2019 International Conference On Computer Aided Design

The Premier Conference Devoted to Technical Innovations in Electronic Design Automation

November 4-7, 2019The Westin Westminster Westminster, CO

MP Associates, Inc.
WEDNESDAY November 06, 1:00pm - 3:00pm | Westminster III
EVENT TYPE: SPECIAL SESSION
SESSION 9D
Tensor Methods for Accelerated Machine Learning and Electronic Design Automation
Moderator:
Wei Zhang - Hong Kong Univ. of Science and Technology
Organizers:
Zhiru Zhang - Cornell Univ.
Zheng Zhang - Univ. of California, Santa Barbara
As a high-order generalization of matrices and vectors, tensor computation has emerged as a powerful tool to solve high-dimensional problems in scientific computing, machine learning, data mining and optimal control. Despite its superior efficiency in overcoming the curse of dimensionality, the research of tensor computation in hardware accelerators and in EDA is still in the early stage. Recently we note that an active body of research has emerged in both academia and industry, ranging from hardware design to EDA algorithms, making encouraging progress in both theory and engineering applications. This special session will present a number of these research efforts, with topics ranging from new tensor-based EDA algorithms (wafer data analysis, circuit modeling and simulation, parasitic extraction, uncertainty quantification) to tensor hardware accelerators (tensorized DNN on edge devices, architecture for tensor-based DNN). We envision that the discussions in this session will lead to novel cross-layer solutions the enable highly efficient tensor-based machine learning systems and EDA frameworks. The proposed special session is a good match for ICCAD, since curse of dimensionality is a fundamental challenge in many EDA algorithms and machine learning accelerators. According to the initial feedbacks, we feel that this special session will open the door of several new research directions and trigger many follow-up research activities.

9D.1DEEPEYE: A Deeply Tensor Compressed Neural Network hardware accelerator
 Speaker: Hao Yu - Southern Univ. of Science and Technology
 Authors: Yuan Cheng - Shanghai Jiaotong University
Guangya Li - Southern Univ. of Science and Technology
Ngai Wong - Univ. of Hong Kong
Hai-Bao Chen - Shanghai Jiao Tong Univ.
Hao Yu - Southern Univ. of Science and Technology
9D.2High-performance Hardware Architecture for Tensor Singular Value Decomposition
 Speaker: Bo Yuan - Rutgers Univ.
 Authors: Chuanhua Deng - Rutgers Univ.
Xiaoyang Liu - Columbia Univ.
Miao Yin - Rutgers Univ.
Xiaodong Wang - Columbia Univ.
Bo Yuan - Rutgers Univ.
9D.3Tensor Methods for Generating Compact Uncertainty Quantification and Deep Learning Models
 Speaker: Chunfeng Cui - Univ. of California, Santa Barbara
 Authors: Chunfeng Cui - Univ. of California, Santa Barbara
Cole Hawkins - Univ. of California, Santa Barbara
Zheng Zhang - Univ. of California, Santa Barbara
9D.4Tensor Methods for Modeling Nonlinear Circuits and Systems
 Speaker: Ngai Wong - Univ. of Hong Kong
 Author: Ngai Wong - Univ. of Hong Kong
9D.5Facilitating Deployment Of A Wafer-Based Analytic Software Using Tensor Methods
 Speaker: Li-C. Wang - Univ. of California, Santa Barbara
 Authors: Li-C. Wang - Univ. of California, Santa Barbara
Chuanhe (Jay) Shan - Univ. of California, Santa Barbara
Ahmed Wahba - Univ. of California, Santa Barbara
9D.6Tensor Train Accelerated Electromagnetic Modeling of Interconnects
 Speaker: Vladimir Okhmatovski - Univ. of Manitoba
 Authors: Shucheng Zheng - Univ. of Manitoba
Zhuotong Chen - Univ. of California, Santa Barbara
Vladimir Okhmatovski - Univ. of Manitoba