I am an assistant professor at the Computer Science Department at the University of Iowa since August 2014. I was a researcher at NEC Laboratories America, Inc. Before that, I was a Machine Learning Researcher at GE Global Research. I received my Ph.D. degree in Computer Science from Michigan State University in 2012. Here is my Google Scholar Citations and CV.

• News
• Research
• Publications
• Software
• Services
• Teaching

I am interested in machine learning and optimization and its applications to big data analytics.
My current research topics include:

• Non-Convex Optimization Algorithms
  To develope provable and practical optimization algorithms for solving non-convex problems in machine learning
  [Unified Momentum] , [NEON] , [Adaptive NCD], [Stagewise Optimization for DL] , [Stagewise Katyusha] , [Non-Convex Concave Min-Max] , [Non-Convex Non-Concave Min-Max]


• Faster Convergent Algorithms by Leveraging Error Bound Condition
  To develope faster convergent optimization algorithms by leveraging local and global error bound conditions
  [RSG] , [HOPS] , [ASSG], [Adaptive Admm] , [adaSVRG] , [adaAGC] , [SadaGrad] , [LogProj] , [ada-Frank-Wolfe]


• Large-scale Stochastic Optimization
  To develope efficient algorithms for stochastic optimization for learning from large-scale high-dimensional data.


• Online Optimization
  To develope new online convex optimization algorithms.


• Deep Learning
  To understand, design and optimize deep neural networks for large-scale image classification.
  See demo on flower, food and vehicle recognition, collaborating with NEC Labs America.


• Randomized Algorithms for Big Data Analytics
  Developing randomized algorithms for solving big data machine learning problems. See here.


• Distributed Optimization for Big Data
  Developed a practical distributed optimization for solving big data classification and regression problems. See Software.

Some old research topics during PhD studies include:

• Social Network Analysis
  Developed algorithms for detecting communities and their dynamic evolutions in Social Networks.


• Learning from Noisy Labels
  Developed algorithms and theories for learning from Noisy labels.


• Multiple Kernel Learning
  Developed efficient and robust algorithms for multiple kernel learning.


• Improved Bounds for the Nystrom method
  Developed several improved bounds for the Nystrom method under larger eigengap condition and power law eigenvalue distribution; and the theory of their applications to large-scale kernel learning.