Dynamic Convolution

Follow this topic Ask about this topic

Dynamic convolution studies how neural networks can generate or select convolution filters from input content, rather than applying the same weights everywhere. Common approaches use attention or gating to mix kernel banks, generate local or global context-conditioned filters, or build adaptive filters from structured bases and geometric or frequency cues. These methods are used in image classification, medical segmentation, image fusion, forecasting, speech processing, and language models, where they often improve on fixed convolutions. A central tension is balancing fine-grained adaptation and robustness across inputs or settings against the extra computation and complexity of producing dynamic filters.

Latest papers

16 papers1 paper in the last four weeks, with none the four weeks before. 0.0% of all new papers.

MCL: Meta Convolution Layer
Oct 8, 2026Naim Reza, Md Al Amin, Ho Yub JungDynamic Convolution Convolutional Neural Networks
Department of Computer Engineering, Chosun University, Gwangju 61452, Republic of Korea
Core-KAN: Continuous Vision Kernels with Kolmogorov-Arnold Networks
Aug 20, 2026Lan Guo, Mengling Li, Haoran Li +4Convolutional Neural Networks Dynamic Convolution
School of Information Science and Engineering, Lanzhou University · Department of Data Science and AI, Monash University · School of Computing and Information Technology, University of Wollongong
Efficient Noise Context-Aware Representation for Low-Dose CT Denoising
Aug 11, 2026Minwoo Yu, N. Robert Bennett, Jongduk Baek +1Dynamic Convolution Computed Tomography
Department of Artificial Intelligence, Yonsei University, South Korea · Department of Radiology, Stanford University, Stanford, CA 94305 USA · Department of Electrical Engineering, Stanford University, Stanford, CA 94305 USA
Towards Array-Invariant Speech Enhancement via Geometry-Aware Dynamic Convolution
Jul 21, 2026Zhenglong Liu, Wangyou Zhang, Chenda Li +1Dynamic Convolution Speech Enhancement
Auditory Cognition and Computational Acoustics Lab Shanghai Jiao Tong University, Shanghai, China · VUI Labs
Seeing Through WiFi: Lightweight Human Pose Estimation with Dynamic Kernel Attention
Jul 3, 2026Toan D. Gian, Van-Dinh Nguyen, Vo Phi Son +5Dynamic Convolution Channel Attention
Smart Green Transformation Center (GREEN-X), VinUniversity, Vietnam · Trinity College Dublin, Ireland · University of Oulu, Finland +3
CNN Models for Microphone Array Covariance Matrix Upsampling and Acoustic Imaging
Jul 1, 2026Marianthi Adamopoulou, Parthasaarathy Sudarsanam, David Diaz-Guerra +5Dynamic Convolution Convolutional Neural Networks
Dept. of Comp. and Elec. Engineering Mid Sweden University Sundsvall, Sweden · Audio Research Group, Tampere University Tampere, Finland
Dynamic Short Convolutions Improve Transformers
Jun 2, 2026Oliver Sieberling, Bharat Runwal, Rameswar Panda +1Dynamic Convolution Transformer
Massachusetts Institute of Technology · MIT-IBM Watson AI Lab
Synthetic Aperture Radar Image Change Detection Based on Global Dynamic Context-Aware Network
May 16, 2026Baogui Huan, Chuanzheng Gong, Dezhong Chen +3Dynamic Convolution Data Augmentation
State Key Laboratory of Physical Oceanography, Ocean University of China, Qingdao 266100, China · Department of Electrical and Computer Engineering, Mississippi State University, Starkville, MS 39762 USA
RAFNet: Region-Aware Fusion Network for Pansharpening
May 4, 2026Jianing Zhang, Zijian Zhou, Kai SunDynamic Convolution Frequency-Domain Feature Learning
College of Artificial Intelligence, Xi’an Jiaotong University, Xi’an 710049, China · School of Mathematics and Statistics, Xi’an Jiaotong University, Xi’an 710049, China
EAPFusion: Intrinsic Evolving Auxiliary Prior Guidance for Infrared and Visible Image Fusion
May 3, 2026Zhenyu Sun, Luobin Zhang, Axi Niu +2Dynamic Convolution Multi-Source Image Fusion
Northwestern Polytechnical University, Xi’an, China
Beyond Convolution: A Taxonomy of Structured Operators for Learning-Based Image Processing
Mar 12, 2026Simone CammarasanaConvolutional Neural Networks Dynamic Convolution
CNR-IMATI, Via De Marini 6, Genova, Italy
SGDC: Structurally-Guided Dynamic Convolution for Medical Image Segmentation
Feb 26, 2026Bo Shi, Wei-ping Zhu, M. N. S. SwamyMedical Image Segmentation Dynamic Convolution
Department of Electrical and Computer Engineering, Concordia University
Cross-Task Benchmarking of CNN Architectures
Feb 25, 2026Kamal Sherawat, Vikrant BhatiConvolutional Neural Networks Dynamic Convolution
Virginia Tech, ECE 6524 - Deep Learning
PFGNet: A Fully Convolutional Frequency-Guided Peripheral Gating Network for Efficient Spatiotemporal Predictive Learning
Feb 24, 2026Xinyong Cai, Changbin Sun, Yong Wang +2Spatiotemporal Forecasting Video Prediction
College of Computer Science, Sichuan University · Department of Data and Systems Engineering, The University of Hong Kong
Dynamic High-frequency Convolution for Infrared Small Target Detection
Feb 3, 2026Ruojing Li, Chao Xiao, Qian Yin +5Dynamic Convolution Frequency-Domain Feature Learning
The authors are with the College of Electronic Science and Technology, National University of Defense Technology, Changsha 410073, China.
Conditional Clifford-Steerable CNNs for PDE Modeling
Oct 15, 2025Bálint László Szarvas, Maksim ZhdanovDynamic Convolution PDE Surrogate Modeling
University of Amsterdam · 2AMLab.

Oct 8, 2026J/K move · Enter open · S save

MCL: Meta Convolution Layer

Naim Reza, Md Al Amin, Ho Yub Jung

Department of Computer Engineering, Chosun University, Gwangju 61452, Republic of Korea

Dynamic Convolution Convolutional Neural Networks Image Classification

Open paper AskPDF arXiv

Dynamic convolution enhances convolutional neural networks (CNNs) by adapting kernels to input content, but it expresses the effective kernel as a linear mixture of a small number of basis kernels, which limits expressivity and complicates optimization as the mixture size grows. In this work, we revisit dynamic convolution from a functional perspective and propose the Meta Convolution Layer (MCL), which directly models the convolutional kernel as an input-conditioned function W(x) realized via a high-order polynomial expansion. Leveraging nested residual blocks inspired by deep polynomial networks, MCL implements a structured polynomial meta-network that generates a single input-adaptive kernel, thereby decoupling representational power from the explicit number of mixture kernels and alleviating training instability. MCL is a plug-in addition with standard convolutions and can be seamlessly integrated into both CNN and transformer backbones. Experimental evaluation shows that adding MCL improves the Top-1 accuracy of Resnet- 18, Resnet-50 and ResNet-101 by 6.61%, 3.42% and 3.05% on the ImageNet dataset. Moreover, the proposed method significantly boosts the accuracy of Resnet and Wide-Resnet variants on CIFAR-10 and CIFAR-100 datasets. Additionally, the proposed method outperforms previous methods on fine-grained visual classification tasks using Swin and ViT backbones. These results demonstrate that high-order polynomial kernel generation is a powerful and scalable alternative to linear mixture based dynamic convolution.