Deep Model Compression and Inference Speedup of Sum-Product Networks on Tensor Trains

Journal article (2020)

Authors

Ching Yun Ko The University of Hong Kong
Cong Chen The University of Hong Kong
Zhuolun He The University of Hong Kong
Yuke Zhang University of South Carolina
K. Batselier Team Jan-Willem van Wingerden - Mechanical, Maritime and Materials Engineering
Ngai Wong The University of Hong Kong
Research Group
Team Jan-Willem van Wingerden (Mechanical, Maritime and Materials Engineering) (TU Delft)
Copyright: © 2020 Ching Yun Ko, Cong Chen, Zhuolun He, Yuke Zhang, K. Batselier, Ngai Wong
DOI: https://doi.org/10.1109/TNNLS.2019.2928379
More Info
expand_more

Published Date

2020

Language

English

Reuse Rights

Other than for strictly personal use, it is not permitted to download, forward or distribute the text or part of it, without the consent of the author(s) and/or copyright holder(s), unless the work is under an open content license such as Creative Commons.

Faculty

Mechanical, Maritime and Materials Engineering

Department

Delft Center for Systems and Control

Research Group

Team Jan-Willem van Wingerden

Abstract

Sum-product networks (SPNs) constitute an emerging class of neural networks with clear probabilistic semantics and superior inference speed over other graphical models. This brief reveals an important connection between SPNs and tensor trains (TTs), leading to a new canonical form which we call tensor SPNs (tSPNs). Specifically, we demonstrate the intimate relationship between a valid SPN and a TT. For the first time, through mapping an SPN onto a tSPN and employing specially customized optimization techniques, we demonstrate improvements up to a factor of 100 on both model compression and inference speedup for various data sets with negligible loss in accuracy.

Files

08793233.pdf
(.pdf | 0.848 Mb)
- Embargo expired in 09-02-2020