Recent research on fine-tuning vision-language models has demonstrated impressive
performance in various downstream tasks. However, the challenge of obtaining accurately …

Direct alignment algorithms (DAAs), such as direct preference optimization (DPO), have
become popular alternatives for Reinforcement Learning from Human Feedback (RLHF) …

Understanding the geographical context of unstructured textual data is a key challenge in
information extraction. In many applications, however, simple identification of toponyms is …

Deep neural networks suffer from overfitting when training samples contain inaccurate
annotations (noisy labels), leading to suboptimal performance. In addressing this challenge …

Continual learning (CL) seeks to preserve and transfer knowledge across a sequence of
tasks, mitigating catastrophic forgetting of past tasks while guiding learning for new ones …