Romero, David W. (author), Bruintjes, R. (author), Bekkers, Erik J. (author), Tomczak, Jakub M. (author), Hoogendoorn, Mark (author), van Gemert, J.C. (author) When designing Convolutional Neural Networks (CNNs), one must select the size of the convolutional kernels before training. Recent works show CNNs benefit from different kernel sizes at different layers, but exploring all possible combinations is unfeasible in practice. A more efficient approach is to learn the kernel size during training....
conference paper 2022
