Convolutional Networks with Dense Connectivity
arXiv:2001.02394
Abstract
Recent work has shown that convolutional networks can be substantially deeper, more accurate, and efficient to train if they contain shorter connections between layers close to the input and those close to the output. In this paper, we embrace this observation and introduce the Dense Convolutional Network (DenseNet), which connects each layer to every other layer in a feed-forward fashion.Whereas traditional convolutional networks with L layers have L connections - one between each layer and its subsequent layer - our network has L(L+1)/2 direct connections. For each layer, the feature-maps of all preceding layers are used as inputs, and its own feature-maps are used as inputs into all subsequent layers. DenseNets have several compelling advantages: they alleviate the vanishing-gradient problem, encourage feature reuse and substantially improve parameter efficiency. We evaluate our proposed architecture on four highly competitive object recognition benchmark tasks (CIFAR-10, CIFAR-100, SVHN, and ImageNet). DenseNets obtain significant improvements over the state-of-the-art on most of them, whilst requiring less parameters and computation to achieve high performance.
Journal(PAMI) version of DenseNet(CVPR'17)
References in corpus (12)
- Batch Normalization: Accelerating Deep Network Training by Reducing Internal Covariate Shift
- Striving for Simplicity: The All Convolutional Net
- Wide Residual Networks
- Training Very Deep Networks
- Resnet in Resnet: Generalizing Residual Architectures
- Semi-Supervised Learning with Ladder Networks
- Training Deep Nets with Sublinear Memory Cost
- Xception: Deep Learning with Depthwise Separable Convolutions
- Bridging the Gaps Between Residual Learning, Recurrent Neural Networks and Visual Cortex
- AdaNet: Adaptive Structural Learning of Artificial Neural Networks
- Deconstructing the Ladder Network Architecture
- Log-DenseNet: How to Sparsify a DenseNet