Special Topics in Information Sciences and Systems: Theory of Deep Weakly Supervised Learning

Graduate level course focusing on theoretical foundations of deep learning with a focus on weakly supervised aspects. The class covers the theoretical analysis tools used in deep learning including stochastic gradient, uniform convergence theory and statistical learning. The course focuses primarily on unsupervised learning and weakly supervised learning. Topics include generative models, self-training, transfer learning, representation learning, semi-supervised learning and other forms of self-supervised learning. Prior knowledge on statistics, linear algebra, probability theory and optimization is required.