Euclidean Neural Network

Euclidean neural networks (ENNs) are a class of machine learning models traditionally designed for data residing in Euclidean space, but recent research increasingly addresses their limitations when dealing with data exhibiting non-Euclidean geometries, such as those found in graphs, manifolds, or hyperbolic spaces. Current research focuses on adapting ENNs to these non-Euclidean settings through techniques like Riemannian geometry, hyperbolic geometry, and the development of specialized architectures (e.g., Riemannian ResNets, hyperbolic collaborative filtering) that incorporate geometric properties into the network's structure and operations. This work is significant because it expands the applicability of neural networks to a wider range of complex data types, improving performance in various fields including computer vision, drug discovery, and recommendation systems.