ReDiS: Reasoning-Diagnosis Self-Evolution for Multimodal Reasoning in Remote Sensing

Key points

• We introduce Reasoning--Diagnosis Self-Evolution (ReDiS), a framework that separates reasoning exploration from candidate evaluation while allowing the two processes to evolve jointly.1
• ReDiS instantiates two role-specific experts from a shared multimodal backbone: a Reasoner that explores diverse visual evidence--answer candidates and a Diagnoser that jointly evaluates the complete candidate set and provides credibility-aware feedback.1
• To handle groups in which all sampled answers are incorrect and group-relative advantages vanish, we further introduce Failure-Triggered Recovery (FTR), which provides minimal outcome-level correction without prescribing intermediate reasoning trajectories.1
• Experiments on LHRS-Bench, MME-RS, and CHOICE show consistent improvements over Qwen3.5 backbones at 0.8B, 2B, and 4B scales.1

Sources (1)

• [1]ReDiS: Reasoning-Diagnosis Self-Evolution for Multimodal Reasoning in Remote Sensing
  arXiv (AI, ML, NLP, CV, robotics, multi-agent) · Oct 13, 01:38 PM

  Multimodal large language models (MLLMs) have shown strong potential for remote sensing understanding, yet reliable reasoning remains challenging when tasks require fine-grained visual grounding and multi-step inference.

  We introduce Reasoning--Diagnosis Self-Evolution (ReDiS), a framework that separates reasoning exploration from candidate evaluation while allowing the two processes to evolve jointly.

Extractive summary: sentences quoted from the sources.