Towards Understanding Normalization in Neural ODEs
This work addresses a gap in understanding normalization for neural ODEs, which is incremental but provides specific performance gains for researchers in this domain.
The paper tackled the problem of understanding how normalization techniques affect neural ODEs, achieving 93% accuracy on CIFAR-10, which is the highest reported accuracy for neural ODEs on this task.
Normalization is an important and vastly investigated technique in deep learning. However, its role for Ordinary Differential Equation based networks (neural ODEs) is still poorly understood. This paper investigates how different normalization techniques affect the performance of neural ODEs. Particularly, we show that it is possible to achieve 93% accuracy in the CIFAR-10 classification task, and to the best of our knowledge, this is the highest reported accuracy among neural ODEs tested on this problem.