MemFly: On-the-Fly Memory Optimization via Information Bottleneck
Fuente:
arXiv
Enregistré dans:
| Auteurs principaux: | , , , , , , |
|---|---|
| Format: | Preprint |
| Publié: |
2026
|
| Sujets: | |
| Accès en ligne: | |
| Tags: |
Ajouter un tag
Pas de tags, Soyez le premier à ajouter un tag!
|
| _version_ | 1866908820140195840 |
|---|---|
| author | Zhang, Zhenyuan Jia, Xianzhang Yang, Zhiqin Song, Zhenbo Xue, Wei Han, Sirui Guo, Yike |
| author_facet | Zhang, Zhenyuan Jia, Xianzhang Yang, Zhiqin Song, Zhenbo Xue, Wei Han, Sirui Guo, Yike |
| contents | Long-term memory enables large language model agents to tackle complex tasks through historical interactions. However, existing frameworks encounter a fundamental dilemma between compressing redundant information efficiently and maintaining precise retrieval for downstream tasks. To bridge this gap, we propose MemFly, a framework grounded in information bottleneck principles that facilitates on-the-fly memory evolution for LLMs. Our approach minimizes compression entropy while maximizing relevance entropy via a gradient-free optimizer, constructing a stratified memory structure for efficient storage. To fully leverage MemFly, we develop a hybrid retrieval mechanism that seamlessly integrates semantic, symbolic, and topological pathways, incorporating iterative refinement to handle complex multi-hop queries. Comprehensive experiments demonstrate that MemFly substantially outperforms state-of-the-art baselines in memory coherence, response fidelity, and accuracy. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2602_07885 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | MemFly: On-the-Fly Memory Optimization via Information Bottleneck Zhang, Zhenyuan Jia, Xianzhang Yang, Zhiqin Song, Zhenbo Xue, Wei Han, Sirui Guo, Yike Artificial Intelligence Machine Learning Long-term memory enables large language model agents to tackle complex tasks through historical interactions. However, existing frameworks encounter a fundamental dilemma between compressing redundant information efficiently and maintaining precise retrieval for downstream tasks. To bridge this gap, we propose MemFly, a framework grounded in information bottleneck principles that facilitates on-the-fly memory evolution for LLMs. Our approach minimizes compression entropy while maximizing relevance entropy via a gradient-free optimizer, constructing a stratified memory structure for efficient storage. To fully leverage MemFly, we develop a hybrid retrieval mechanism that seamlessly integrates semantic, symbolic, and topological pathways, incorporating iterative refinement to handle complex multi-hop queries. Comprehensive experiments demonstrate that MemFly substantially outperforms state-of-the-art baselines in memory coherence, response fidelity, and accuracy. |
| title | MemFly: On-the-Fly Memory Optimization via Information Bottleneck |
| topic | Artificial Intelligence Machine Learning |
| url | https://arxiv.org/abs/2602.07885 |