论文标题
人工神经网络的自适应卷积内核
Adaptive Convolution Kernel for Artificial Neural Networks
论文作者
论文摘要
通过使用固定和单个尺寸的堆叠卷积层（通常3 $ \ times $ 3）内核来构建许多深神网络。本文介绍了一种训练卷积内核大小的方法，以单层提供不同尺寸的内核。该方法利用了可以在基本网格中生长或收缩的可区分的，因此可逆转的高斯信封。我们的实验将所提出的自适应层与简单的两层网络，更深的残留网络和U-NET体系结构中的普通卷积层进行了比较。流行的图像分类数据集（例如MNIST，MNIST杂交，CIFAR-10，时尚和``野外面）''中的结果表明，自适应内核可以对普通卷积内核提供统计上显着的改进。 Oxford-Pets数据集中的分割实验表明，在具有单个7 $ \ times $ 7自适应层的U形网络中替换单个普通卷积层可以提高其学习成绩和推广能力。
Many deep neural networks are built by using stacked convolutional layers of fixed and single size (often 3$\times$3) kernels. This paper describes a method for training the size of convolutional kernels to provide varying size kernels in a single layer. The method utilizes a differentiable, and therefore backpropagation-trainable Gaussian envelope which can grow or shrink in a base grid. Our experiments compared the proposed adaptive layers to ordinary convolution layers in a simple two-layer network, a deeper residual network, and a U-Net architecture. The results in the popular image classification datasets such as MNIST, MNIST-CLUTTERED, CIFAR-10, Fashion, and ``Faces in the Wild'' showed that the adaptive kernels can provide statistically significant improvements on ordinary convolution kernels. A segmentation experiment in the Oxford-Pets dataset demonstrated that replacing a single ordinary convolution layer in a U-shaped network with a single 7$\times$7 adaptive layer can improve its learning performance and ability to generalize.
