Vector quantization using <i>k</i> ‐means clustering neural network

Vector Quantization (VQ) is a clustering problem in the fields of signal processing, source coding, information theory etc. Taking advantage recent advances field deep neural networks, this paper investigates performance between VQ problems and networks. A k-means-based network architecture for presented to solve problems. By applying learning implementation convergence optimization, (algorithm) purpose proposed. In practice, proposed quantifies vectors over set stacked layers, overcoming exponential complexity methods by trainable parameters. Experiments show that work can improve results without human intervention, outperforms traditional modified VQ.