Skip to content
conftrace_
2026 EMNLP EMNLP 2026

Interactive Memory Learning for Long-Term Conversations

AbstractRecent advancements in large language models have significantly enhanced the capabilities of agents in modeling long-term conversations. Despite these successes, existing approaches typically adopt a static heuristic paradigm, where information is passively archived without adaptive memory valuation. Consequently, these methods fail to self-evolve or align their memory management with evolving user needs. To address this, we propose **ICML** (**I**ntera**C**tive **M**emory **L**earning), a multi-agent framework that transforms the memory mechanism from a passive archive into a learnable, interactive memory policy. Specifically, we first employ a session synthesis pipeline to generate expert data, facilitating rapid test-time adaptation in unseen scenarios. Building on this, **ICML** utilizes an online reinforcement learning mechanism where a Planner agent selectively encodes high-value information and a Trigger agent dynamically retrieves it to optimize response quality, whereby the two agents co-evolve through continuous interaction feedback. Crucially, both agents are synchronized through a delayed reward mechanism that propagates future feedback back to earlier storage decisions, ensuring memory policies are precisely aligned with user expectations. Experimental results demonstrate that **ICML** significantly outperforms strong baselines, exhibiting the unique capability to continuously improve response quality as interactions accumulate.

More from EMNLP 2026

All EMNLP 2026 papers →