Band-limited Training and Inference for Convolutional Neural Networks
arXiv:1911.09287
Abstract
The convolutional layers are core building blocks of neural network architectures. In general, a convolutional filter applies to the entire frequency spectrum of the input data. We explore artificially constraining the frequency spectra of these filters and data, called band-limiting, during training. The frequency domain constraints apply to both the feed-forward and back-propagation steps. Experimentally, we observe that Convolutional Neural Networks (CNNs) are resilient to this compression scheme and results suggest that CNNs learn to leverage lower-frequency components. In particular, we found: (1) band-limited training can effectively control the resource usage (GPU and memory); (2) models trained with band-limited layers retain high prediction accuracy; and (3) requires no modification to existing training algorithms or neural network architectures to use unlike other compression schemes.
arXiv commentPublished at International Conference on Machine Learning (ICML)
Cited by in corpus (7)
- A Fourier Perspective on Model Robustness in Computer Vision
- SSD-GAN: Measuring the Realness in the Spatial and Spectral Domains
- Band-limited Soft Actor Critic Model
- Analysis of Random Perturbations for Robust Convolutional Neural Networks
- Machine Learning enabled Spectrum Sharing in Dense LTE-U/Wi-Fi Coexistence Scenarios
- Robust Learning with Frequency Domain Regularization
- Progressive Compressed Records: Taking a Byte out of Deep Learning Data