Session

· Contributors · Organizations · Search Program · My Agenda

Research Manuscript: Boosting Neural Networks through Innovative Approximation and Compression

Session Chairs

Ganapati Bhat

Washington State University

Xiaoxiao Liu

Event TypeResearch Manuscript

Topics

Keywords

TimeWednesday, July 12th3:30pm - 5:30pm PDT

Location3010, 3rd Floor

DescriptionDeep learning techniques have enjoyed ever-increasing accuracy and capabilities thanks to their massive tunable scaling parameters. However, this progress comes at the expense of computation, energy, and carbon footprint requirements. This session provides answers to a fundamental question. Do we have to perform every single operation to achieve the target performance? The papers will show innovative approaches to eliminate and approximate repetitive operations for dramatic efficiency gains with a minimal performance impact.

Presentations

3:30pm - 3:40pm PDT	Lightning Talk: How to Boost Deep Neural Networks for Computer Vision Author Soonhoi Ha
3:40pm - 3:55pm PDT	Flex-SFU: Accelerating DNN Activation Functions by Non-Uniform Piecewise Approximation Authors Enrico Reggiani Renzo Andri Lukas Cavigelli
3:55pm - 4:10pm PDT	A Memory-Efficient Edge Inference Accelerator with XOR-based Model Compression Authors Hyunseung Lee Jihoon Hong Soosung Kim Seung Yul Lee Jae W. Lee
4:10pm - 4:25pm PDT	Dynamic Sparse Training via Balancing the Exploration-Exploitation Trade-off Authors Shaoyi Huang Bowen Lei Dongkuan Xu Hongwu Peng Yue Sun Mimi Xie Caiwen Ding
4:25pm - 4:40pm PDT	Algorithms and Hardware for Efficient Processing of Logic-based Neural Networks Authors Jingkai Hong Arash Fayyazi Amirhossein Esmaili Mahdi Nazemi Massoud Pedram
4:40pm - 4:55pm PDT	HBP: Hierarchically Balanced Pruning and Accelerator Co-Design for Efficient DNN Inference Authors Ao Ren Yuhao Wang Tao Zhang Jiaxing Shi Duo Liu Xianzhang Chen Yujuan Tan Yuan Xie
4:55pm - 5:10pm PDT	Efficient Non-Linear Adder for Stochastic Computing with Approximate Spatial-Temporal Sorting Network Authors Yixuan Hu Tengyu Zhang Meng Li Renjie Wei Liangzhen Lai Yuan Wang Runsheng Wang Ru Huang