Self-structuring hidden control neural model for speech recognition

The majority of neural models for pattern recognition have fixed architecture during training. A typical consequence is nonoptimal and often too large networks. A self-structuring hidden control (SHC) neural model for pattern recognition that establishes a near-optimal architecture during training is proposed. A network architecture reduction of approximately 80-90% in terms of the number of hidden processing elements (PEs) is typically achieved. The SHC model combines self-structuring architecture generation with nonlinear prediction and hidden Markov modeling. A theorem for self-structuring neural models that states that these models are universal approximators and thus relevant for real-world pattern recognition is presented. Using SHC models containing as few as five hidden PEs each for an isolated word recognition task resulted in a recognition rate of 98.4%. SHC models can furthermore be applied to continuous speech recognition