Striving for Simplicity: The All Convolutional Net
This work simplifies CNN design for object recognition, potentially benefiting researchers and practitioners by reducing architectural complexity while maintaining performance.
The paper questions the necessity of max-pooling in convolutional neural networks for object recognition, finding it can be replaced by a convolutional layer with increased stride without accuracy loss, and proposes an all-convolutional architecture achieving competitive or state-of-the-art results on datasets like CIFAR-10, CIFAR-100, and ImageNet.
Most modern convolutional neural networks (CNNs) used for object recognition are built using the same principles: Alternating convolution and max-pooling layers followed by a small number of fully connected layers. We re-evaluate the state of the art for object recognition from small images with convolutional networks, questioning the necessity of different components in the pipeline. We find that max-pooling can simply be replaced by a convolutional layer with increased stride without loss in accuracy on several image recognition benchmarks. Following this finding -- and building on other recent work for finding simple network structures -- we propose a new architecture that consists solely of convolutional layers and yields competitive or state of the art performance on several object recognition datasets (CIFAR-10, CIFAR-100, ImageNet). To analyze the network we introduce a new variant of the "deconvolution approach" for visualizing features learned by CNNs, which can be applied to a broader range of network structures than existing approaches.