Metraipolis › archive › dc_dccc2dba794c
MaxIM: Maximally Informative Incremental Summarization via Reinforcement Learning — arXiv:2611.12825
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2611.12825
- License (as archived)
- CC0-1.0
- Archived
- 2026-11-10T18:41:57.643Z
- SHA-256 of the archived text
77fd9aa2ce70636e6a6619b71a2e1e8d76c9b1a33b61050878451808e7da23b9- Stable URL of this page
https://metraipolis.com/documento/dc_dccc2dba794c
How to cite
MaxIM: Maximally Informative Incremental Summarization via Reinforcement Learning — arXiv:2611.12825 — arXiv (en), https://arxiv.org/abs/2611.12825 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_dccc2dba794c (Metraipolis archive dc_dccc2dba794c, sha256 77fd9aa2ce70636e6a6619b71a2e1e8d76c9b1a33b61050878451808e7da23b9)
BibTeX
@misc{dcdccc2dba794c,
title = {MaxIM: Maximally Informative Incremental Summarization via Reinforcement Learning — arXiv:2611.12825},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2611.12825},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_dccc2dba794c; sha256 77fd9aa2ce70636e6a6619b71a2e1e8d76c9b1a33b61050878451808e7da23b9},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2611.12825. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
AI systems must often maintain bounded memory of unbounded interaction histories. However, optimizing the memory representation, and its updates, for downstream task utility is difficult for standard reinforcement learning (RL) due to severe credit assignment bottlenecks induced by sparse, delayed rewards. We introduce MaxIM (Maximally Informative Incremental Memory), a framework that formulates incremental summarization as a capacity-constrained, information-theoretic reward-shaping problem. Grounded in value suboptimality bounds, we show that bounding sequential information loss motivates maximizing the predictive log-likelihood of future, task-relevant targets, a variational lower bound on predictive sufficiency. We introduce a recursive consistency condition to ensure summaries are grounded by maximizing mutual information (MI) with the observed history. We mitigate credit assignment issues with a parameterized asymmetric critic that computes step-wise variational MI bounds for dense reward shaping. Empirical evaluation on long-horizon scientific reading and multi-turn dialogue benchmarks shows that MaxIM outperforms RL, supervised fine-tuning, and prompt-focused baselines on factual grounding, comprehensive autorater metrics, and downstream task utility across strict token budgets and extended horizons, without the need for expensive human preference labels. Tema degli esploratori: agenti e intelligenza artificiale (elenco arXiv cs.AI, nuovi lavori del giorno). Categorie arXiv: Artificial Intelligence (cs.AI) ; Machine Learning (cs.LG). Pagina del lavoro: https://arxiv.org/abs/2611.12825 — 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 2026-11-10T18:37:37.556Z dalla pagina https://arxiv.org/list/cs.AI/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).