Music genre classification using traditional and relational approaches Jorge Valverde-Rebaza, Aurea Soriano, Lilian Berton, Maria Cristina F. Oliveira and Alneu de Andrade Lopes Laboratory of Computational Intelligence (LABIC) Laboratory of Visualization, Imaging and Computer Graphics (VICG) University of S˜ ao Paulo (USP) Brazil October 2014
Outline Introduction 1 Method 2 Experimental Evaluation 3 Conclusion 4 Jorge Valverde-Rebaza et al. Music genre classification using traditional and relational . . . 2 / 20
Outline Introduction 1 Method 2 Music Features Graph construction Graph construction Experimental Evaluation 3 Datasets Experimental setup Results Conclusion 4 Jorge Valverde-Rebaza et al. Music genre classification using traditional and relational . . . 3 / 20
Introduction Many music collections, typically very large Jorge Valverde-Rebaza et al. Music genre classification using traditional and relational . . . 4 / 20
Introduction Many music collections, typically very large Manual music classification: a non-expert person can identify the genre of a music with 72% accuracy after hearing 3 seconds of it [Perrot and Gjerdigen, 1999]. Classifying a large collection demands time, effort and expertise Jorge Valverde-Rebaza et al. Music genre classification using traditional and relational . . . 4 / 20
Introduction Many music collections, typically very large Manual music classification: a non-expert person can identify the genre of a music with 72% accuracy after hearing 3 seconds of it [Perrot and Gjerdigen, 1999]. Classifying a large collection demands time, effort and expertise Automatic music classification : Solutions achieve high accuracy (ranging from 63 to 84%) [Shao et al., 2004, Pampalk et al., 2005, Scaringella and Mlynek, 2005, Yaslan and Cataltepe, 2009, Poria et al., 2013] require music files with the same size applied on controlled environments, i.e. without considering class imbalance employed traditional classifiers Jorge Valverde-Rebaza et al. Music genre classification using traditional and relational . . . 4 / 20
Outline Introduction 1 Method 2 Music Features Graph construction Graph construction Experimental Evaluation 3 Datasets Experimental setup Results Conclusion 4 Jorge Valverde-Rebaza et al. Music genre classification using traditional and relational . . . 5 / 20
Music feature extraction Feature extracted from MIDI description Jorge Valverde-Rebaza et al. Music genre classification using traditional and relational . . . 6 / 20
Music feature extraction Feature extracted from MIDI description (12)Histograms Distance: DTW Jorge Valverde-Rebaza et al. Music genre classification using traditional and relational . . . 6 / 20
Music feature extraction Feature extracted from MIDI description (8) Moments (4) Melody and (4) Rhythm: the (12)Histograms mean, standard deviation, entropy and uniformity Distance: DTW Distance: Euclidean Jorge Valverde-Rebaza et al. Music genre classification using traditional and relational . . . 6 / 20
Music feature extraction Feature extracted from MIDI description (248)Structure Identification of patterns from (8) Moments chord sequences. Generates (4) Melody and (4) Rhythm: the (12)Histograms vectors of different sizes. mean, standard deviation, entropy and uniformity Distance: DTW Distance: DTW Distance: Euclidean Jorge Valverde-Rebaza et al. Music genre classification using traditional and relational . . . 6 / 20
Music feature extraction Feature extracted from MIDI description (248)Structure Identification of patterns from (8) Moments chord sequences. Generates (4) Melody and (4) Rhythm: the (12)Histograms vectors of different sizes. mean, standard deviation, entropy and uniformity Distance: DTW Distance: DTW Distance: Euclidean Jorge Valverde-Rebaza et al. Music genre classification using traditional and relational . . . 6 / 20
Graph construction Graph construction techniques Jorge Valverde-Rebaza et al. Music genre classification using traditional and relational . . . 7 / 20
Graph construction Graph construction techniques kNN Connect each vertex only to its k nearest neighbors Jorge Valverde-Rebaza et al. Music genre classification using traditional and relational . . . 7 / 20
Graph construction Graph construction techniques kNN mutual-kNN Two vertices are Connect each vertex only connected only if the to its k nearest neighbors neighborhood pertinence condition is met by both Jorge Valverde-Rebaza et al. Music genre classification using traditional and relational . . . 7 / 20
