Avoid representational bottlenecks, especially early in the network.(简单说就是feature map的大小要慢慢的减少。)网络
Higher dimensional representations are easier to process locally within a network. Increasing the activations per tile in a convolutional network allows for more disentangled features. The resulting networks will train faster.(在网络较深层应该利用更多的feature map,有利于容纳更多的分解特征。这样能够加速训练)性能
Spatial aggregation can be done over lower dimensional embeddings without much or any loss in representational power.(也就是bottleneck layer的设计)lua
Balance the width and depth of the network.(Increasing both the width and the depth of the network can contribute to higher quality networks.同时增长网络的深度和宽度)spa