Towards Better Orthogonality Regularization with Disentangled Norm in Training Deep CNNs.
Changhao WuShenan ZhangFangsong LongZiliang YinTuo LengPublished in: CoRR (2023)
Keyphrases
- test set
- training set
- training data
- regularization method
- mixed norm
- feature space
- trace norm
- supervised learning
- training samples
- early stopping
- tikhonov regularization
- cellular neural networks
- penalty term
- convex optimization
- sparsity inducing
- structured sparsity
- inverse problems
- data sets
- prior information
- back propagation
- neural network