MemQ: Integrating Q-Learning into Self-Evolving Memory Agents over Provenance DAGs

2026-05-08

AI Summary

Existing episodic memory approaches for large language model agents evaluate memory retrieval quality in isolation, neglecting the influence of inter-memory dependency chains on subsequent generation. This work proposes a novel credit assignment mechanism that integrates the TD(λ) eligibility trace from reinforcement learning with a provenance-aware directed acyclic graph (DAG) of memories, introducing structure-based credit decay formalized within an exogenous-context Markov decision process (EC-MDP). By replacing temporal distance with structural proximity for credit allocation, the method achieves state-of-the-art success rates across six benchmark tasks, with particularly pronounced gains in multi-step tasks—up to a 5.7 percentage point improvement—demonstrating its effectiveness in modeling complex memory dependencies.

Abstract

Episodic memory allows LLM agents to accumulate and retrieve experience, but current methods treat each memory independently, i.e., evaluating retrieval quality in isolation without accounting for the dependency chains through which memories enable the creation of future memories. We introduce MemQ, which applies TD($λ$) eligibility traces to memory Q-values, propagating credit backward through a provenance DAG that records which memories were retrieved when each new memory was created. Credit weight decays as $(γλ)^d$ with DAG depth $d$, replacing temporal distance with structural proximity. We formalize the setting as an Exogenous-Context MDP, whose factored transition decouples the exogenous task stream from the endogenous memory store. Across six benchmarks, spanning OS interaction, function calling, code generation, multimodal reasoning, embodied reasoning, and expert-level QA, MemQ achieves the highest success rate on all six in generalization evaluation and runtime learning, with gains largest on multi-step tasks that produce deep and relevant provenance chains (up to +5.7~pp) and smallest on single-step classification (+0.77~pp) where single-step updates already suffice. We further study how $γ$ and $λ$ interact with the EC-MDP structure, providing principled guidance for parameter selection and future research. Code will be available soon.

Problem

• episodic memory
• memory dependency
• provenance DAG
• LLM agents
• credit assignment

Innovation

• Q-learning
• provenance DAG
• eligibility traces
• episodic memory
• Exogenous-Context MDP