Aligned Structured Sparsity Learning for Efficient Image Super-Resolution

Neural Information Processing Systems 

Lightweight image super-resolution (SR) networks have obtained promising results with moderate model size.