Classification and Ranking of Fermi Lat Gamma-Ray Sources From the 3fgl Catalog Using Machine Learning Techniques

Abstract

We apply a number of statistical and machine learning techniques to classify and rank gamma-ray sources from the Third Fermi Large Area Telescope Source Catalog (3FGL), according to their likelihood of falling into the two major classes of gamma-ray emitters: pulsars (PSR) or active galactic nuclei (AGNs). Using 1904 3FGL sources that have been identified/associated with AGNs (1738) and PSR (166), we train (using 70% of our sample) and test (using 30%) our algorithms and find that the best overall accuracy (>96%) is obtained with the Random Forest (RF) technique, while using a logistic regression (LR) algorithm results in only marginally lower accuracy. We apply the same techniques on a subsample of 142 known gamma-ray pulsars to classify them into two major subcategories: young (YNG) and millisecond pulsars (MSP). Once more, the RF algorithm has the best overall accuracy (∼90%), while a boosted LR analysis comes a close second. We apply our two best models (RF and LR) to the entire 3FGL catalog, providing predictions on the likely nature of unassociated sources, including the likely type of pulsar (YNG or MSP). We also use our predictions to shed light on the possible nature of some gamma-ray sources with known associations (e.g., binaries, supernova remnants/pulsar wind nebulae). Finally, we provide a list of plausible X-ray counterparts for some pulsar candidates, obtained using Swift, Chandra, and XMM. The results of our study will be of interest both for in-depth follow-up searches (e.g., pulsar) at various wavelengths and for broader population studies.

3fgl Demographics Outside the Galactic Plane Using Supervised Machine Learning: Pulsar and Dark Matter Subhalo Interpretations

“…Table 3 lists these candidates. To check for consistency, we compared our list of candidates with results from spectral fitting of 3FGL unassociated sources (Bertoni et al 2015), as well as with pulsar predictions using a combination of Random Forest and Logistic Regression (Saz Parkinson et al 2016). The list of Galactic candidates at high Galactic latitude is in good agreement with both of these works.…”

A Novel Framework for Gamma-ray Source Classification using Automatic Feature Selection

“…We evaluate our framework with data from the 3FGL Catalog [2] and compare our results with those based on the method in [1]. Our experiments are divided into two parts: the PSR/AGN classification and the YNG/MSP classification.…”

“…There are three main stages in our framework shown in Figure 1 which illustrates (i) the preprocessing stage, (ii) the feature selection stage and (iii) the classification stage. In our framework, RFs are used for feature selection without prior knowledge after the input data is cleaned and pre-processed in the same way as [1]. In the classification stage of our framework, boosted logistic regression (LR) with the features automatically selected in stage (ii) is used to build the prediction model.…”

“…RF has been widely used for classification in astronomy (e.g. [1][9]). Since RFs generally improve the performance of the classifiers and reduce overfitting of standard decision trees in the training set, we adopt it as the feature selection algorithm in our work.…”

“…[1][5] [6]). However, since the features were selected manually in these works, it is unlikely that the power of automatic classification has been fully exploited.…”