A Bayesian approach for initialization of weights in backpropagation neural net with application to character recognition
This work addresses convergence issues in neural network training for character recognition, but it is incremental as it builds on existing initialization techniques.
The paper tackles the problem of slow convergence in neural network training by proposing a Bayesian weight initialization method based on a customized Kalman filter, applied to character recognition, and reports improved convergence rates compared to random initialization and other methods.
Convergence rate of training algorithms for neural networks is heavily affected by initialization of weights. In this paper, an original algorithm for initialization of weights in backpropagation neural net is presented with application to character recognition. The initialization method is mainly based on a customization of the Kalman filter, translating it into Bayesian statistics terms. A metrological approach is used in this context considering weights as measurements modeled by mutually dependent normal random variables. The algorithm performance is demonstrated by reporting and discussing results of simulation trials. Results are compared with random weights initialization and other methods. The proposed method shows an improved convergence rate for the backpropagation training algorithm.