Systems for Machine Learning

1 Course Unit

The course covers advanced topics in machine learning systems with an emphasis on the interplay between models and system-level support. This course surveys recent advances in efficient machine learning computing techniques including model compression, pruning, quantization, neural architecture search, knowledge distillation, distributed training, and parallelism. Discussion-oriented classes focus on in-depth analysis of readings. Final project and paper required. Appropriate for graduate and advanced undergraduate students. After completing this course, students should be able to: 1) understand general research and development trends in machine learning systems; 2) develop intuition on how to optimize algorithms with hardware in mind; 3) read machine learning system papers critically; 4) write constructive paper reviews; 5) design and execute a research project to address an open research problem in machine learning systems; 6) develop self- learning skills for continuous growth beyond the course. Prerequisites: Students must take (CIS 4190 or CIS 4480 or CIS 4710 or CIS 5190 or CIS 5200 or CIS 5480 or CIS 5710 or ESE 3050 or ESE 5320) before taking ESE 6390. 1 Course Unit