Training-Adaptive Convolutional Sparse Coding via Information Bottleneck for Robust Visual Representation

arXiv:2609.19122, 2026