Comparison of score normalization methods applied to multi-label classification

Our paper deals with the multi-label text classification of the newspaper articles, where the classifier must decide if a document does or does not belong to each topic from the predefined topic set. A generative classifier is used to tackle this task and the problem with finding a threshold for the positive classification is mainly addressed. This threshold can vary for each document depending on the content of the document (words used, length of the document, etc.). An extensive comparison of the score normalization methods, primary proposed in the speaker identification/verification task, for robustly finding the threshold defining the boundary between the “correct” and the “incorrect” topics of a document is presented. Score normalization methods (based on World Model and Unconstrained Cohort Normalization) applied to the topic identification task has shown an improvement of results in our former experiments, therefore in this paper an in-depth experiments with more score normalization techniques applied to the multi-label classification were performed. Thorough analysis of the effects of the various parameters setting is presented.

[1]  William J. Byrne,et al.  Large vocabulary ASR for spontaneous czech in the MALACH project , 2003, INTERSPEECH.

[2]  Saso Dzeroski,et al.  An extensive experimental comparison of methods for multi-label learning , 2012, Pattern Recognit..

[3]  Ronald Rosenfeld,et al.  Using story topics for language model adaptation , 1997, EUROSPEECH.

[4]  Zbynek Zajíc,et al.  Score Normalization Methods Applied to Topic Identification , 2014, TSD.

[5]  Aladdin M. Ariyaeeinia,et al.  Score normalisation applied to open-set, text-independent speaker identification , 2003, INTERSPEECH.

[6]  Ludek Müller,et al.  Automatic Lemmatizer Construction with Focus on OOV Words Lemmatization , 2005, TSD.

[7]  Zhi-Hua Zhou,et al.  A k-nearest neighbor based algorithm for multi-label classification , 2005, 2005 IEEE International Conference on Granular Computing.

[8]  Roland Auckenthaler,et al.  Score Normalization for Text-Independent Speaker Verification Systems , 2000, Digit. Signal Process..

[9]  Lucie Skorkovská,et al.  Dynamic Threshold Selection Method for Multi-label Newspaper Topic Identification , 2013, TSD.

[10]  Jiebo Luo,et al.  Learning multi-label scene classification , 2004, Pattern Recognit..

[11]  Jan Svec,et al.  System for fast lexical and phonetic spoken term detection in a Czech cultural heritage archive , 2011, EURASIP J. Audio Speech Music. Process..

[12]  Sunita Sarawagi,et al.  Discriminative Methods for Multi-labeled Classification , 2004, PAKDD.

[13]  Adi Wahyu Pribadi,et al.  Automatic news articles classification in Indonesian language by using Naive Bayes Classifier method , 2009, iiWAS.

[14]  Daniel Soutner,et al.  Web Text Data Mining for Building Large Scale Language Modelling Corpus , 2011, TSD.

[15]  Douglas A. Reynolds,et al.  Speaker Verification Using Adapted Gaussian Mixture Models , 2000, Digit. Signal Process..

[16]  Biing-Hwang Juang,et al.  The use of cohort normalized scores for speaker verification , 1992, ICSLP.

[17]  Arnon D. Cohen,et al.  On cohort selection for speaker verification , 2003, INTERSPEECH.

[18]  Ludek Müller,et al.  Benefit of Proper Language Processing for Czech Speech Retrieval in the CL-SR Task at CLEF 2006 , 2006, CLEF.

[19]  Shingo Kuroiwa,et al.  Category Classification and Topic Discovery of Japanese and English News Articles , 2006, MFCSIT.

[20]  Hinrich Schütze,et al.  Introduction to information retrieval , 2008 .

[21]  Pavel Ircing,et al.  Automatic Topic Identification for Large Scale Language Modeling Data Filtering , 2011, TSD.

[22]  Grigorios Tsoumakas,et al.  Multi-Label Classification: An Overview , 2007, Int. J. Data Warehous. Min..