A new competitive learning approach based on an equidistortion principle for designing optimal vector quantizers.
Naonori UedaRyohei NakanoPublished in: Neural Networks (1994)
Keyphrases
- competitive learning
- codebook design
- vector quantizer
- vector quantization
- neural network
- information theoretic
- image coding
- self organizing maps
- entropy constrained
- unsupervised learning
- data clustering
- dynamic programming
- feature selection
- k means
- variable length
- finite state
- high dimensional
- image processing
- joint optimization