6DCNN with roto-translational convolution filters for volumetric data processing

TL;DR

This paper introduces 6DCNN, a 6D convolutional neural network that effectively captures spatial patterns in volumetric data by leveraging roto-translational filters and Fourier space operations, improving recognition accuracy.

Contribution

The paper presents a novel 6D CNN architecture with SE(3)-equivariant message passing, reducing computational complexity and enhancing pattern recognition in 3D data.

Findings

Outperforms baseline models on CASP protein datasets

Achieves state-of-the-art results in 3D pattern recognition

Demonstrates efficiency of Fourier space operations in 6D convolutions

Abstract

In this work, we introduce 6D Convolutional Neural Network (6DCNN) designed to tackle the problem of detecting relative positions and orientations of local patterns when processing three-dimensional volumetric data. 6DCNN also includes SE(3)-equivariant message-passing and nonlinear activation operations constructed in the Fourier space. Working in the Fourier space allows significantly reducing the computational complexity of our operations. We demonstrate the properties of the 6D convolution and its efficiency in the recognition of spatial patterns. We also assess the 6DCNN model on several datasets from the recent CASP protein structure prediction challenges. Here, 6DCNN improves over the baseline architecture and also outperforms the state of the art.