Kernel Convolution

Kernel convolution, a fundamental operation in convolutional neural networks (CNNs), is being actively refined to improve performance across diverse applications. Current research focuses on adapting kernel behavior, including dynamically adjusting kernel size and shape based on input features (e.g., frequency in sound processing, or distortion in fisheye images), and learning adaptive kernel weights to selectively emphasize relevant information. These advancements enhance CNN capabilities in areas such as image denoising, medical image segmentation, and graph-structured data analysis, leading to improved accuracy and efficiency compared to traditional methods.

Papers

June 8, 2024

Diversifying and Expanding Frequency-Adaptive Convolution Kernels for Sound Event Detection
Hyeonuk Nam, Seong-Hu Kim, Deokki Min, Junhyeok Lee, Yong-Hwa Park
Sound Event Detection Polyphonic Sound Cosine Kernel Kernel Convolution

February 2, 2024

Convolution kernel adaptation to calibrated fisheye
Bruno Berenguel-Baeta, Maria Santos-Villafranca, Jesus Bermudez-Cameo, Alejandro Perez-Yus, Jose J. Guerrero
Convolutional Neural Network Convolution Kernel Fisheye Camera Perspective Image Kernel Convolution

January 22, 2024

Learning to Approximate Adaptive Kernel Convolution on Graphs
Jaeyoon Sim, Sooyeon Jeon, InJun Choi, Guorong Wu, Won Hwa Kim
LeArning Abstract Graph Drawing Graph Convolution Diffusion Learning Diffusion Kernel Kernel Convolution

February 10, 2023

A deep convolutional neural network for salt-and-pepper noise removal using selective convolutional blocks
Ahmad Ali Rafiee, Mahmoud Farhang
Convolutional Neural Network Deep Convolutional Neural Network Noise Removal Kernel Convolution

February 7, 2023

Linear-scaling kernels for protein sequences and small molecules outperform deep learning while providing uncertainty quantitation and improved interpretability
Jonathan Parkinson, Wei Wang
Deep Learning Inherent Interpretability Gaussian Process Protein Sequence Small Molecule Linear Kernel Kernel Convolution

November 5, 2022

ESKNet-An enhanced adaptive selection kernel convolution for breast tumors segmentation
Gongping Chen, Lu Zhou, Jianxun Zhang, Xiaotao Yin, Liang Cui, Yu Dai
Breast Ultrasound Breast Tumor Segmentation Kernel Convolution

Kernel Convolution

Papers

Diversifying and Expanding Frequency-Adaptive Convolution Kernels for Sound Event Detection

Convolution kernel adaptation to calibrated fisheye

Learning to Approximate Adaptive Kernel Convolution on Graphs

A deep convolutional neural network for salt-and-pepper noise removal using selective convolutional blocks

Linear-scaling kernels for protein sequences and small molecules outperform deep learning while providing uncertainty quantitation and improved interpretability

ESKNet-An enhanced adaptive selection kernel convolution for breast tumors segmentation