Topological data analysis and machine learning

Daniel Leykam,Dimitris G. Angelakis
DOI: https://doi.org/10.1080/23746149.2023.2202331
2023-07-25
Abstract:Topological data analysis refers to approaches for systematically and reliably computing abstract ``shapes'' of complex data sets. There are various applications of topological data analysis in life and data sciences, with growing interest among physicists. We present a concise yet (we hope) comprehensive review of applications of topological data analysis to physics and machine learning problems in physics including the detection of phase transitions. We finish with a preview of anticipated directions for future research.
Mesoscale and Nanoscale Physics,Optics,Quantum Physics
What problem does this paper attempt to address?
The paper primarily discusses Topological Data Analysis (TDA) and its applications in the fields of physics and machine learning. Its core objectives are: 1. **Introduction to the basic concepts and techniques of TDA**: The paper first introduces the fundamental ideas of TDA, which involve systematically defining and studying the "shape" of data sets from complex data collections. This is achieved by constructing graphs in graph theory or higher-dimensional simplicial complexes to quantify these shapes, and by calculating the Betti numbers of these structures to capture the number of holes in different dimensions. 2. **Applications of TDA in physical problems**: Subsequently, the paper outlines the role of TDA in identifying different phase transitions of matter and detecting novel states of matter. Specifically, it can help physicists discover and characterize novel phenomena emerging in complex systems, such as unsupervised detection of phase transitions. 3. **Combining TDA with machine learning techniques**: The paper also explores methods of integrating TDA into larger machine learning pipelines for analyzing the characteristics of physical systems. This includes using features extracted by TDA as inputs for machine learning algorithms to classify and predict physical phenomena. In summary, the key issues this paper attempts to address are: - How to effectively apply TDA to analyze and understand the structural characteristics of complex physical systems; - How to combine TDA with modern machine learning techniques to enhance the understanding and predictive capabilities of physical phenomena; - Exploring the potential future applications of TDA in physics research, including the application of novel TDA methods. Through these contents, the paper aims to provide readers with a comprehensive overview of the latest advancements in TDA within the fields of physics and machine learning, and to guide future related research.