Machine Learning Acceleration and Neuromorphic Computing

3 credits · Graduate

The rapidly growing size of neural networks adopted in modern artificial intelligence (AI) applications makes accelerating computations of machine learning algorithms a critical need of the industry. This course will introduce various approaches to design high-efficient neural network models and to include hardware constraints in the efficient neural network designs. We will also discuss the hardware techniques that can accelerate the computations of neural networks on different computing platforms such as GPU, FPGA, and ASIC. Bio-inspired computing and neuromorphic computing will be also discussed. The course is a mix of lectures, labs, & projects. Prerequisite: ECE 250D/COMPSCI 250D, or ECE 552/COMPSCI 550, or permission of instructor.