Bitwidth-Adaptive Quantization-Aware Neural Network Training: A Meta-Learning Approach.
Jiseok YounJaehun SongHyung-Sin KimSaewoong BahkPublished in: ECCV (12) (2022)
Keyphrases
- meta learning
- neural network training
- adaptive quantization
- shape coding
- neural network
- training algorithm
- rate distortion
- inductive learning
- learning tasks
- low bit rate
- model selection
- subband coding
- image compression
- decision trees
- subband
- optimization method
- machine learning algorithms
- data mining
- machine learning
- metamodel
- base classifiers
- feature selection
- particle swarm optimisation
- image quality
- bit rate
- active learning
- motion estimation
- training samples
- learning process
- high dimensional
- feature space
- supervised learning
- learning algorithm
- back propagation