Zhe Li and Talal Riaz, 4th year PhD students advised by Professors Tianbao Yang and Sriram Pemmaraju, respectively, have been named a recipient of the Graduate College Post-Comprehensive Research Award for the spring 2017 semester. In addition, Ananda Guneratne, 1st year PhD student advised by Professor Aaron Stump, was appointed as a University of Iowa Recruitment Fellow beginning in the spring 2017.
Zhe's research work with Professor Yang lies at the intersection between optimization and machine learning. In machine learning, the ultimate goal is to learn a mapping function with the lower generalization risk. However, most traditional methods solve this problem by Empirical Risk Minimization (ERM) or Regularized Empirical Risk Minimization (RERM), which does not necessarily lead to the lowest generalization risk. Our research work is aiming to directly minimize the generalization error via the proposed sampling strategy, which could improve the performance of the learned model and meanwhile lead a faster convergence for both shallow learning and deep learning.
Talal’s research focuses on Distributed Algorithms, the study of algorithms executed in a distributed fashion over many processors (e.g., computers, mobile devices, wireless sensors etc.) that can communicate with each other. He has focused on the distributed Maximal Independent Set (MIS) problem. MIS is a fundamental problem in Distributed Algorithms sometimes referred to as its “crown jewel”. It has applications in map labeling, molecular biology, routing, scheduling, and many other real world systems. Recently, he has been working on problems in distributed setting where congestion is the key bottleneck. These problems cannot be solved quickly because the volume of information needed cannot reach any processor fast enough. In this setting he is exploring the impact on the quality of the produced solution when very few rounds of communication are allowed.
The Graduate College Post-Comprehensive Research Award program provides an opportunity for advanced doctoral students to benefit from protected and supported time to pursue their scholarly research activities. The award is intended to recognize students with distinguished academic achievement during their early graduate training. These achievements should be evident from a combination of outstanding academic performance in coursework, as well as early scholarly research activities. Students who have held teaching assistantships in the previous two semesters will have priority.