Research map / Architectures and efficiency / CNN design

kVNN: Learnable Volterra Network Kernels

NeurIPS 2026 · Architectures and efficiency › CNN design · approach: Large and dynamic kernel design

Read the paperCheck your idea against 108,967 papers

AI-extracted summary. Written by AI from the paper's abstract only. Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Proposes kVNN, a kernelized Volterra neural network with learnable multi-kernel representations to model higher-order nonlinear interactions in deep learning.

Problem

Modeling higher-order interactions among variables, scales, or modalities in large-scale deep learning is complex; existing approaches struggle to capture nonlinear interactions beyond pairwise relations.

Approach

A kernelized Volterra Neural Network whose Volterra kernels are learnable multi-kernel representations, enabling explicit higher-order nonlinear interaction modeling within a neural architecture.

Experimental setup

Builds on: Volterra series

Closest papers by meaning

• Conquering the CNN Over-Parameterization Dilemma: A Volterra Filtering Approach for Action Recognition (AAAI 2020 · similarity 0.87)
• Is Kernel Prediction More Powerful than Gating in Convolutional Neural Networks? (ICML 2024 · similarity 0.85)
• Three Mechanisms of Feature Learning in a Linear Network (ICLR 2025 · similarity 0.84)
• Kernel Learning with Adversarial Features: Numerical Efficiency and Adaptive Regularization (NeurIPS 2025 · similarity 0.83)
• From Kernels to Features: A Multi-Scale Adaptive Theory of Feature Learning (ICML 2025 · similarity 0.83)
• Scalable Neural Network Kernels (ICLR 2024 · similarity 0.83)