NeurIPS 2019
Sun Dec 8th through Sat the 14th, 2019 at Vancouver Convention Center
Paper ID:719
Title:Convolution with even-sized kernels and symmetric padding


		
This paper proposes a technique to improve convolutional neural networks. The technique relies on the use of symmetric padding to address the shift problem in even-sized convolutions. The reviewers found the method to be sound and the experimental validation on CIFAR-10, CIFAR-100, and ImageNet to be convincing. The concerns raised by the reviewers were later addressed by the rebuttal.