Unsupervised learning using topological data augmentation
Oleksandr Balabanov, Mats Granath
Abstract
The paper applies the concept of data augmentation to the study of topological states of matter. Because of the rigorous mathematical structure of topology, the authors show that data augmentation based on continuous deformations can be a powerful procedure for analyzing topological features and extracting topological indices using machine learning.
Topics & Concepts
Topological data analysisComputer scienceUnsupervised learningArtificial intelligenceTopology (electrical circuits)MathematicsTraining setPattern recognition (psychology)Noisy dataData pointMachine learningTheoretical computer scienceSynthetic dataSequence (biology)Topological Materials and PhenomenaQuantum many-body systemsSurface and Thin Film Phenomena