Error-Tolerant Computation for Voting Classifiers With Multiple Classes

e-Archivo Repository

Show simple item record Liu, Shanshan Reviriego Vasallo, Pedro Montuschi, Paolo Lombardi, Fabrizio 2021-02-09T15:43:58Z 2021-02-09T15:43:58Z 2020-09-21
dc.identifier.bibliographicCitation S. Liu, P. Reviriego, P. Montuschi and F. Lombardi, "Error-Tolerant Computation for Voting Classifiers With Multiple Classes," in IEEE Transactions on Vehicular Technology, vol. 69, no. 11, pp. 13718-13727, Nov. 2020
dc.identifier.issn 0018-9545
dc.description.abstract In supervised learning, labeled data are provided as inputs and then learning is used to classify new observations. Error tolerance should be guaranteed for classifiers when they are employed in critical applications. A widely used type of classifiers is based on voting among instances (referred to as single voter classifiers) or multiple voters (referred to as ensemble classifiers). When the classifiers are implemented on a processor, Time-Based Modular Redundancy (TBMR) techniques are often used for protection due to the inflexibility of the hardware. In TBMR, any single error can be handled at the cost of additional computing either once for detection or twice for correction after detection; however, this technique increases the computation overhead by at least 100%. The Voting Margin (VM) scheme has recently been proposed to reduce the computation overhead of TBMR, but this scheme has only been utilized for k Nearest Neighbors ( k NNs) classifiers with two classes. In this paper, the VM scheme is extended to multiple classes, as well as other voting classifiers by exploiting the intrinsic robustness of the algorithms. k NNs (that is a single voter classifier) and Random Forest (RF) (that is an ensemble classifier) are considered to evaluate the proposed scheme. Using multiple datasets, the results show that the proposed scheme significantly reduces the computation overhead by more than 70% for k NNs with good classification accuracy and by more than 90% for RF in all cases. However, when extended to multiple classes, the VM scheme for k NNs is not efficient for some datasets. In this paper, a new protection scheme referred to as k + 1 NNs is presented as an alternative option to provide efficient protection in those scenarios. In the new scheme, the computation overhead can be further reduced at the cost of allowing a very low percentage of errors that can modify the classification outcome.
dc.description.sponsorship This work was supported in part by the ACHILLES Project PID2019-104207RB-I00 and the Go2Edge network RED2018-102585-T funded by the Spanish Ministry of Economy and Competitivity, in part by the Department of Research and Innovation of Madrid Regional Authority, in part by the EMPATIA-CM Research Project (Reference Y2018/TCS-5046), and in part by NSF under Grants CCF-1953961 and 1812467
dc.language.iso eng
dc.publisher IEEE
dc.rights © 2020 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission.
dc.subject.other Machine learning
dc.subject.other Voting classifier
dc.subject.other Error tolerance
dc.subject.other k nearest neighbors
dc.subject.other Random forest
dc.title Error-Tolerant Computation for Voting Classifiers With Multiple Classes
dc.type article
dc.subject.eciencia Telecomunicaciones
dc.rights.accessRights openAccess
dc.relation.projectID Gobierno de España. PID2019-104207RB-I00
dc.relation.projectID Gobierno de España. RED2018-102585-T
dc.relation.projectID Comunidad de Madrid. Y2018/TCS-5046/EMPATIA-CM
dc.type.version acceptedVersion
dc.identifier.publicationfirstpage 13718
dc.identifier.publicationissue 11
dc.identifier.publicationlastpage 13727
dc.identifier.publicationtitle IEEE Transactions on Vehicular Technology
dc.identifier.publicationvolume 69
dc.identifier.uxxi AR/0000026482
dc.contributor.funder Ministerio de Economía y Competitividad (España)
dc.contributor.funder Comunidad de Madrid
dc.affiliation.dpto UC3M. Departamento de Ingeniería Telemática
dc.affiliation.instituto UC3M. Instituto Universitario de Estudios de Género
dc.affiliation.grupoinv UC3M. Grupo de Investigación: Network Technologies
 Find Full text

Files in this item

*Click on file's image for preview. (Embargoed files's preview is not supported)

This item appears in the following Collection(s)

Show simple item record