Apr 3 (Wed) @ 5:00pm: "High-Performance Deep Learning Systems via DL Sparsity and DL Compiler," Guyue Huang, ECE PhD Defense

https://ucsb.zoom.us/j/5124483174

Abstract

Deep learning (DL) has achieved impressive results in many tasks across recognition, decision, and content generation. The widespread adoption of DL requires efficient systems that can support large-scale training and reduce the cost of inference. As rigid hardware scaling starts to show diminishing returns for DL system performance improvement, the computation efficiency and algorithm efficiency becomes critical. This defense investigates two key technologies to advance high-performance DL systems, named DL compiler and DL sparsity. I will present compiler designs, model pruning techniques, GPU architecture, and network architecture that enhance the performance and energy efficiency of DL systems significantly. I will close the talk by summarizing my methodology and discussing about future directions.

Bio

Guyue Huang is a Ph.D. candidate at the Department of Electrical and Computer Engineering, UC Santa Barbara. Her research interest is high-performance deep learning systems. She received her B.E. degree from Tsinghua University, Beijing, China in 2020.

Hosted by: Professor Zheng Zhang

Submitted by: Guyue Huang <guyue@ucsb.edu>