Recently, despite the unprecedented success of large pre-trained visual-language models
(VLMs) on a wide range of downstream tasks, the real-world unsupervised domain …

Graph Neural Networks (GNNs) show promising results for graph tasks. However, existing
GNNs' generalization ability will degrade when there exist distribution shifts between testing …

Out-of-distribution (OOD) problems in few-shot classification (FSC) occur when novel
classes sampled from testing distributions differ from base classes drawn from training …

In database of recommender systems, users' ratings for most items are usually missing,
resulting in selection bias when users selectively choose items to rate. To address this …

Catastrophic forgetting emerges as a critical challenge when fine-tuning multi-modal large
language models (MLLMs), where improving performance on unseen tasks often leads to a …

Large-scale vision-language (VL) models have demonstrated remarkable generalization
capabilities for downstream tasks through prompt tuning. However, the mechanisms behind …

Due to privacy or patent concerns, a growing number of large models are released without
granting access to their training data, making transferring their knowledge inefficient and …

Graph Neural Networks (GNNs) show promising results for graph tasks. However, existing
GNNs' generalization ability will degrade when there exist distribution shifts between testing …

Gradient-based meta-learning (GBML) algorithms can quickly adapt to new tasks by
transferring the learned meta-knowledge while assuming that all tasks come from the same …

Recent advances achieved by deep learning models rely on the independent and identically
distributed assumption, hindering their applications in real-world scenarios with domain …