论文标题
训练训练的交流的超平面布置有偏见
Hyperplane Arrangements of Trained ConvNets Are Biased
论文作者
论文摘要
我们通过对卷积层诱导的超平面布置进行经验研究，研究了训练有素的卷动杆在其卷积层的预催化空间中学到的功能的几何特性。我们介绍了有关训练有素网络的权重的统计数据，以研究本地安排并将其与培训动态联系起来。我们观察到，受过训练的交流表显示出对常规超平面构型的明显统计偏差。此外，我们发现显示出偏置配置的层对于在CIFAR10，CIFAR100和Imagenet上训练的架构的验证性能至关重要。
We investigate the geometric properties of the functions learned by trained ConvNets in the preactivation space of their convolutional layers, by performing an empirical study of hyperplane arrangements induced by a convolutional layer. We introduce statistics over the weights of a trained network to study local arrangements and relate them to the training dynamics. We observe that trained ConvNets show a significant statistical bias towards regular hyperplane configurations. Furthermore, we find that layers showing biased configurations are critical to validation performance for the architectures considered, trained on CIFAR10, CIFAR100 and ImageNet.
