Most implementations of Neural Machine Translation (NMT) models employ a padding strategy when processing a mini-batch to make all sentences in a mini-batch have the same length. This enables an ...

We introduce O-SegNet- the robust encoder and decoder architecture for objects segmentation from high-resolution aerial imagery data to address this challenge. The proposed O-SegNet architecture ...

We introduce Fourier mixed window attention (FWin) to the self-attention blocks in the encoder and decoder of Informer. Specifically, we replace the ProbSparse self-attention blocks in the encoder and ...