Quasi-Conformal Convolution : A Learnable Convolution for Deep Learning on Riemann Surfaces
This addresses the problem of analyzing complex geometric data for researchers and practitioners in fields like computer vision and medical imaging, offering a novel framework that is not purely incremental but builds on existing theories.
The paper tackles the challenge of defining convolution on non-Euclidean domains like Riemann surfaces by introducing Quasi-conformal Convolution (QCC), which uses trainable quasi-conformal mappings to adapt convolution operators, and demonstrates superior performance in tasks such as image classification on curvilinear surfaces and medical applications like craniofacial analysis and lesion segmentation.
Deep learning on non-Euclidean domains is important for analyzing complex geometric data that lacks common coordinate systems and familiar Euclidean properties. A central challenge in this field is to define convolution on domains, which inherently possess irregular and non-Euclidean structures. In this work, we introduce Quasi-conformal Convolution (QCC), a novel framework for defining convolution on Riemann surfaces using quasi-conformal theories. Each QCC operator is linked to a specific quasi-conformal mapping, enabling the adjustment of the convolution operation through manipulation of this mapping. By utilizing trainable estimator modules that produce Quasi-conformal mappings, QCC facilitates adaptive and learnable convolution operators that can be dynamically adjusted according to the underlying data structured on Riemann surfaces. QCC unifies a broad range of spatially defined convolutions, facilitating the learning of tailored convolution operators on each underlying surface optimized for specific tasks. Building on this foundation, we develop the Quasi-Conformal Convolutional Neural Network (QCCNN) to address a variety of tasks related to geometric data. We validate the efficacy of QCCNN through the classification of images defined on curvilinear Riemann surfaces, demonstrating superior performance in this context. Additionally, we explore its potential in medical applications, including craniofacial analysis using 3D facial data and lesion segmentation on 3D human faces, achieving enhanced accuracy and reliability.