MemGuard: Preventing Memory Contamination in Long-Term Memory-Augmented Large Language Models
Fuente:
arXiv
Saved in:
| Main Authors: | , , , , , , , , , |
|---|---|
| Format: | Preprint |
| Published: |
2026
|
| Subjects: | |
| Online Access: | |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866916053361098752 |
|---|---|
| author | Ha, Hyeonjeong Kim, Jeonghwan Qian, Cheng Liu, Jiayu Campbell, William M. Wu, Yue Zhang, Yuji McKeown, Kathleen Hakkani-Tur, Dilek Ji, Heng |
| author_facet | Ha, Hyeonjeong Kim, Jeonghwan Qian, Cheng Liu, Jiayu Campbell, William M. Wu, Yue Zhang, Yuji McKeown, Kathleen Hakkani-Tur, Dilek Ji, Heng |
| contents | Memory-augmented large language models extend reasoning beyond a fixed context window by maintaining long-term memory across interactions. However, existing memory systems often collapse stable user facts, episodic events, and behavioral rules into a shared space, allowing functionally distinct memories to be retrieved and used as interchangeable evidence. We identify this failure mode as heterogeneous memory contamination, where context-specific events become overgeneralized claims, or semantically relevant but functionally incompatible memories mislead generation. To this end, we introduce MemGuard, a type-aware memory framework that preserves functional memory boundaries during memory construction and retrieval. It assigns each memory an explicit functional role at write time, maintains relations across type-isolated memories, and selectively composes evidence only from necessary memory types, reducing contamination from irrelevant or functionally incompatible evidence. Across hallucination and long-horizon conversation benchmarks, MemGuard improves memory reliability by up to 28.27% while retrieving up to 5.8x fewer memory tokens than prior methods. These results suggest that reliable long-term reasoning depends on principled organization and selective use of heterogeneous memory. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2605_28009 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | MemGuard: Preventing Memory Contamination in Long-Term Memory-Augmented Large Language Models Ha, Hyeonjeong Kim, Jeonghwan Qian, Cheng Liu, Jiayu Campbell, William M. Wu, Yue Zhang, Yuji McKeown, Kathleen Hakkani-Tur, Dilek Ji, Heng Computation and Language Artificial Intelligence Machine Learning Memory-augmented large language models extend reasoning beyond a fixed context window by maintaining long-term memory across interactions. However, existing memory systems often collapse stable user facts, episodic events, and behavioral rules into a shared space, allowing functionally distinct memories to be retrieved and used as interchangeable evidence. We identify this failure mode as heterogeneous memory contamination, where context-specific events become overgeneralized claims, or semantically relevant but functionally incompatible memories mislead generation. To this end, we introduce MemGuard, a type-aware memory framework that preserves functional memory boundaries during memory construction and retrieval. It assigns each memory an explicit functional role at write time, maintains relations across type-isolated memories, and selectively composes evidence only from necessary memory types, reducing contamination from irrelevant or functionally incompatible evidence. Across hallucination and long-horizon conversation benchmarks, MemGuard improves memory reliability by up to 28.27% while retrieving up to 5.8x fewer memory tokens than prior methods. These results suggest that reliable long-term reasoning depends on principled organization and selective use of heterogeneous memory. |
| title | MemGuard: Preventing Memory Contamination in Long-Term Memory-Augmented Large Language Models |
| topic | Computation and Language Artificial Intelligence Machine Learning |
| url | https://arxiv.org/abs/2605.28009 |