Metraipolis › archive › dc_beb60fd00992

Interactive Memory Learning for Long-Term Conversations — arXiv:2609.17088

Source: arXiv (en)
Original: https://arxiv.org/abs/2609.17088
License (as archived): CC0-1.0
Archived: 2026-10-15T18:32:20.868Z
SHA-256 of the archived text: 8a6c551b1096c7615bf08dd640c51586119589d3b80c2e34ef362dfb27c23d9c
Stable URL of this page: https://metraipolis.com/documento/dc_beb60fd00992

How to cite

Interactive Memory Learning for Long-Term Conversations — arXiv:2609.17088 — arXiv (en), https://arxiv.org/abs/2609.17088 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_beb60fd00992 (Metraipolis archive dc_beb60fd00992, sha256 8a6c551b1096c7615bf08dd640c51586119589d3b80c2e34ef362dfb27c23d9c)

BibTeX

@misc{dcbeb60fd00992,
  title = {Interactive Memory Learning for Long-Term Conversations — arXiv:2609.17088},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2609.17088},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_beb60fd00992; sha256 8a6c551b1096c7615bf08dd640c51586119589d3b80c2e34ef362dfb27c23d9c},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.17088. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Recent 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 (InteraCtive Memory Learning), 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.

Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-09-16 in cs.AI).
Categorie arXiv: Artificial Intelligence (cs.AI) ; Computation and Language (cs.CL).
Pagina del lavoro: https://arxiv.org/abs/2609.17088 — autori e testo completo sono lì (qui omessi i nomi: nessun dato personale nel deposito).
Licenza: titolo e abstract sono metadati arXiv distribuiti in CC0 1.0; il testo completo ha la licenza scelta dagli autori, indicata sulla pagina.
Letto il 2026-10-15T18:03:09.506Z dalla pagina https://arxiv.org/catchup/cs.AI/2026-09-16?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).