Adversarial Training and Reinforcement Learning for Robust Reasoning in Large Language Models. JDIAS 2026, 1 (3).