Interaction history as a source of compositionality in emergent communication

Abstract

In this paper, we explore interaction history as a particular source of pressure for achieving emergent compositional communication in multi-agent systems. We propose a training regime implementing template transfer, the idea of carrying over learned biases across contexts. In the presented method, a sender-receiver dyad is first trained with a disentangled pair of objectives, and then the receiver is transferred to train a new sender with a standard objective. Unlike other methods (e.g. the obverter algorithm), the template transfer approach does not require imposing inductive biases on the architecture of the agents. We experimentally show the emergence of compositional communication using topographical similarity, zero-shot generalization and context-independence as evaluation metrics. The presented approach is connected to an important line of work in semiotics and developmental psycholinguistics: it supports a conjecture that compositional communication is scaffolded on simpler communication protocols.

Recursive Neural Networks with Bottlenecks Diagnose (Non-)Compositionality

“…Other instantiations of the TRE are from literature on language emergence in signalling games, where the degree of compositionality of that language is measured. Korbak et al (2020) contrast TRE and six other compositionality metrics for signalling games where the colour and shape of an object are communicated. Examples of such metrics are topographic similarity, positional disentanglement and context independence.…”

Models of symbol emergence in communication: a conceptual review and a guide for avoiding local minima

“…The goal is not to faithfully portray actual processes but rather to support certain philosophical positions by demonstrating the theoretical possibility of the bottom-up emergence of conventions. In recent years, this thread has been developed to account for more complex communication processes (Barrett & Skyrms, 2017;Barrett, Skyrms, & Mohseni, 2019;Korbak, Zubek, Kuciński, Miłoś, & Rączaszek-Leonardi, 2021;Skyrms, 2010).…”

“…Zubek et al (2017) studied reactions of a population of agents to a certain trajectory of changes in the environment and demonstrated that the structure of language categories is affected by history. Korbak et al (2021) presented a developmentally motivated model in which an agent goes through two stages of learning, which helps it develop a compositional language.…”

RLupus: Cooperation through emergent communication in The Werewolf social deduction game

“…Study of compositionality Compositionality is a highly desirable property in both human and artificial languages [Baroni, 2020], as it allows for the generalization of concepts and ideas [Pelletier, 1994, Janssen andPartee, 1997]. However, defining compositionality can be challenging [Korbak et al, 2020, Andreas, 2019, and its necessity for generalization has been debated in the literature . In their study, investigate input reconstruction in a simplified signaling game.…”

“…Other works, such as Choi et al [2018] and , attribute the emergence of compositionality to environmental constraints rather than specific model architecture, highlighting the importance of constructing appropriate settings for language development. For example, Korbak et al [2019] employs curriculum learning, gradually increasing the difficulty of referential games, and reports an emergence of compositionality using topographic similarity and zero-shot generalization accuracy. Similarly, the introduction of iterative learning, see Section 2.3, in the language development process, has been shown to lead to increasingly compositional languages with each generation [Cogswell et al, 2019, Ren et al, 2020, Tieleman et al, 2019, Chaabouni et al, 2019a.…”