Search

The identification of new white dwarfs (WDs) polluted with heavy elements is important since they provide a valuable tool for inferring chemical properties of putative planetary systems accreting material on their surfaces. The Gaia space mission has provided us with an unprecedented amount of astrometric, photometric, and low resolution (XP) spectroscopic data for millions of newly discovered stellar sources, among them thousands of WDs. In order to find WDs among this data and to identify which ones have metals in their atmospheres, we propose a methodology based on an unsupervised artificial intelligence technique called Self-Organizing Maps (SOM). In our approach a nonlinear high-dimensional dataset is projected on a 2D grid map where similar elements fall into the same neuron. By applying this method, we obtained a clean sample of 66,337 WDs. We performed an automatic spectral classification analysis to them, obtaining 143 bona fide polluted WD candidates not previously classified in the literature. The majority of them are cool WDs and we identify in their XP spectra several metallic lines such as Ca, Mg, Na, Li, and K. The fact that we obtain similar precision metrics than those achieved with recent supervised techniques highlights the power of our unsupervised approach to mine the Gaia archives for hidden treasures to follow-up spectroscopically with higher resolution., Comment: Accepted for publication in The Astrophysical Journal (ApJ). 15 pages, 13 figures

Accreting massive black holes (MBHs, with M$_\bullet > 10^3$ M$_{\odot}$) are known for their panchromatic emission, spanning from radio to gamma rays. While MBHs accreting at significant fractions of their Eddington rate are readily detectable, those accreting at much lower rates in radiatively inefficient modes often go unnoticed, blending in with other astrophysical sources. This challenge is particularly relevant for gas-starved MBHs in external galaxies and those possibly wandering in the Milky Way. We present HOLESOM, a machine learning-powered tool based on the Self-Organizing Maps (SOMs) algorithm, specifically designed to identify slowly-accreting MBHs using sparse photometric data. Trained on a comprehensive set of $\sim$ 20, 000 spectral energy distributions (SEDs), HOLESOM can (i) determine if the photometry of a source is consistent with slowly-accreting MBHs and (ii) estimate its black hole mass and Eddington ratio, including uncertainties. We validate HOLESOM through extensive tests on synthetic data and real-world cases, including Sagittarius A* (Sgr A*), demonstrating its effectiveness in identifying slowly-accreting MBHs. Additionally, we derive analytical relations between radio and X-ray luminosities to further constrain physical parameters. The primary strength of HOLESOM lies in its ability to accurately identify MBH candidates, which can then be targeted for follow-up photometric and spectroscopic observations. Fast and scalable, HOLESOM offers a robust framework for automatically scanning large multi-wavelength datasets, making it a valuable tool for unveiling hidden MBH populations in the local Universe., Comment: 17 pages, 11 figures, 2 tables, submitted to The Astrophysical Journal. Comments and feedback on this work are very welcome

Modern wide field radio surveys typically detect millions of objects. Techniques based on machine learning are proving to be useful for classifying large numbers of objects. The self-organizing map (SOM) is an unsupervised machine learning algorithm that projects a many-dimensional dataset onto a two- or three-dimensional lattice of neurons. This dimensionality reduction allows the user to visualize common features of the data better and develop algorithms for classifying objects that are not otherwise possible with large datasets. To this aim, we use the PINK implementation of a SOM. PINK incorporates rotation and flipping invariance so that the SOM algorithm may be applied to astronomical images. In this cookbook we provide instructions for working with PINK, including preprocessing the input images, training the model, and offering lessons learned through experimentation. The problem of imbalanced classes can be improved by careful selection of the training sample and increasing the number of neurons in the SOM (chosen by the user). Because PINK is not scale-invariant, structure can be smeared in the neurons. This can also be improved by increasing the number of neurons in the SOM. We also introduce pyink, a Python package used to read and write PINK binary files, assist in common preprocessing operations, perform standard analyses, visualize the SOM and preprocessed images, and create image-based annotations using a graphical interface. A tutorial is also provided to guide the user through the entire process. We present an application of PINK to VLA Sky Survey (VLASS) images. We demonstrate that the PINK is generally able to group VLASS sources with similar morphology together. We use the results of PINK to estimate the probability that a given source in the VLASS QuickLook Catalogue is actually due to sidelobe contamination., Comment: Accepted for publication in Astronomy and Computing

The current and upcoming large data volume galaxy surveys require the use of machine learning techniques to maximize their scientific return. This study explores the use of Self-Organizing Maps (SOMs) to estimate galaxy parameters with a focus on handling cases of missing data and providing realistic probability distribution functions for the parameters. We train a SOM with a simulated mass-limited lightcone assuming a ugrizYJHKs+IRAC dataset, mimicking the Hyper Suprime-Cam (HSC) Deep joint dataset. For parameter estimation, we derive SOM likelihood surfaces considering photometric errors to derive total (statistical and systematic) uncertainties. We explore the effects of missing data including which bands are particular critical to the accuracy of the derived parameters. We demonstrate that the parameter recovery is significantly better when the missing bands are "filled-in" rather than if they are completely omitted. We propose a practical method for such recovery of missing data., Comment: 22 pages, 12 figures, 1 table, accepted for publication in The Astronomical Journal

Identifying phases of flight is important in the field of general aviation, as knowing which phase of flight data is collected from aircraft flight data recorders can aid in the more effective detection of safety or hazardous events. General aviation flight data for phase of flight identification is usually per-second data, comes on a large scale, and is class imbalanced. It is expensive to manually label the data and training classification models usually faces class imbalance problems. This work investigates the use of a novel method for minimally supervised self-organizing maps (MS-SOMs) which utilize nearest neighbor majority votes in the SOM U-matrix for class estimation. Results show that the proposed method can reach or exceed a naive SOM approach which utilized a full data file of labeled data, with only 30 labeled datapoints per class. Additionally, the minimally supervised SOM is significantly more robust to the class imbalance of the phase of flight data. These results highlight how little data is required for effective phase of flight identification.

