Inference, learning and attention mechanisms that exploit and preserve sparsity in convolutional networks
While CNNs naturally lend themselves to densely sampled data, and sophisticated implementations are available, they lack the ability to efficiently process sparse data. In this …
Timo hackel
