SqueezeBlock: A Transparent Weight Compression Scheme for Deep Neural Networks.
Mo SongJiajun WuYuhao DingHayden Kwok-Hay SoPublished in: ICFPT (2023)
Keyphrases
- compression scheme
- neural network
- image compression
- data compression
- compression ratio
- compression algorithm
- weight update
- text compression
- entropy coding
- quantization scheme
- run length encoding
- jpeg xr
- back propagation
- artificial neural networks
- neural network model
- vector quantization
- computer vision
- subband
- binary images
- multiscale
- image processing
- jpeg ls
- huffman coding
- machine learning