Searched for: subject%3A%22attention%22
(1 - 1 of 1)
document
Zhu, B. (author), Hofstee, H.P. (author), Lee, Jinho (author), Al-Ars, Z. (author)
Attention mechanism has been regarded as an advanced technique to capture long-range feature interactions and to boost the representation capability for convolutional neural networks. However, we found two ignored problems in current attentional activations-based models: the approximation problem and the insufficient capacity problem of the...
conference paper 2021