The PhD/Doctoral Seminar in Machine Learning Theory is designed to provide advanced students with a rigorous foundation in the theoretical principles underlying modern machine learning. The seminar emphasizes critical engagement with cutting-edge research; fostering a deep understanding of mathematical frameworks; statistical guarantees; and computational limits in learning systems. Participants will explore topics such as generalization bounds; optimization theory; statistical learning theory; and the interplay between algorithms and data distributions. A key educational benefit of this seminar is the development of analytical skills necessary to assess and contribute to theoretical advances in machine learning. Students will learn to rigorously evaluate proofs; identify open problems; and articulate novel research directions. The seminar encourages active participation through presentations; peer feedback; and collaborative discussions; ensuring students refine their ability to communicate complex ideas clearly and effectively. The seminar also bridges theory and practice by examining how theoretical insights inform real-world applications. Students will analyze case studies where theoretical breakthroughs have led to practical improvements in model performance; robustness; or efficiency. This connection helps participants appreciate the broader impact of their research and prepares them to address challenges at the intersection of theory and implementation. By the end of the seminar; students will have strengthened their ability to conduct independent theoretical research; critically engage with literature; and contribute original ideas to the field. The seminar is ideal for those pursuing academic careers or research-oriented roles in industry; as it cultivates the intellectual rigor and creativity required to advance the frontiers of machine learning theory.
