Nonparametric Topological Layers in Neural Networks
This addresses a bottleneck in adopting topological methods for neural networks, potentially improving efficiency and performance for researchers in topological data analysis and machine learning.
The paper tackles the problem of constructing topological layers in neural networks that rely on Euclidean coordinate systems, which are time-consuming to parametrize and lack theoretical foundations, leading to suboptimal performance. It proposes a learnable topological layer that operates in a general metric space (Hilbert space), eliminating hyperparameters and costly parametrization.
Various topological techniques and tools have been applied to neural networks in terms of network complexity, explainability, and performance. One fundamental assumption of this line of research is the existence of a global (Euclidean) coordinate system upon which the topological layer is constructed. Despite promising results, such a \textit{topologization} method has yet to be widely adopted because the parametrization of a topologization layer takes a considerable amount of time and more importantly, lacks a theoretical foundation without which the performance of the neural network only achieves suboptimal performance. This paper proposes a learnable topological layer for neural networks without requiring a Euclidean space; Instead, the proposed construction requires nothing more than a general metric space except for an inner product, i.e., a Hilbert space. Accordingly, the according parametrization for the proposed topological layer is free of user-specified hyperparameters, which precludes the costly parametrization stage and the corresponding possibility of suboptimal networks.