مسابقه‌ی بین‌المللی برنامه‌سازی دانش‌جویی

In this talk, I will explore the foundational concepts of machine learning, including supervised and unsupervised scenarios based on data types, generalization, and learning architectures such as neural networks. We will then dive into the critical role of optimization in training models, followed by an introduction to robust optimization and its variant, distributionally robust optimization (DRO), highlighting their significance in enhancing model reliability. Finally, I will discuss our latest research on leveraging DRO to incorporate unlabeled data, thereby improving generalization and model performance in machine learning applications.