Modeling parallel feed-forward based compression network

Abstract

The objective of this paper is to model a simple feed-forward type three-layered neural network based model for image and video data compression. The input to the neural network is the original data while the output is the reconstructed data that is close to the input. If the amount of data required to store the hidden unit values and the connection weights to the output layer of the neural network is less than the original data then compression is said to be achieved. During training if the desired accuracy is not obtained one or more hidden layer neurons are added. An acceleration technique is used which performs a local adaptation of the weight-updates according to the behavior of the error function. This leads to an efficient and transparent adaptation process. The implementation of this model is done on a parallel cluster of computer environment to increase the learning efficiency of the network. Experimental results show that the model provides better results than conventional compression techniques.

Keywords:

• Feed-forward network
• Image and video compression
• Parallel cluster
• Compression ratio
• Acceleration technique

Keywords:

• Neural nets
• Learning and adaptive systems