Designing enhanced classifiers using prior process knowledge: Regularized maximum-likelihood

Mohammad Shahrokh Esfahani, Amin Zollanvari, Byung Jun Yoon, Edward R. Dougherty

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Abstract

We propose a novel optimization-based paradigm for designing enhanced classifiers. The proposed paradigm allows us to incorporate available prior process knowledge into classifier design, thereby improving the performance of the resulting classifiers. In this work, we focus on dynamical systems that can be represented as finite-state multi-dimensional stochastic processes that possess labeled steady-state distributions. Given prior operational knowledge of the process, our goal is to build a classifier that can accurately label future observations obtained from the steady-state, by utilizing both the available prior knowledge and the training data. Simulation results show that the proposed paradigm yields improved classifiers that outperform traditional classifiers that use only training data.

Original languageEnglish
Title of host publicationProceedings - IEEE International Workshop on Genomic Signal Processing and Statistics
Pages91-94
Number of pages4
Publication statusPublished - 2011
Externally publishedYes
Event2011 IEEE International Workshop on Genomic Signal Processing and Statistics, GENSIPS'11 - San Antonio, TX, United States
Duration: Dec 4 2011Dec 6 2011

Other

Other2011 IEEE International Workshop on Genomic Signal Processing and Statistics, GENSIPS'11
CountryUnited States
CitySan Antonio, TX
Period12/4/1112/6/11

Fingerprint

Stochastic Processes
Maximum likelihood
Classifiers
Random processes
Labels
Dynamical systems

ASJC Scopus subject areas

  • Biochemistry, Genetics and Molecular Biology (miscellaneous)
  • Computational Theory and Mathematics
  • Signal Processing
  • Biomedical Engineering

Cite this

Esfahani, M. S., Zollanvari, A., Yoon, B. J., & Dougherty, E. R. (2011). Designing enhanced classifiers using prior process knowledge: Regularized maximum-likelihood. In Proceedings - IEEE International Workshop on Genomic Signal Processing and Statistics (pp. 91-94). [6169451]

Designing enhanced classifiers using prior process knowledge : Regularized maximum-likelihood. / Esfahani, Mohammad Shahrokh; Zollanvari, Amin; Yoon, Byung Jun; Dougherty, Edward R.

Proceedings - IEEE International Workshop on Genomic Signal Processing and Statistics. 2011. p. 91-94 6169451.

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Esfahani, MS, Zollanvari, A, Yoon, BJ & Dougherty, ER 2011, Designing enhanced classifiers using prior process knowledge: Regularized maximum-likelihood. in Proceedings - IEEE International Workshop on Genomic Signal Processing and Statistics., 6169451, pp. 91-94, 2011 IEEE International Workshop on Genomic Signal Processing and Statistics, GENSIPS'11, San Antonio, TX, United States, 12/4/11.
Esfahani MS, Zollanvari A, Yoon BJ, Dougherty ER. Designing enhanced classifiers using prior process knowledge: Regularized maximum-likelihood. In Proceedings - IEEE International Workshop on Genomic Signal Processing and Statistics. 2011. p. 91-94. 6169451
Esfahani, Mohammad Shahrokh ; Zollanvari, Amin ; Yoon, Byung Jun ; Dougherty, Edward R. / Designing enhanced classifiers using prior process knowledge : Regularized maximum-likelihood. Proceedings - IEEE International Workshop on Genomic Signal Processing and Statistics. 2011. pp. 91-94
@inproceedings{9294669443eb4e66a6eaea36aab251c7,
title = "Designing enhanced classifiers using prior process knowledge: Regularized maximum-likelihood",
abstract = "We propose a novel optimization-based paradigm for designing enhanced classifiers. The proposed paradigm allows us to incorporate available prior process knowledge into classifier design, thereby improving the performance of the resulting classifiers. In this work, we focus on dynamical systems that can be represented as finite-state multi-dimensional stochastic processes that possess labeled steady-state distributions. Given prior operational knowledge of the process, our goal is to build a classifier that can accurately label future observations obtained from the steady-state, by utilizing both the available prior knowledge and the training data. Simulation results show that the proposed paradigm yields improved classifiers that outperform traditional classifiers that use only training data.",
author = "Esfahani, {Mohammad Shahrokh} and Amin Zollanvari and Yoon, {Byung Jun} and Dougherty, {Edward R.}",
year = "2011",
language = "English",
isbn = "9781467304900",
pages = "91--94",
booktitle = "Proceedings - IEEE International Workshop on Genomic Signal Processing and Statistics",

}

TY - GEN

T1 - Designing enhanced classifiers using prior process knowledge

T2 - Regularized maximum-likelihood

AU - Esfahani, Mohammad Shahrokh

AU - Zollanvari, Amin

AU - Yoon, Byung Jun

AU - Dougherty, Edward R.

PY - 2011

Y1 - 2011

N2 - We propose a novel optimization-based paradigm for designing enhanced classifiers. The proposed paradigm allows us to incorporate available prior process knowledge into classifier design, thereby improving the performance of the resulting classifiers. In this work, we focus on dynamical systems that can be represented as finite-state multi-dimensional stochastic processes that possess labeled steady-state distributions. Given prior operational knowledge of the process, our goal is to build a classifier that can accurately label future observations obtained from the steady-state, by utilizing both the available prior knowledge and the training data. Simulation results show that the proposed paradigm yields improved classifiers that outperform traditional classifiers that use only training data.

AB - We propose a novel optimization-based paradigm for designing enhanced classifiers. The proposed paradigm allows us to incorporate available prior process knowledge into classifier design, thereby improving the performance of the resulting classifiers. In this work, we focus on dynamical systems that can be represented as finite-state multi-dimensional stochastic processes that possess labeled steady-state distributions. Given prior operational knowledge of the process, our goal is to build a classifier that can accurately label future observations obtained from the steady-state, by utilizing both the available prior knowledge and the training data. Simulation results show that the proposed paradigm yields improved classifiers that outperform traditional classifiers that use only training data.

UR - http://www.scopus.com/inward/record.url?scp=84863715202&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84863715202&partnerID=8YFLogxK

M3 - Conference contribution

SN - 9781467304900

SP - 91

EP - 94

BT - Proceedings - IEEE International Workshop on Genomic Signal Processing and Statistics

ER -