The DOI matches its Crossref record; 4 of the 4 references it cites were found in a registry or an index.
Explicit Memory Learning with Expectation Maximization
Open accessAbstract
Large Language Models (LLMs) have revolutionized the landscape of natural language processing, demonstrating remarkable abilities across various complex tasks.However, their stateless nature limits the capability to retain information across interactions, hindering performance in scenarios requiring historical context recall.To mitigate this, current approaches primarily use explicit memory to allow LLMs to store useful information, which is accessible, readable, and interpretable.Nevertheless, explicit memory lacks the reliable learning mechanisms of implicit memory, which can be optimized end-to-end.To harness the benefits of both, we introduce EM 2 , a novel framework enhancing explicit memory updates via the Expectation-Maximization (EM) algorithm.EM 2 treats memory as a latent variable, ensuring continual learning and improvement during updates.Experimental results on streaming inference tasks demonstrate that EM 2 outperforms existing methods without memory or with static external memory.Our in-depth analysis highlights that EM 2 significantly enhances performance across various backbones and memory strategies, providing a robust solution for advancing LLM memory management and enabling explicit memory to learn and improve similarly to implicit memory.
Overview
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
Claims other papers make about this one
Each is checked against this paper’s own text: does it say what the citing paper says it does?
EM-based formulations ensure memory consistency under distribution shifts.
“and EM-based formulations ensure memory consistency under distribution shifts (Yin et al., 2024).”
From From Storage to Experience: A Survey on the Evolution of LLM Agent Memory Mechanisms (2026) · cites this paper as Yin et al., 2024
Citations
Cites
The publisher hasn’t deposited this paper’s reference list.
Cited by
Sorted by citations
MemIndex: Agentic Event-based Distributed Memory Management for Multi-agent Systems
Saleh, A., Tarkoma, S., Lindgren, A., Donta, P. K., Dustdar, S., Pirttikangas, S., & Lovén, L. (2025) · ACM Transactions on Autonomous and Adaptive Systems · doi:10.1145/3774946 · Cited by 11
How Memory Management Impacts LLM Agents: An Empirical Study of Experience-Following Behavior
Xiong, Z., Lin, Y., Xie, W., He, P., Liu, Z., Tang, J., Lakkaraju, H., & Xiang, Z. (2026) · Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers) · doi:10.18653/v1/2026.acl-long.27 · Cited by 4
From Storage to Experience: A Survey on the Evolution of LLM Agent Memory Mechanisms
Luo, J., Tian, Y., Cao, C., Luo, Z., Lin, H., Li, K., Kong, C., Yang, R., & Ma, J. (2026) · Findings of the Association for Computational Linguistics: ACL 2026 · doi:10.18653/v1/2026.findings-acl.2069 · Cited by 3
“Preference-aware updates integrate short-term variation with long-term trends (Sun et al., 2025b), and EM-based formulations ensure memory consistency under distribution shifts (Yin et al., 2024).”
Memory Injection Attacks on LLM Agents via Query-Only Interaction
Dong, S., Xu, S., He, P., Li, Y., Tang, J., Liu, T., Liu, H., & Xiang, Z. (2025) · arXiv · doi:10.48550/arxiv.2503.03704 · Cited by 124
“Recent studies have focused on enhancing the effectiveness of memory storage and utilization by developing innovative memory management strategies [44, 47, 53].”
Evaluating Memory in LLM Agents via Incremental Multi-Turn Interactions
Hu, Y., Wang, Y., & McAuley, J. (2025) · arXiv · doi:10.48550/arxiv.2507.05257 · Cited by 1
“In contrast, memory encoded in model parameters [44, 46, 51] remains largely within academic research and is typically less capable than proprietary memory systems equipped on closed-sourced API models.”
Metadata
- Authors
- Yin, Z., Sun, Q., Guo, Q., Zeng, Z., Cheng, Q., Qiu, X., & Huang, X.
- Affiliations
- Fudan University (CN)
- Shanghai Artificial Intelligence Laboratory (CN)
- Venue
- Proceedings of the 2024 Conference on Empirical Methods in Natural Language Processing
- Published
- 2024
- Pages
- 16618-16635
- OpenAlex
- W4404782344
- Cited by
- 1
- References
- 0
- Citations by year
- 2025: 1
- FWCI
- 0.29
- Type
- conference paper
- Language
- en
- Open access
- hybrid
- doi.org
- Licence
- cc-by
- Locations
- Conference on Empirical Methods in Natural Language Processing (EMNLP) (publishedVersion, cc-by)
- Funders
- National Natural Science Foundation of China (62236004)
- Fudan University
- Topics
- Physical Sciences › Computer Science › Artificial Intelligence › Topic Modeling
- Physical Sciences › Computer Science › Artificial Intelligence › Natural Language Processing Techniques
- Physical Sciences › Computer Science › Artificial Intelligence › Speech and dialogue systems
- Keywords
- declarative memory, large language models, expectation-maximization algorithm, memory updating, continual learning, streaming inference, memory management
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
