Please use this identifier to cite or link to this item: http://localhost:8080/xmlui/handle/123456789/2201
Full metadata record
DC FieldValueLanguage
dc.contributor.authorPushpa M-
dc.contributor.authorKarpagavalli S-
dc.date.accessioned2020-10-13T09:06:30Z-
dc.date.available2020-10-13T09:06:30Z-
dc.date.issued2017-
dc.identifier.urihttps://www.sciencedirect.com/science/article/pii/S1877050917319440-
dc.identifier.urihttps://doi.org/10.1016/j.procs.2017.09.116-
dc.identifier.urihttp://localhost:8080/xmlui/handle/123456789/2201-
dc.description.abstractMost of the supervised learning task has been carried out using single label classification and solved as binary or multiclass classification problems. The hierarchical relationship among the classes leads to Multi- Label (ML) classification which is learning from a set of instances that are associated with a set of labels. In Tamil language, phonemes fall into different categories according to place and manner of articulation. This motivates the application of multi-label classification methods to classify Tamil phonemes. Experiments are carried out using Binary Relevance (BR) and Label Powerset (LP) and BR’s improvement Classifier Chains (CC) methods with different base classifiers and the results are analysed.en_US
dc.language.isoenen_US
dc.publisherIn Proceedings of 7th International Conference on Advances in Computing & Communications (ICACC August 2017), Kochi, Elsevier Procedia Computer Science.en_US
dc.subjectSupervised learningen_US
dc.subjectMulti-label classificationen_US
dc.subjectSpeech recognitionen_US
dc.subjectPhoneme classificationen_US
dc.titleMULTI-LABEL CLASSIFICATION- PROBLEM TRANSFORMATION METHODS IN TAMIL PHONEME CLASSIFICATIONen_US
dc.typeBooken_US
Appears in Collections:International Conference

Files in This Item:
File Description SizeFormat 
MULTI-LABEL CLASSIFICATION- PROBLEM TRANSFORMATION METHODS IN TAMIL PHONEME CLASSIFICATION.docx10.34 kBMicrosoft Word XMLView/Open


Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.