| U. Bodenhausen: Automatic Structuring of Neural Networks for Spatio-Temporal Real-World Applications, Ph.D thesis, University of Karlsruhe, June 1994 |
....phonemes and the size of the feature space phonemes cover varies greatly, constant codebook size leads to suboptimal allocation of resources. We have therefore suggested [12] to adapt the codebook size of each phoneme according to the amount and the distribution of the training data, similar to [15]. During training, the size of the codebook is incrementally increased. Some quality criterion determines when to stop the process of increasing the codebook. We compared a variance criterion based on the average distance between data points and their nearest codebook vector with a prediction ....
U. Bodenhausen: Automatic Structuring of Neural Networks for Spatio-Temporal Real-World Applications, Ph.D thesis, University of Karlsruhe, June 1994
Online articles have much greater impact More about CiteSeer.IST Add search form to your site Submit documents Feedback
CiteSeer.IST - Copyright Penn State and NEC