It’s About Time: Temporal References in Emergent Communication

This work studies how artificial agents can develop temporal references in their communication. The experiments show that a small architectural change, rather than a change to the loss function alone, enables agents to express temporal relations reliably.

In cybersecurity, communicating not only what is happening but when and in what sequence is essential for coordinated detection, incident response, and distributed autonomous systems. The paper therefore provides a basis for agents that can share temporal context while making their interactions more understandable and verifiable.