Rede Neural Probabilística para a Classificação de Atividades Econômicas

• Ano de defesa: 2008
• Autor(a) principal: Ciarelli, Patrick Marques
• Tipo de documento: Dissertação
• Tipo de acesso: Acesso aberto
• Idioma: por
• Instituição de defesa: Universidade Federal do Espírito Santo; BR; Mestrado em Engenharia Elétrica; Centro Tecnológico; UFES; Programa de Pós-Graduação em Engenharia Elétrica
• Palavras-chave em Português: Reconhecimento de padrões.; Redes neurais (Computação); Eletrônica Industrial, Sistemas e Controles Eletrônicos; 621.3
• Link de acesso: http://repositorio.ufes.br/handle/10/4056

Thisworkpresents anapproach basedonArtificialNeuralNetworksforproblemsofmulti-label classification.In particular,was useda modifiedversionof ProbabilisticNeuralNetworkto handlesuch problems.In experiments carriedoutin variousdatabasesknownin theliterature,theProbabilisticNeuralNetworkproposalpresenteda performancecomparable,andsometimesevensuperiorto otheralgorithmsspecializedin thistype ofproblem.Asthemainfocusof thisworkwas thestudyof strategiesforautomatictextclassi-ficationof economicactivitiesthenwerealsoconductedexperiments usinga databaseofeconomicactivities.However,unlike of databasesusedpreviously, thisdatabaseshowsahugenumber of categoriesandfewsamplesof trainingby category, which increasesthedegreeof difficulty thisproblem.In theexperiments wereusedto ProbabilisticNeuralNetworkproposal,theclassifierMulti-label k-NearestNeighbor anda GeneticAlgorithmforoptimizationof theparameters.Themetricsusedto evaluationof performancehaveshownthattheresultsof ProbabilisticNeuralNetworkweresuperiorandcomparabletotheresultsobtainedby theMulti-label k-NearestNeighbor,showingthattheapproachusedin thisworkis promising.