The learning dynamics of deep neural networks are not well understood. The information bottleneck (IB) theory proclaimed separate fitting and compression phases. But they have since been heavily debated. We comprehensively analyze the learning dynamics by investigating a layer's reconstruction ability of the input and prediction performance based on the evolution of parameters during training. We empirically show the existence of three phases using common datasets and architectures such as ResNet and VGG: (i) near constant reconstruction loss, (ii) decrease, and (iii) increase. We also derive an empirically grounded data model and prove the existence of phases for single-layer networks. Technically, our approach leverages classical complexity analysis. It differs from IB by relying on measuring reconstruction loss rather than information theoretic measures to relate information of intermediate layers and inputs. Our work implies a new best practice for transfer learning: We show empirically that the pre-training of a classifier should stop well before its performance is optimal.

通过对参数的演化，我们全面分析了深度神经网络的学习动态，发现存在三个阶段：接近恒定的重建损失、下降和上升。我们还通过经验实证建立了数据模型，并对单层神经网络证明了阶段的存在。我们的工作为迁移学习提供了新的最佳实践：通过实验证明预训练的分类器在性能达到最优之前应该停止。

理解和利用神经网络的学习阶段