Close

Session

Research Manuscript: Cost-Efficient Analog and Stochastic Computing Techniques for Deep Learning
Session Chair
Event TypeResearch Manuscript
Keywords
AI/ML Design: Circuits and Architecture
Digital and Analog Circuits
Topics
Design
TimeThursday, July 14th1:30pm - 3:00pm PDT
Location3007, Level 3
DescriptionThis session presents a fundamental challenge: Cost-efficient and practical implementation of deep learning techniques. The solutions offered to this challenge span a broad range, from novel stochastic computation techniques to analog memristive neural network accelerators. The research presented in this session also describes how domain-specific knowledge and cost-efficient methods can help analog and RF circuit parameter optimization and stochastic time to digital conversion.