I am a Ph.D. student in the Computer Science and Engineering (CSE) department at the University of Michigan, Ann Arbor. I joined the Compilers Creating Custom Processors (CCCP) research group in Fall of 2014 and started working under supervision of Prof. Scott Mahlke. I received my Bachelor's degree in Electrical Engineering from the Sharif University of Technology in 2014.
My research interests lie in neural networks, neuromorphic accelerators and approximate computing. Neural networks are compute-intensive, which makes them power hungry and slow. I am investigating how we can improve the performance and energy efficiency of these networks without decreasing their accuracy.
The main goal of approximate computing is trading a small amount of accuracy or quality of results for a considerable improved performance and reduced energy consumption. My projects in this area could be categorized to two different categories. First, using neural networks or other approximation techniques to approximate real applications and bound their error. Second, taking advantages of various approximation methods to improve the performance and efficiency of neural networks.