Kervolutional Neural Networks

Abstract

Convolutional neural networks (CNNs) have enabled the state-of-the-art performance in many computer vision tasks. However, little effort has been devoted to establishing convolution in non-linear space. Existing works mainly leverage on the activation layers, which can only provide point-wise non-linearity. To solve this problem, a new operation, kervolution (kernel convolution), is introduced to approximate complex behaviors of human perception systems leveraging on the kernel trick. It generalizes convolution, enhances the model capacity, and captures higher order interactions of features, via patch-wise kernel functions, but without introducing additional parameters. Extensive experiments show that kervolutional neural networks (KNN) achieve higher accuracy and faster convergence than baseline CNN.

arXiv commentoral paper in CVPR 2019

References in corpus (3)

• High-Speed Tracking with Kernelized Correlation Filters
• Ultra-Wideband Aided Fast Localization and Mapping System
• Non-Iterative SLAM

Cited by in corpus (7)

• Graph Optimization Approach to Range-based Localization
• PyPose: A Library for Robot Learning with Physics-based Optimization
• Online Visual Place Recognition via Saliency Re-identification
• Visual Memorability for Robotic Interestingness via Unsupervised Online Learning
• Unsupervised Online Learning for Robotic Interestingness with Visual Memory
• AirCode: A Robust Object Encoding Method
• Polynomial degree reduction in the ${\mathcal{L}}^2$-norm on a symmetric interval for the canonical basis