Joint Optimization of Reasoning and Dual-Memory for Self-Learning Diagnostic Agent

Bingxuan Li, Simo Du, Yue Guo, University of Illinois Urbana-Champaign, Jacobi Medical Center [email protected], [email protected] · 2026

Read Joint Optimization of Reasoning and Dual-Memory for Self-Learning Diagnostic Agent by Bingxuan Li, Simo Du, Yue Guo, University of Illinois Urbana-Champaign, Jacobi Medical Center [email protected], [email protected] (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2604-07269.txt.

Description

Li, Du, and Guo build SEA, a diagnostic agent that learns from experience instead of treating each case in isolation. Its dual-memory module stores reusable diagnostic patterns, and a reinforcement-training framework jointly tunes reasoning and memory management. On a standard benchmark, SEA reaches 92.46 percent accuracy, beating the strongest baseline by 19.6 points. On a long-horizon set, it also finishes best, and expert review finds its consolidated rules clinically sound, useful, and trustworthy.

Catalog record

CREATORBingxuan Li, Simo Du, Yue Guo, University of Illinois Urbana-Champaign, Jacobi Medical Center [email protected], [email protected]
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE59470
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2604.07269

Browse the full library · About the Tower of Babel