Learning with Limited Supervision - Generative Modeling Approaches

Explore cutting-edge approaches to machine learning with limited supervision in this seminar presented by Stanford University's Assistant Professor Stefano Ermon. Delve into innovative generative modeling techniques designed to overcome the challenges of scarce or non-existent labeled training data. Discover how humans can learn from minimal examples or high-level instructions, and examine methods for leveraging prior knowledge, such as physics laws or simulators, to provide weak forms of supervision. Gain insights into the emerging field of computational sustainability and learn about Professor Ermon's award-winning research in probabilistic modeling, inference, and optimization. This 1-hour 12-minute seminar offers a valuable opportunity to understand the future direction of machine learning in contexts where traditional large-scale labeled datasets are not available.