Some well known theoretical results concerning the universal approximation property of MLP neural networks with one hidden layer have shown that for any function / from [0,1]n to R, only the output layer weights depend on f. We use this result to propose a network architecture called the predictive Kohonen map allowing to design a new speech features extractor. We give experimental results of this approach on a phonemes recognition task. © Springer-Verlag Berlin Heidelberg 2005.
CITATION STYLE
Gas, B., Chetouani, M., Zarader, J. L., & Charbuillet, C. (2005). Predictive Kohonen map for speech features extraction. In Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) (Vol. 3697 LNCS, pp. 793–798). https://doi.org/10.1007/11550907_125
Mendeley helps you to discover research relevant for your work.