Bagging Based Feature Selection for Dimensional Affect Recognition in the continuous Emotion Space

This paper exploits the Bagging based feature selection method on the baseline audio features provided by AVEC2012 challenge competition. The selected features are input to SVR and RVM regression models, respectively, to estimate the affect dimensions arousal, valence, expectation, and power embedded in the audio speech. Experiments have been carried out on the word based and frame based baseline features, respectively, and the Pearson correlations between the estimated affect dimensions and their ground-truth labels are compared to those from the traditional correlation based feature selection (CFS) method with BestFirst or sequential floating forward selection (SFFS) algorithm. Experimental results show that both on word based and frame based baseline feature selection obtains the best accuracy in estimating the affect dimensions, while keeping the lowest number of features.

[1]  Johannes Wagner,et al.  From Physiological Signals to Emotions: Implementing and Comparing Selected Methods for Feature Extraction and Classification , 2005, 2005 IEEE International Conference on Multimedia and Expo.

[2]  K. Scherer,et al.  The World of Emotions is not Two-Dimensional , 2007, Psychological science.

[3]  F. J. Anscombe,et al.  Graphs in Statistical Analysis , 1973 .

[4]  Constantine Kotropoulos,et al.  Fast sequential floating forward selection applied to emotional speech features estimated on DES and SUSAS data collections , 2006, 2006 14th European Signal Processing Conference.

[5]  Björn W. Schuller,et al.  AVEC 2012: the continuous audio/visual emotion challenge , 2012, ICMI '12.

[6]  Leo Breiman,et al.  Bagging Predictors , 1996, Machine Learning.

[7]  Michael E. Tipping,et al.  Fast Marginal Likelihood Maximisation for Sparse Bayesian Models , 2003 .

[8]  Björn W. Schuller,et al.  Combining Long Short-Term Memory and Dynamic Bayesian Networks for Incremental Emotion-Sensitive Artificial Listening , 2010, IEEE Journal of Selected Topics in Signal Processing.

[9]  Arman Savran,et al.  Combining video, audio and lexical indicators of affect in spontaneous conversation via particle filtering , 2012, ICMI '12.

[10]  Mark A. Hall,et al.  Correlation-based Feature Selection for Discrete and Numeric Class Machine Learning , 1999, ICML.

[11]  Leo Breiman,et al.  Bagging Predictors , 1996, Machine Learning.

[12]  J. Russell.  A circumplex model of affect. , 1980 .

[13]  Björn W. Schuller,et al.  AVEC 2011-The First International Audio/Visual Emotion Challenge , 2011, ACII.

[14]  Josef Kittler,et al.  Floating search methods in feature selection , 1994, Pattern Recognit. Lett..

[15]  Nasser M. Nasrabadi,et al.  Pattern Recognition and Machine Learning , 2006, Technometrics.

[16]  Chih-Jen Lin,et al.  LIBSVM: A library for support vector machines , 2011, TIST.

[17]  T. Dalgleish.  Basic Emotions , 2004 .