1 paper
Ikumi Okubo, Keisuke Sugiura, Hiroki Matsutani
Transformer has been adopted to image recognition tasks and shown to outperform CNNs and RNNs while it suffers from high training cost and computational complexity. To address thes…