2026-09-20T06:38:46Zhttps://dspace.mit.edu/server/oai/request

oai:dspace.mit.edu:1721.1/1180342022-01-13T07:54:01Zcom_1721.1_7582com_1721.1_7581col_1721.1_131023

Vivienne Sze. Yang, Tien-Ju Massachusetts Institute of Technology. Department of Electrical Engineering and Computer Science. Massachusetts Institute of Technology. Department of Electrical Engineering and Computer Science 2018-09-17T15:54:41Z 2018-09-17T15:54:41Z 2018 2018 http://hdl.handle.net/1721.1/118034 1051458923 Thesis: S.M., Massachusetts Institute of Technology, Department of Electrical Engineering and Computer Science, 2018. Cataloged from PDF version of thesis. Includes bibliographical references (pages 49-57). Neural networks are indispensable to state-of-the-art artificial intelligence algorithms. However, its high accuracy comes at the cost of high computational complexity. This leads to the high operating cost of data centers and also hinders its deployment on mobile devices. In this thesis, we propose an algorithm to address this problem. The proposed algorithm uses progressive barriers to automatically and progressively simplify a pre-trained neural network until the target complexity is met while maximizing the accuracy. Along with the neural network that meets the target complexity, the algorithm also generates a family of simplified networks with different accuracy-complexity trade-offs, which allows for dynamic network selection and further study. Experiment results show that the algorithm achieves better accuracy-complexity trade-offs on a highly compact MobileNet architecture, compared with state-of-the-art automated network simplification approaches. For image classification on the ImageNet dataset, the algorithm reduces the number of multiply-accumulate operations by 1.68x while achieving 0.9% higher accuracy. by Tien-Ju Yang. S.M. 57 pages eng Massachusetts Institute of Technology MIT theses are protected by copyright. They may be viewed, downloaded, or printed from this source but further reproduction or distribution in any format is prohibited without written permission. http://dspace.mit.edu/handle/1721.1/7582 Electrical Engineering and Computer Science. Neural network simplification using a progressive barrier based approach Thesis application/pdf false Publication unknown unknown <Publication xmlns="https://www.openaire.eu/cerif-profile/1.1/" id="3db881cc-3d4a-4adc-a8ec-ef563db12bff"> <Type xmlns="https://www.openaire.eu/cerif-profile/vocab/COAR_Publication_Types">http://purl.org/coar/resource_type/c_1843</Type> <Language>eng</Language> <Title>Neural network simplification using a progressive barrier based approach</Title> <PublishedIn> <Publication> </Publication> </PublishedIn> <PublicationDate>2018</PublicationDate> <Authors> <Author> <DisplayName>Yang, Tien-Ju</DisplayName> <Affiliation> <OrgUnit> </OrgUnit> </Affiliation> </Author> </Authors> <Editors> </Editors> <Publishers> <Publisher> <DisplayName>Massachusetts Institute of Technology</DisplayName> <OrgUnit /> </Publisher> </Publishers> <License>http://dspace.mit.edu/handle/1721.1/7582</License> <Keyword>Electrical Engineering and Computer Science.</Keyword> <Abstract>Neural networks are indispensable to state-of-the-art artificial intelligence algorithms. However, its high accuracy comes at the cost of high computational complexity. This leads to the high operating cost of data centers and also hinders its deployment on mobile devices. In this thesis, we propose an algorithm to address this problem. The proposed algorithm uses progressive barriers to automatically and progressively simplify a pre-trained neural network until the target complexity is met while maximizing the accuracy. Along with the neural network that meets the target complexity, the algorithm also generates a family of simplified networks with different accuracy-complexity trade-offs, which allows for dynamic network selection and further study. Experiment results show that the algorithm achieves better accuracy-complexity trade-offs on a highly compact MobileNet architecture, compared with state-of-the-art automated network simplification approaches. For image classification on the ImageNet dataset, the algorithm reduces the number of multiply-accumulate operations by 1.68x while achieving 0.9% higher accuracy.</Abstract> <Access xmlns="http://purl.org/coar/access_right" > </Access> </Publication>