Quantizing Signals for Linear Classification. Ezzeldin, Y. H, Fragouli, C., & Diggavi, S. In 2019 IEEE International Symposium on Information Theory (ISIT), pages 912–916, 2019. IEEE. doi abstract bibtex 1 download In many machine learning applications, once we have learned a classifier, in order to apply it, we may still need to gather features from distributed sensors over communication constrained channels. In this paper, we propose a polynomial complexity algorithm for feature quantization tailored to minimizing the classification error of a linear classifier. Our scheme produces scalar quantizers that are well-tailored to delay-sensitive applications, operates on the same training data used to learn the classifier, and allows each distributed sensor to operate independently of each other. Numerical evaluation indicates up to 65% benefits over alternative approaches. Additionally, we provide an example where, jointly designing the linear classifier and the quantization scheme, can outperform sequential designs.
@inproceedings{ezzeldin2019quantizing,
abstract = {In many machine learning applications, once we have learned a classifier, in order to apply it, we may still need to gather features from distributed sensors over communication constrained channels. In this paper, we propose a polynomial complexity algorithm for feature quantization tailored to minimizing the classification error of a linear classifier. Our scheme produces scalar quantizers that are well-tailored to delay-sensitive applications, operates on the same training data used to learn the classifier, and allows each distributed sensor to operate independently of each other. Numerical evaluation indicates up to 65% benefits over alternative approaches. Additionally, we provide an example where, jointly designing the linear classifier and the quantization scheme, can outperform sequential designs.},
author = {Ezzeldin, Yahya H and Fragouli, Christina and Diggavi, Suhas},
booktitle = {2019 IEEE International Symposium on Information Theory (ISIT)},
organization = {IEEE},
pages = {912--916},
tags = {conf,DML,CEDL},
title = {Quantizing Signals for Linear Classification},
type = {4},
doi = {10.1109/ISIT.2019.8849589},
year = {2019}
}
Downloads: 1
{"_id":"zyMabdBnxYyCCoJ5i","bibbaseid":"ezzeldin-fragouli-diggavi-quantizingsignalsforlinearclassification-2019","author_short":["Ezzeldin, Y. H","Fragouli, C.","Diggavi, S."],"bibdata":{"bibtype":"inproceedings","type":"4","abstract":"In many machine learning applications, once we have learned a classifier, in order to apply it, we may still need to gather features from distributed sensors over communication constrained channels. In this paper, we propose a polynomial complexity algorithm for feature quantization tailored to minimizing the classification error of a linear classifier. Our scheme produces scalar quantizers that are well-tailored to delay-sensitive applications, operates on the same training data used to learn the classifier, and allows each distributed sensor to operate independently of each other. Numerical evaluation indicates up to 65% benefits over alternative approaches. Additionally, we provide an example where, jointly designing the linear classifier and the quantization scheme, can outperform sequential designs.","author":[{"propositions":[],"lastnames":["Ezzeldin"],"firstnames":["Yahya","H"],"suffixes":[]},{"propositions":[],"lastnames":["Fragouli"],"firstnames":["Christina"],"suffixes":[]},{"propositions":[],"lastnames":["Diggavi"],"firstnames":["Suhas"],"suffixes":[]}],"booktitle":"2019 IEEE International Symposium on Information Theory (ISIT)","organization":"IEEE","pages":"912–916","tags":"conf,DML,CEDL","title":"Quantizing Signals for Linear Classification","doi":"10.1109/ISIT.2019.8849589","year":"2019","bibtex":"@inproceedings{ezzeldin2019quantizing,\n abstract = {In many machine learning applications, once we have learned a classifier, in order to apply it, we may still need to gather features from distributed sensors over communication constrained channels. In this paper, we propose a polynomial complexity algorithm for feature quantization tailored to minimizing the classification error of a linear classifier. Our scheme produces scalar quantizers that are well-tailored to delay-sensitive applications, operates on the same training data used to learn the classifier, and allows each distributed sensor to operate independently of each other. Numerical evaluation indicates up to 65% benefits over alternative approaches. Additionally, we provide an example where, jointly designing the linear classifier and the quantization scheme, can outperform sequential designs.},\n author = {Ezzeldin, Yahya H and Fragouli, Christina and Diggavi, Suhas},\n booktitle = {2019 IEEE International Symposium on Information Theory (ISIT)},\n organization = {IEEE},\n pages = {912--916},\n tags = {conf,DML,CEDL},\n title = {Quantizing Signals for Linear Classification},\n type = {4},\n doi = {10.1109/ISIT.2019.8849589},\n year = {2019}\n}\n\n","author_short":["Ezzeldin, Y. H","Fragouli, C.","Diggavi, S."],"key":"ezzeldin2019quantizing","id":"ezzeldin2019quantizing","bibbaseid":"ezzeldin-fragouli-diggavi-quantizingsignalsforlinearclassification-2019","role":"author","urls":{},"metadata":{"authorlinks":{}},"downloads":1},"bibtype":"inproceedings","biburl":"https://research.seas.ucla.edu/licos/files/2023/02/publicactions_v1.bib","dataSources":["ZyBHKCcCyS7Qk9n3w","hicKnsKYNEFXC4CgH","jxCYzXXYRqw2fiEXQ","wCByFFrQMyRwfzrJ6","yuqM5ah4HMsTyDrMa","YaM87hGQiepg5qijZ","n9wmfkt5w8CPqCepg","soj2cS6PgG8NPmWGr","FaDBDiyFAJY5pL28h","ycfdiwWPzC2rE6H77","2BHqTGHtDg7BjRAJQ","XpqzCLvCgsYBqbTpK","QNNZyNQzfeETJyTbL","Z6bBhKezwKd3pPWRc"],"keywords":[],"search_terms":["quantizing","signals","linear","classification","ezzeldin","fragouli","diggavi"],"title":"Quantizing Signals for Linear Classification","year":2019,"downloads":1}