Metraipolis › archive › dc_5842be370985
ThinkFlow: Self-Evolving Probabilistic Latent Memory for Lifelong Conversational Agents — arXiv:2609.17010
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2609.17010
- License (as archived)
- CC0-1.0
- Archived
- 2026-10-15T18:32:17.927Z
- SHA-256 of the archived text
a27236e8a03c9827458006dcedc240226064997e1b042a5f0a640501e2dc4503- Stable URL of this page
https://metraipolis.com/documento/dc_5842be370985
How to cite
ThinkFlow: Self-Evolving Probabilistic Latent Memory for Lifelong Conversational Agents — arXiv:2609.17010 — arXiv (en), https://arxiv.org/abs/2609.17010 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_5842be370985 (Metraipolis archive dc_5842be370985, sha256 a27236e8a03c9827458006dcedc240226064997e1b042a5f0a640501e2dc4503)
BibTeX
@misc{dc5842be370985,
title = {ThinkFlow: Self-Evolving Probabilistic Latent Memory for Lifelong Conversational Agents — arXiv:2609.17010},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2609.17010},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_5842be370985; sha256 a27236e8a03c9827458006dcedc240226064997e1b042a5f0a640501e2dc4503},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.17010. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
Lifelong conversational agents rely on memory systems to maintain deep, context-aware interactions with users. However, existing explicit textual memory pipelines suffer from a severe information bottleneck, often losing subtle behavioral patterns and emotional shifts. Furthermore, being typically static post-deployment, they cannot autonomously adapt to personal habits and preferences without manual feedback. Cognitive science, however, suggests that humans maintain mental models purely in a latent space and continuously refine them through predictive coding. Inspired by this, we propose \textbf{ThinkFlow}, a novel end-to-end latent memory framework for lifelong conversational agents. ThinkFlow bypasses the text bottleneck by dynamically compressing conversational flows into probabilistic latent memory skills, autonomously consolidating complex user states into disentangled, continuous vectors without semantic interference. To break this barrier, we introduce a test-time evolution paradigm. By coupling teacher-guided latent alignment to bootstrap the initial state with a self-supervised next-user-utterance prediction task for continuous refinement, the framework successfully overcomes cold-start challenges and achieves label-free lifelong personalization. Extensive experiments on long-term conversation benchmarks demonstrate that ThinkFlow significantly outperforms prevailing memory systems, providing highly personalized and contextually accurate responses over extended multi-session interactions.
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/2609.17010 — 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 2026-10-15T18:03:09.496Z 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).