Work about novel method for topological data analysis will be published at the International Joint Conference on Artificial Intelligence
Researchers from Jagiellonian University, Swansea University and the St. Pölten University of Applied Sciences present a novel method for topological data analysis.
Topological data analysis (TDA) is a powerful framework for the analysis of different types of data, e.g. images, 3D object shapes, motion tracking data, and social media graphs. TDA enables the extraction of robust intrinsic data properties, which help to better understand the data. Especially in the case of noisy data, TDA is a promising approach for the extraction of compact and robust descriptions.
One limitation of TDA is that the obtained data descriptions are difficult to combine with today’s machine learning techniques. This hinders the development of artificial intelligence (AI) techniques based on TDA.
Novel method for TDA
The novelty of the proposed approach is that it makes it possible to efficiently combine TDA with arbitrary machine learning methods. It thereby opens the door for novel AI solutions based on TDA.
This was made possible by combining the so called ”bag-of-words” approach, which was originally developed for content-based text and image retrieval, with numerical representations obtained by TDA. The developed approach is not only much faster than related approaches but also partly outperforms related approaches in terms of machine learning performance.
The work will be published at the International Joint Conference on Artificial Intelligence – one of the premier (A*) conferences in the area of Artificial Intelligence.
A preprint of the publication can be found here.