Fundamental performance limits (and algorithms) for dictionary learning (e.g., matrix factorization), ranking, and deep learning architectures;
Learning in the presence of privacy constraints;
Learning in the large alphabet regime;
Learning of graphical models and other statistical models.
Working in traditional topics in Shannon's information theory of interest to the PI will also be highly encouraged. Some sample topics include:
Multi-user information theory;
Strong converse and second-order asymptotics;
Error exponent analysis and the method of types;
Each position is expected to last from one to three years. The earliest start date is in March 2018. The candidate is expected to have a PhD in electrical engineering, computer science or applied mathematics and a strong publication record in information theory, signal processing, or machine learning. The candidate will work closely with Dr. Vincent Tan and will use the postdoctoral stint to develop a strong research profile that will enable him/her to find a good faculty position after the postdoctoral stint.
For more information about Vincent Tan's research interests, please visit his homepage at;