Hongxin Wei

Assistant Professor@Southern University of Science and Technology·中国·OpenReview

研究方向

Uncertainty Quantification · Conformal prediction · Model calibration · Data privacy · Data optimization · Weakly supervised learning · noisy labels · Open-world machine learning · OOD detection · Reliable deep learning

Hongxin Wei

Detecting Distillation Data from Reasoning Models

Multi-Condition Conformal Selection

PAC Reasoning: Controlling the Performance Loss for Efficient Reasoning

Selective Labeling with False Discovery Rate Control

High-Power Training Data Identification with Provable Statistical Guarantees

Conformal Regression under Distribution Shift: A Reinforcement Learning Method for Adaptive Uncertainty Quantification

Unlocking the Pre-Trained Model as a Dual-Alignment Calibrator for Post-Trained LLMs

DT-Pro: Proactive Decision Transformers with Implicit Latent Space Planning

Exploring Imbalanced Annotations for Effective In-Context Learning

Parametric Scaling Law of Tuning Bias in Conformal Prediction

C-Adapter: Adapting Deep Classifiers for Efficient Conformal Prediction Sets

Your Pre-trained LLM is Secretly an Unsupervised Confidence Calibrator

Exploring Learning Complexity for Efficient Downstream Dataset Pruning

How Contaminated Is Your Benchmark? Measuring Dataset Leakage in Large Language Models with Kernel Divergence

Exploring the Noise Robustness of Online Conformal Prediction

Fine-tuning can Help Detect Pretraining Data from Large Language Models

Local-Prompt: Extensible Local Prompts for Few-Shot Out-of-Distribution Detection

Automatic Dataset Construction (ADC): Sample Collection, Data Curation, and Beyond

Understanding and Mitigating Miscalibration in Prompt Tuning for Vision-Language Models

Robust Online Conformal Prediction under Uniform Label Noise

Understanding and Mitigating Miscalibration in Prompt Tuning for Vision-Language Models

Analytic Continual Test-Time Adaptation for Multi-Modality Corruption

Delving into Temperature Scaling for Adaptive Conformal Prediction

Defending Membership Inference Attacks via Privacy-aware Sparsity Tuning

Conformal Prediction for Deep Classifier via Truncating

Improving Ordinal Conformal Prediction by Stepwise Adaptive Posterior Alignment