Developmentally motivated emergence of compositional communication via template transfer

TL;DR

This paper introduces a template transfer training regime that enables multi-agent systems to develop compositional communication without architectural biases, inspired by developmental and psycholinguistic theories.

Contribution

It presents a novel training method that transfers learned biases across contexts to promote emergent compositional language in agents, avoiding architectural inductive biases.

Findings

Emergence of compositional communication demonstrated

Achieved zero-shot generalization in communication tasks

Communication protocols are context-independent

Abstract

This paper explores a novel approach to 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 our method, a sender-receiver pair is first trained with disentangled loss functions and then the receiver is transferred to train a new sender with a standard loss. Unlike other methods (e.g. the obverter algorithm), our 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.