A search for dark matter among Fermi-LAT unidentified sources with systematic features in machine learning

Abstract

Around one third of the point-like sources in the Fermi-LAT catalogs remain as unidentified sources (unIDs) today. Indeed, these unIDs lack a clear, univocal association with a known astrophysical source. If dark matter (DM) is composed of weakly interacting massive particles (WIMPs), there is the exciting possibility that some of these unIDs may actually be DM sources, emitting gamma rays from WIMPs annihilation. We propose a new approach to solve the standard, Machine Learning (ML) binary classification problem of disentangling prospective DM sources (simulated data) from astrophysical sources (observed data) among the unIDs of the 4FGL Fermi-LAT catalogue. We artificially build two systematic features for the DM data which are originally inherent to observed data: the detection significance and the uncertainty on the spectral curvature. We do it by sampling from the observed population of unIDs, assuming that the DM distributions would, if any, follow the latter. We consider different ML models: Logistic Regression, Neural Network (NN), Naive Bayes and Gaussian Process, out of which the best, in terms of classification accuracy, is the NN, achieving around $93.3\% \pm 0.7\%$ performance. Other ML evaluation parameters, such as the True Negative and True Positive rates, are discussed in our work. Applying the NN to the unIDs sample, we find that the degeneracy between some astrophysical and DM sources can be partially solved within this methodology. Nonetheless, we conclude that there are no DM source candidates among the pool of 4FGL Fermi-LAT unIDs.

Classification of Blazar Candidates of Unknown Type in Fermi 4LAC by Unanimous Voting from Multiple Machine-learning Algorithms

“…Following Gammaldi et al (2023), we did a five-fold CV with 20 repetitions on the data set. In each iteration, the test set consists of 401 sources having 267 BL Lac objects and 134 FSRQs.…”

“…First, we split our final data set into two parts-the training set and the test set. Following Gammaldi et al (2023), we performed a five-fold CV with 20 repetitions on the data set. In each iteration, we took 80% of data for the training set, and the remaining 20% was kept aside as a test set.…”

Searching for dark matter subhalos in the Fermi-LAT catalog with Bayesian neural networks

“…By processing complete spectra, we can leverage the ability of neural networks to extract all relevant information from low-level features. Our work thus differs from previous studies, such as [37], which used synthetic features to search for dark matter subhalo candidates.…”

“…In the recent ref. [37], the authors base their classification on the spectral features expected from a generic dark matter spectrum [15], and artificially build systematic features sampling from the real observed sources to improve the performance of the network. Constraints on dark matter properties based on subhalo searches usually formulate their results as a function of the possible number of candidates among UNID sources [14,27].…”

Classification of Fermi-LAT unidentified gamma-ray sources using catboost gradient boosting decision trees

“…Nevertheless, in principle none of these exotic sources are expected within the AGN-like sample, as DM presents a highly curved spectrum, which would be classified as PSR-like Mirabal (2013). Such sources can also be studied with ML techniques Gammaldi et al (2023).…”