Xumin Liu

Mixing Expertise with Confidence: A Mixture of Expert Framework for Robust Multi-Modal Continual Learner

Bayesian-Informed Diverse Sampling for Calibration of Fine-Tuned Foundation Models with Evidential Ensembles

Knowledge Exchange with Confidence: Cost-Effective LLM Integration for Reliable and Efficient Visual Question Answering

Parameter-Efficient Multi-Source Domain-Adaptive Prompt Tuning for Open-Vocabulary Object Detection

Multi-Source Knowledge-Fusion for Source-Free Domain Adaptation in Object Detection

Learning to Compose: Continual Visual QA through a Dual-Purpose Mixture-of-Experts Framework

Diversity–Ambiguity Exploration for Weakly Supervised Video Anomaly Detection

Vulnerability-Aware Parameter-Efficient Fine-Tuning for Enhanced Adversarial Robustness

InsCal: Calibrated Multi-Source Fully Test-Time Prompt Tuning for Object Detection

Avoid Being a Shortcut Learner through Library-Based Re-Learning