Study of advanced topics of current interest in the field of deep learning, with an emphasis on understanding the network architecture of the pre-trained deep learning models. Selected topics from the following areas will be covered, with an emphasis on practical applications: computer vision, speech recognition, natural language processing, reinforcement learning, and deep learning tools.

This course introduces statical machine learning methods from a theoretical perspective. Topics include the maximum likelihood inference, regularization principle, risk minimization framework, cross-validation, high dimensional inference and nonparametric methods.

By completing this capstone project the students will get an opportunity to apply the knowledge and skills in data science to some real industrial projects. In particular, it will test their skills in data acquisition, data visualization, data wrangling, data organization, data modeling and data analysis. Unlike the other courses in the Data Science series, in this course, the students are expected to explore some projects that are more real and open-ended.