UltraPINK -- New possibilities to explore Self-Organizing Kohonen Maps

TL;DR

The paper tackles the challenge of exploring massive, heterogeneous astronomical datasets by combining unsupervised Self-Organizing Maps with manual exploration. It introduces UltraPINK, a datatype-agnostic, web-based frontend built around the PINK SOM framework, featuring an abstract interaction design, metadata support, and multiple inspection tools. Key contributions include a Django-based interface for project and dataset management, capabilities to view best-matching data points and outliers, labeling mechanisms, and two alternative map views, along with metadata integration via Astropy types and sky-position visualization with ALADIN. The work presents a functional prototype that aims to serve as a flexible infrastructure for astronomical data analysis and outlines planned extensions to support more data types, image manipulation, and multi-dimensional data, enabling broader interoperability with existing tools. Overall, UltraPINK demonstrates how datatype-agnostic, ML-assisted exploration can complement manual analysis to manage the complexity of modern astronomical data.

Abstract

Unsupervised learning algorithms like self-organizing Kohonen maps are a promising approach to gain an overview among massive datasets. With UltraPINK, researchers can train, inspect, and explore self-organizing maps, whereby the toolbox of interaction possibilities grows continually. Key feature of UltraPINK is the consideration of versality in astronomical data. By keeping the operations as abstract as possible and using design patterns meant for abstract usage, we ensure that data is compatible with UltraPINK, regardless of its type, formatting, or origin. Future work on the application will keep extending the catalogue of exploration tools and the interfaces towards other established applications to process astronomical data. Ultimatively, we aim towards a solid infrastructure for data analysis in astronomy.