Graph construction Graph construction techniques regular-kNN kNN mutual-kNN Two vertices are Connect each vertex only connected only if the All the vertices have the to its k nearest neighbors neighborhood pertinence same degree condition is met by both Jorge Valverde-Rebaza et al. Music genre classification using traditional and relational . . . 7 / 20
Graph construction Graph construction techniques regular-kNN kNN mutual-kNN Two vertices are Connect each vertex only connected only if the All the vertices have the to its k nearest neighbors neighborhood pertinence same degree condition is met by both Jorge Valverde-Rebaza et al. Music genre classification using traditional and relational . . . 7 / 20
Classifiers Classifiers Jorge Valverde-Rebaza et al. Music genre classification using traditional and relational . . . 8 / 20
Classifiers Classifiers Traditional Map input data to a category Decision trees, na¨ ıve Bayes, neural networks, support vector machine, etc Jorge Valverde-Rebaza et al. Music genre classification using traditional and relational . . . 8 / 20
Classifiers Classifiers Traditional Map input data to a category Decision trees, na¨ ıve Bayes, neural networks, support vector machine, etc Jorge Valverde-Rebaza et al. Music genre classification using traditional and relational . . . 8 / 20
Classifiers Classifiers Traditional Relational Map input data to a Map relational input data to a category category Decision trees, na¨ ıve Weighted vote relational neighbor, Bayes, neural networks, network-only Bayes, probabilistic support vector machine, relational neighbor, network-only etc link based [Macskassy, 2007] Jorge Valverde-Rebaza et al. Music genre classification using traditional and relational . . . 8 / 20
Classifiers Classifiers Traditional Relational Map input data to a Map relational input data to a category category Decision trees, na¨ ıve Weighted vote relational neighbor, Bayes, neural networks, network-only Bayes, probabilistic support vector machine, relational neighbor, network-only etc link based [Macskassy, 2007] Jorge Valverde-Rebaza et al. Music genre classification using traditional and relational . . . 8 / 20
Outline Introduction 1 Method 2 Music Features Graph construction Graph construction Experimental Evaluation 3 Datasets Experimental setup Results Conclusion 4 Jorge Valverde-Rebaza et al. Music genre classification using traditional and relational . . . 9 / 20
Datasets Tabela: Music genre distribution for the music collection considered Genre # Tracks Classical 31 Brazilian Backcountry 243 Pop/Rock 550 Jazz 95 Total 919 Jorge Valverde-Rebaza et al. Music genre classification using traditional and relational . . . 10 / 20
Experimental setup Feature vectors Histogram Moments Structure Jorge Valverde-Rebaza et al. Music genre classification using traditional and relational . . . 11 / 20
Experimental setup Feature vectors Histogram Moments Structure Graph construction (1 ≤ k ≤ 15 ) regular-kN kNN mutual-kNN Jorge Valverde-Rebaza et al. Music genre classification using traditional and relational . . . 11 / 20
Experimental setup Feature vectors Histogram Moments Structure Graph construction (1 ≤ k ≤ 15 ) regular-kN kNN mutual-kNN Classifiers Traditional Relational Decision tree (J48) weighted vote relational neighbor (wvrn) Na¨ ıve Bayes (NB) network-only Bayes (no-Bayes) Multilayer perceptron with probabilistic relational neighbor (prn) backpropagation (MLP) network-only link-based Support vector machine (SMO) mode-link (no-lb-mode) count-link (no-lb-count) binary-link (no-lb-binary) class-distribution-link (no-lb-distrib) Jorge Valverde-Rebaza et al. Music genre classification using traditional and relational . . . 11 / 20
CD 1 2 3 4 MLP SMO NB J48 Traditional classification results Tabela: Traditional classifiers performance measured by AUC J48 NB MLP SMO Histogram 0.619 0.607 0.665 0.506 Moments 0.706 0.750 0.771 0.585 Structure 0.738 0.920 0.816 0.724 Average rank 2.667 2.000 1.333 4.000 Jorge Valverde-Rebaza et al. Music genre classification using traditional and relational . . . 12 / 20
Traditional classification results Tabela: Traditional classifiers performance measured by AUC J48 NB MLP SMO Histogram 0.619 0.607 0.665 0.506 Moments 0.706 0.750 0.771 0.585 Structure 0.738 0.920 0.816 0.724 Average rank 2.667 2.000 1.333 4.000 CD 1 2 3 4 MLP SMO NB J48 Figura: Post-hoc test results for traditional classifiers performance Jorge Valverde-Rebaza et al. Music genre classification using traditional and relational . . . 12 / 20
Recommend
More recommend
