Application Example: An Adaptive Neural Network Source Coder
In this chapter, I apply the S-Level network SPAN introduced in Chapter 4 to source coding problems for digital communication. The basic idea is to use SPAN as an active codebook that can grow from scratch to follow the statistics of source signals, capture the local context of the source signal space, and map onto the structure of the network. As a result, when the statistics of the source signals change, the network can dynamically modify its structure to follow the change.
KeywordsVector Quantization Finite State Automaton Optimum Distortion Rate Path Code Neuron Position
Unable to display preview. Download preview PDF.