Compressing the Input for CNNs with the First-Order Scattering Transform

Abstract

We study the first-order scattering transform as a candidate for reducing the signal processed by a convolutional neural network (CNN). We show theoretical and empirical evidence that in the case of natural images and sufficiently small translation invariance, this transform preserves most of the signal information needed for classification while substantially reducing the spatial resolution and total signal size. We demonstrate that cascading a CNN with this representation performs on par with ImageNet classification models, commonly used in downstream tasks, such as the ResNet-50. We subsequently apply our trained hybrid ImageNet model as a base model on a detection system, which has typically larger image inputs. On Pascal VOC and COCO detection tasks we demonstrate improvements in the inference speed and training memory consumption compared to models trained directly on the input image.

References in corpus (5)

• Very Deep Convolutional Networks for Large-Scale Image Recognition
• Spectral Representations for Convolutional Neural Networks
• Towards Image Understanding from Deep Compression without Decoding
• Wavelet Convolutional Neural Networks for Texture Classification
• Audio Texture Synthesis with Scattering Moments

Cited by in corpus (7)

• Joint Time-Frequency Scattering
• Online Learned Continual Compression with Adaptive Quantization Modules
• Efficient Visual Recognition with Deep Neural Networks: A Survey on Recent Advances and New Directions
• Multichannel Orthogonal Transform-Based Perceptron Layers for Efficient ResNets
• Harmonic Networks: Integrating Spectral Information into CNNs
• The Unreasonable Effectiveness of Patches in Deep Convolutional Kernels Methods
• Interferometric Graph Transform for Community Labeling