning | Erscheinungsdatum: | 2020 | Herausgeber: | The Cognitive Science Society | Journal: | Proceedings for the 42nd Annual Meeting of the Cognitive Science Society: Developing a Mind: Learning in Humans, Animals, and Machines, CogSci 2020 | Startseite: | 1779 | Seitenende: | 1785 | Zusammenfassung: | We present a novel framework for building pragmatic artificial agents with explicit and trainable semantic representations, using the Rational Speech Act model. We train our agents on supervised and unsupervised communication games and compare their behavior to literal agents lacking pragmatic abilities. For both types of games pragmatic but not literal agents evolve a mutual exclusivity bias. This provides a computational pragmatic account of mutual exclusivity and points out a possible direction for solving the mutual exclusivity bias challenge posed by Gandhi and Lake (2019). We find that pragmatic reasoning can cause the bias either by promoting lexical constraints during learning, or by affecting online inference. In addition we show that pragmatic abilities lead to faster learning and that this advantage is even stronger when meanings to be communicated follow a more natural distribution as described by Zipf's law. © 2020 The Author(s) |
Beschreibung: | Conference of 42nd Annual Meeting of the Cognitive Science Society: Developing a Mind: Learning in Humans, Animals, and Machines, CogSci 2020 ; Conference Date: 29 July 2020 Through 1 August 2020; Conference Code:182812 |
Externe URL: | https://www.scopus.com/inward/record.uri?eid=2-s2.0-85106169328&partnerID=40&md5=efb874f96dce1ea0f5a172137b33d89b |
Zur Langanzeige