It's About Time: Temporal References in Emergent Communication

TL;DR

This work investigates whether temporal references can emerge in emergent communication and identifies architecture as the critical enabler. By comparing a standard Base LSTM with Temporal and TemporalR variants in Temporal Referential Games, the authors show that sequential batching and temporal processing are necessary for robust temporal referencing, achieving over 95% emergence in many runs, even without explicit temporal losses. Temporal references do not consistently boost task accuracy, but they do not hinder and can support better generalization; topographic similarity remains in a similar range across architectures, indicating that temporal grounding can coexist with compositional structure. The study provides transferable architectural insights for incorporating temporal references into diverse emergent communication settings and highlights the role of sequential processing in learning temporal relationships.

Abstract

Emergent communication studies the development of language between autonomous agents, aiming to improve understanding of natural language evolution and increase communication efficiency. While temporal aspects of language have been considered in computational linguistics, there has been no research on temporal references in emergent communication. This paper addresses this gap, by exploring how agents communicate about temporal relationships. We analyse three potential influences for the emergence of temporal references: environmental, external, and architectural changes. Our experiments demonstrate that altering the loss function is insufficient for temporal references to emerge; rather, architectural changes are necessary. However, a minimal change in agent architecture, using a different batching method, allows the emergence of temporal references. This modified design is compared with the standard architecture in a temporal referential games environment, which emphasises temporal relationships. The analysis indicates that over 95\% of the agents with the modified batching method develop temporal references, without changes to their loss function. We consider temporal referencing necessary for future improvements to the agents' communication efficiency, yielding a closer to optimal coding as compared to purely compositional languages. Our readily transferable architectural insights provide the basis for their incorporation into other emergent communication settings.

Figures (24)

• Figure 1: Structure of the referential game and temporal referential game.
• Figure 2: The Base LSTM sender and receiver architectures.
• Figure 3: The Temporal LSTM sender and receiver architectures, with the temporal modules highlighted in purple.
• Figure 4: Examples of regular and temporal batching strategies.
• Figure 5: The TemporalR LSTM sender and receiver architectures, with the temporal modules highlighted in purple.