convolutional-sparse-coding

This paper proposes a training-adaptive convolutional sparse coding framework that leverages information bottleneck principles for robust visual representation, achieving improved performance on CIFAR and ImageNet under input perturbations.