MACHINE LEARNING AND INTELLIGENCE LAB

We have four papers "PhaseMI: A Motivational Interviewing Dataset for Enhancing Phase Progression in LLM-based Counseling", "MM-JudgeBias: A Benchmark for Evaluating Compositional Biases in MLLM-as-a-Judge", "CURaTE: Continual Unlearning in Real Time with Ensured Preservation of LLM Knowledge", and "Bringing Real-World Relations into Video Generation with Graph-Structured Knowledge" accepted to ACL 2026.

We have six papers “Secure Inference for Diffusion Models via Unconditional Scores”, “When to Ensemble: Identifying Token-Level Points for Stable and Fast LLM Ensembling”, “No Prompt Left Behind: Exploiting Zero-Variance Prompts in LLM Reinforcement Learning via Entropy-Guided Advantage Shaping”, “Delta-XAI: A Unified Framework for Explaining Prediction Changes in Online Time Series Monitoring”, “Soft Equivariance Regularization for Invariant Self-Supervised Learning”, and “Faster Vision Transformers with Adaptive Patches” accepted to ICLR 2026.

We have one paper "Unveiling the Response of Large Vision-Language Models to Visually Absent Tokens" accepted to EMNLP 2025 main.

We have one paper "VQ-FedDiff: Federated Learning Algorithm of Diffusion Models with Client-Specific Vector-Quantized Conditioning" accepted to TPAMI 2025.

We have one paper "Early Timestep Zero-Shot Candidate Selection for Instruction-Guided Image Editing" accepted to ICCV 2025.