Parameter prediction is essential for many applications, facilitating insightful interpretation and decision-making. However, in many real life domains, such as power systems, medicine, and engineering, it can be very expensive to acquire ground truth labels for certain datasets as they may require extensive and expensive laboratory testing. In this work, we introduce a semi-supervised learning approach based on topological projections in self-organizing maps (SOMs), which significantly reduces the required number of labeled data points to perform parameter prediction, effectively exploiting information contained in large unlabeled datasets. Our proposed method first trains SOMs on unlabeled data and then a minimal number of available labeled data points are assigned to key best matching units (BMU). The values estimated for newly-encountered data points are computed utilizing the average of the $n$ closest labeled data points in the SOM's U-matrix in tandem with a topological shortest path distance calculation scheme. Our results indicate that the proposed minimally supervised model significantly outperforms traditional regression techniques, including linear and polynomial regression, Gaussian process regression, K-nearest neighbors, as well as deep neural network models and related clustering schemes.

An intelligent system capable of continual learning is one that can process and extract knowledge from potentially infinitely long streams of pattern vectors. The major challenge that makes crafting such a system difficult is known as catastrophic forgetting - an agent, such as one based on artificial neural networks (ANNs), struggles to retain previously acquired knowledge when learning from new samples. Furthermore, ensuring that knowledge is preserved for previous tasks becomes more challenging when input is not supplemented with task boundary information. Although forgetting in the context of ANNs has been studied extensively, there still exists far less work investigating it in terms of unsupervised architectures such as the venerable self-organizing map (SOM), a neural model often used in clustering and dimensionality reduction. While the internal mechanisms of SOMs could, in principle, yield sparse representations that improve memory retention, we observe that, when a fixed-size SOM processes continuous data streams, it experiences concept drift. In light of this, we propose a generalization of the SOM, the continual SOM (CSOM), which is capable of online unsupervised learning under a low memory budget. Our results, on benchmarks including MNIST, Kuzushiji-MNIST, and Fashion-MNIST, show almost a two times increase in accuracy, and CIFAR-10 demonstrates a state-of-the-art result when tested on (online) unsupervised class incremental learning setting.

In sufficiently complex tasks, it is expected that as a side effect of learning to solve a problem, a neural network will learn relevant abstractions of the representation of that problem. This has been confirmed in particular in machine vision where a number of works showed that correlations could be found between the activations of specific units (neurons) in a neural network and the visual concepts (textures, colors, objects) present in the image. Here, we explore the use of self-organizing maps as a way to both visually and computationally inspect how activation vectors of whole layers of neural networks correspond to neural representations of abstract concepts such as `female person' or `realist painter'. We experiment with multiple measures applied to those maps to assess the level of representation of a concept in a network's layer. We show that, among the measures tested, the relative entropy of the activation map for a concept compared to the map for the whole data is a suitable candidate and can be used as part of a methodology to identify and locate the neural representation of a concept, visualize it, and understand its importance in solving the prediction task at hand., Comment: Published in proceedings of K-CAP 2023

Leachates from garbage dumps can significantly compromise their surrounding area. Even if the distance between these and the populated areas could be considerable, the risk of affecting the aquifers for public use is imminent in most cases. For this reason, the delimitation and monitoring of the leachate plume are of significant importance. Geoelectric data (resistivity and IP), and surface methane measurements, are integrated and classified using an unsupervised Neural Network to identify possible risk zones in areas surrounding a landfill. The Neural Network used is a Kohonen type, which generates; as a result, Self-Organizing Classification Maps or SOM (Self-Organizing Map). Two graphic outputs were obtained from the training performed in which groups of neurons that presented a similar behaviour were selected. Contour maps corresponding to the location of these groups and the individual variables were generated to compare the classification obtained and the different anomalies associated with each of these variables. Two of the groups resulting from the classification are related to typical values of liquids percolated in the landfill for the parameters evaluated individually. In this way, a precise delimitation of the affected areas in the studied landfill was obtained, integrating the input variables via SOMs. The location of the study area is not detailed for confidentiality reasons., Comment: 11 pages, 7 figures

The growing amount of data produced by simulations and observations of space physics processes encourages the use of methods rooted in Machine Learning for data analysis and physical discovery. We apply a clustering method based on Self-Organizing Maps (SOM) to fully kinetic simulations of plasmoid instability, with the aim of assessing its suitability as a reliable analysis tool for both simulated and observed data. We obtain clusters that map well, a posteriori, to our knowledge of the process: the clusters clearly identify the inflow region, the inner plasmoid region, the separatrices, and regions associated with plasmoid merging. SOM-specific analysis tools, such as feature maps and Unified Distance Matrix, provide one with valuable insights into both the physics at work and specific spatial regions of interest. The method appears as a promising option for the analysis of data, both from simulations and from observations, and could also potentially be used to trigger the switch to different simulation models or resolution in coupled codes for space simulations.