论文信息 - Sélection de descripteurs audio pour la classification des sons environnementaux avec des SVMs mono-classe

Sélection de descripteurs audio pour la classification des sons environnementaux avec des SVMs mono-classe

This paper addresses a speciflc audio classiflcation problem for surveillance and security applications. The developed approach uses original audio features including essentially wavelets-based features, and a procedure of multi-class classiflcation using several One-class SVMs. The originality of our work lies primarily in the proposal of a selection procedure of the optimal audio features combinations, as well as a new method of multi-class classiflcation. These new approaches are evaluated on an original and su-ciently large database.

[1] Horst M. Eidenberger,et al. Discrimination and Retrieval of Environmental Sounds , 2005 .

[2] Eric R. Ziegel,et al. The Elements of Statistical Learning , 2003, Technometrics.

[3] Stéphane Canu,et al. Estimation of Minimum Measure Sets in Reproducing Kernel Hilbert Spaces and Applications. , 2006, 2006 IEEE International Conference on Acoustics Speech and Signal Processing Proceedings.

[4] Arthur Gretton,et al. An online support vector machine for abnormal events detection , 2006, Signal Process..

[5] Alexander J. Smola,et al. Learning with Kernels: support vector machines, regularization, optimization, and beyond , 2001, Adaptive computation and machine learning series.

[6] Alain Dufaux. Detection and Recognition of Impulsive Sound Signals , 2001 .

[7] A. Atiya,et al. Learning with Kernels: Support Vector Machines, Regularization, Optimization, and Beyond , 2005, IEEE Transactions on Neural Networks.