Connections Between Pairs of Filters Improve the Accuracy of Convolutional Neural Networks

滤波器对之间的连接提升了卷积神经网络的准确性

Abstract: While researchers continue to find new and improved network structures for CNNs, most of the newly invented architectures still rely on the traditional pattern of stacking convolutional blocks and separating them with pointwise activation functions.

摘要： 尽管研究人员不断为卷积神经网络（CNN）寻找更新、更优的网络结构，但大多数新发明的架构仍然依赖于传统的模式，即堆叠卷积块并使用逐点激活函数将它们分隔开。

However, there are drawbacks to a network purely building on pointwise nonlinearities. One alternative is to introduce a pairwise connection between two filters of a network. Typical connection functions use multiplications or the minimum operation to realize logical AND connections.

然而，仅基于逐点非线性构建的网络存在缺陷。一种替代方案是在网络的两个滤波器之间引入成对连接。典型的连接函数使用乘法或最小值运算来实现逻辑“与”（AND）连接。

In this paper, we go one step further by demonstrating that CNNs can benefit from more general connections, which include parameters that are learned. With such parameters, the network is able to implement different connections in different network layers and better adapt the connection function to the task at hand.

在本文中，我们更进一步，证明了 CNN 可以从更通用的连接中获益，这些连接包含了可学习的参数。通过这些参数，网络能够在不同的网络层中实现不同的连接，并使连接函数更好地适应当前的任务。