Double Viterbi: Weight Encoding for High Compression Ratio and Fast On-Chip Reconstruction for Deep Neural Network.
Daehyun AhnDongsoo LeeTaesu KimJae-Joon KimPublished in: ICLR (Poster) (2019)
Keyphrases
- high compression ratio
- reconstructed image
- neural network
- image compression
- image reconstruction
- image quality
- compression scheme
- compression ratio
- visual quality
- low complexity
- fractal image compression
- compression algorithm
- image transmission
- bit rate
- high speed
- post processing
- lossless compression
- spatial resolution
- data compression
- high resolution
- compressed images
- vector quantization
- compressed data
- motion estimation
- image data
- bitstream
- region of interest
- high frequency
- super resolution
- medical images
- entropy coding
- rate distortion
- multiscale