Analyzing γ rays of the Galactic Center with deep learning

Caron, S.; Gomez-Vargas, G. A.; Hendriks, Luc; Austri, Roberto Ruiz de

doi:10.1088/1475-7516/2018/05/058

Cited by 34 publications

(39 citation statements)

References 63 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…At the moment, several explanations of the GC exist, including DM annihilation [1,6,7,8] and a contribution of unresolved point sources [27,28,29], such as millisecond pulsars (MSPs)…”

Section: Discussionmentioning

confidence: 99%

Diffuse gamma-ray emission modeling and the Galactic center GeV excess

Malyshev¹

2017

Proceedings of 7th International Fermi Symposium — PoS(IFS2017)

View full text Add to dashboard Cite

Excess of gamma rays around the Galactic center (GC) observed in the Fermi Large Area Telescope (LAT) data is one of the most intriguing features in the gamma-ray sky. The spherical morphology and the spectral energy distribution with a peak around a few GeV are consistent with emission from annihilation of dark matter particles. Other possible explanations include a distribution of millisecond pulsars (MSPs). One of the caveats of the MSP hypothesis is the relatively small number of associated MSPs near the GC. In this paper we perform a multi-class classification of Fermi-LAT sources using machine learning and determine the contribution from unassociated MSP-like sources near the GC. The spectral energy distribution, spatial morphology, and the source count distribution are consistent with expectations for a population of MSPs that can explain the gamma-ray excess. Possible caveats of the contribution from the unassociated MSP-like sources are discussed.

show abstract

“…At the moment, several explanations of the GC exist, including DM annihilation [1,6,7,8] and a contribution of unresolved point sources [27,28,29], such as millisecond pulsars (MSPs)…”

Section: Discussionmentioning

confidence: 99%

Diffuse gamma-ray emission modeling and the Galactic center GeV excess

Malyshev¹

2017

Proceedings of 7th International Fermi Symposium — PoS(IFS2017)

View full text Add to dashboard Cite

show abstract

“…1, andΣ −1 i,j is the inverse of the covariance matrix, which was derived in ref. [33]. Note that the derived information on the GCE spectrum in ref.…”

Section: )mentioning

confidence: 99%

“…At lower energies (E < 10 GeV) the GCE might be due to DM annihilation, unresolved millisecond pulsars (MSP), or a combination of both, see for instance refs. [17,33]. According to ref.…”

Section: Introductionmentioning

confidence: 99%

Extended Higgs-portal dark matter and the Fermi-LAT Galactic Center Excess

Casas

Vargas

Moreno

et al. 2018

J. Cosmol. Astropart. Phys.

View full text Add to dashboard Cite

In the present work, we show that the Galactic Center Excess (GCE) emission, as recently updated by the Fermi-LAT Collaboration, could be explained by a mixture of Fermi-bubbles-like emission plus dark matter (DM) annihilation, in the context of a scalar-singlet Higgs portal scenario (SHP). In fact, the standard SHP, where the DM particle, S, only has renormalizable interactions with the Higgs, is non-operational due to strong constraints, especially from DM direct detection limits. Thus we consider the most economical extension, called ESHP (for extended SHP), which consists solely in the addition of a second (more massive) scalar singlet in the dark sector. The second scalar can be integrated-out, leaving a standard SHP plus a dimension-6 operator. Mainly, this model has only two relevant parameters (the DM mass and the coupling of the dim-6 operator). DM annihilation occurs mainly into two Higgs bosons, SS → hh. We demonstrate that, despite its economy, the ESHP model provides an excellent fit to the GCE (with p-value ∼ 0.6−0.7) for very reasonable values of the parameters, in particular, m S 130 GeV. This agreement of the DM candidate to the GCE properties does not clash with other observables and keep the S−particle relic density at the accepted value for the DM content in the universe.

show abstract

“…A CNN utilizes a set of trainable convolutional layers consisting of neurons and nonlinear activation functions to learn complex patterns from data and are for example used in the case of the DeepSource algorithm (Vafaei Sadr et al 2019) developed for source identification in radio data, to which we return to later in the text. Also, we note that similar Deep-Learning techniques are used in Caron et al (2018) in order to analyze gamma-ray signals from the Galactic Center, which is the first time that computer vision/CNN networks are used in the context of gammaray analysis. The U-Net (Ronneberger et al 2015) architecture is a popular CNN model that uses multiscale feature extraction for learning small and large image structures.…”

Section: Introductionmentioning

confidence: 99%

Identification of point sources in gamma rays using U-shaped convolutional neural networks and a data challenge

Caron

Dijkstra

Eckner

et al. 2021

A&A

Self Cite

View full text Add to dashboard Cite

Context. At GeV energies, the sky is dominated by the interstellar emission from the Galaxy. With limited statistics and spatial resolution, accurately separating point sources is therefore challenging. Aims. Here we present the first application of deep learning based algorithms to automatically detect and classify point sources from gamma-ray data. For concreteness we refer to this approach as AutoSourceID. Methods. To detect point sources, we utilized U-shaped convolutional networks for image segmentation and k-means for source clustering and localization. We also explored the Centroid-Net algorithm, which is designed to find and count objects. Using two algorithms allows for a cross check of the results, while a combination of their results can be used to improve performance. The training data are based on 9.5 years of exposure from The Fermi Large Area Telescope (Fermi-LAT) and we used source properties of active galactic nuclei (AGNs) and pulsars (PSRs) from the fourth Fermi-LAT source catalog (4FGL) in addition to several models of background interstellar emission. The results of the localization algorithm are fed into a classification neural network that is trained to separate the three general source classes (AGNs, PSRs, and FAKE sources). Results. We compared our localization algorithms qualitatively with traditional methods and find them to have similar detection thresholds. We also demonstrate the robustness of our source localization algorithms to modifications in the interstellar emission models, which presents a clear advantage over traditional methods. The classification network is able to discriminate between the three classes with typical accuracy of ∼70%, as long as balanced data sets are used in classification training. In this http URL , we publish our training data sets and analysis scripts and invite the community to join the data challenge aimed to improve the localization and classification of gamma-ray point sources.

show abstract

Analyzing γ rays of the Galactic Center with deep learning

Cited by 34 publications

References 63 publications

Diffuse gamma-ray emission modeling and the Galactic center GeV excess

Diffuse gamma-ray emission modeling and the Galactic center GeV excess

Extended Higgs-portal dark matter and the Fermi-LAT Galactic Center Excess

Identification of point sources in gamma rays using U-shaped convolutional neural networks and a data challenge

Contact Info

Product

Resources

About