FadeMem: Biologically-Inspired Forgetting for Efficient Agent Memory

2026-01-26

AI Summary

This work addresses the inefficiencies in large language model (LLM) agents stemming from the absence of human-like selective forgetting mechanisms, which often lead to memory bloat, information overload, or catastrophic forgetting. To this end, we propose FadeMem, a biologically inspired two-tier memory architecture that introduces active forgetting into agent memory systems for the first time. FadeMem employs an adaptive exponential decay strategy modulated by semantic relevance, access frequency, and temporal patterns to enable differentiated memory retention and intelligent integration. Experimental results demonstrate that FadeMem reduces memory storage overhead by 45% across Multi-Session Chat, LoCoMo, and LTI-Bench benchmarks while significantly enhancing multi-hop reasoning and retrieval performance.

Technology Category

• Cognitive Modeling & Cognitive Systems: Agent Architectures
• Machine Learning: Large Multimodal Models (LMMs)
• Multiagent Systems: Modeling other Agents

Application Category

• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions
• User Modeling, Personalization and Recommendation: Large Language Models (LLM) for user modeling and recommendation
• Search and Retrieval-Augmented AI: Agentic search

Abstract

Large language models deployed as autonomous agents face critical memory limitations, lacking selective forgetting mechanisms that lead to either catastrophic forgetting at context boundaries or information overload within them. While human memory naturally balances retention and forgetting through adaptive decay processes, current AI systems employ binary retention strategies that preserve everything or lose it entirely. We propose FadeMem, a biologically-inspired agent memory architecture that incorporates active forgetting mechanisms mirroring human cognitive efficiency. FadeMem implements differential decay rates across a dual-layer memory hierarchy, where retention is governed by adaptive exponential decay functions modulated by semantic relevance, access frequency, and temporal patterns. Through LLM-guided conflict resolution and intelligent memory fusion, our system consolidates related information while allowing irrelevant details to fade. Experiments on Multi-Session Chat, LoCoMo, and LTI-Bench demonstrate superior multi-hop reasoning and retrieval with 45\% storage reduction, validating the effectiveness of biologically-inspired forgetting in agent memory systems.

Problem

• memory limitations
• selective forgetting
• catastrophic forgetting
• information overload
• agent memory

Innovation

• biologically-inspired forgetting
• adaptive memory decay
• dual-layer memory hierarchy
• semantic relevance modulation
• LLM-guided memory fusion