Aidong Zhang

Toward Faithful Retrieval-Augmented Generation with Sparse Autoencoders

Chart-RVR: Reinforcement Learning with Verifiable Rewards for Explainable Chart Reasoning

Rectifying Shortcut Behaviors in Preference-based Reward Learning

NeuronTune: Towards Self-Guided Spurious Bias Mitigation

What is the Visual Cognition Gap between Humans and Multimodal LLMs?

Benchmarking Visual Cognition of Multimodal LLMs via Matrix Reasoning

Review and Rebuttal: Zero-shot In-context Adversarial Learning for Improving Research Ideation

Mitigating Spurious Bias with Last-Layer Selective Activation Retraining

MixNAM: Advancing Neural Additive Models with Mixture of Experts

Learn from Known Unknowns: A Unified Empirical Bayesian Framework for Improving Group Robustness