Introduction

GammaLearn & Cherenkov Telescope Array Observatory (CTAO)

**Exploring the Universe at Very High Energies (VHE)**

* CTAO * Next-generation ground-based observatory for gamma-ray astronomy * Increased sensitivity * First Large-Sized Telescope (LST-1) operational * GammaLearn project * Collaboration between LAPP (CNRS) and LISTIC (USMB) * Fosters innovative methods in AI for CTAO * https://purl.org/gammalearn

[Science with the Cherenkov Telescope Array.](https://www.worldscientific.com/doi/abs/10.1142/10986) WORLD SCIENTIFIC, 2019

Gamma-ray astronomy

**Observation of the universe in the gamma-ray segment of the electromagnetic spectrum (> 0.1 MeV)**

Energy ranges * CTAO: 20 GeV to 300 TeV * LST prevailing in the lowest energies Scientific objectives * Understanding the origin and role of relativistic cosmic particles * Probing extreme environments (e.g. black holes, neutron stars) * Multi-messenger analysis (neutrinos, gravitational waves and cosmic rays) * Exploring frontiers in physics (e.g. dark matter)

Werner Hofmann. “Perspectives from CTA in relativistic astrophysics”. In: Fourteenth Marcel Grossmann Meeting - MG14. Ed. by Massimo Bianchi, Robert T. Jansen, and Remo Ruffini. Jan. 2018, pp. 223–242

Reconstruction workflow

The reconstruction procedure

Discrimination patterns

→ Different Cherenkov light emission patterns make distinction possible in acquired images

---

https://www.iap.kit.edu/corsika/71.php

Standard approaches

Hillas+Random Forest (RF)

* Morphological prior hypothesis: ellipsoidal integrated signal * Uses Hillas parameters (moments) * Leverages multiple RFs * Pros * Fast and robust * Cons * Necessitates image cleaning * Limited at lower energy levels * In production on LST-1 (baseline)

---

A. M. Hillas. [Cerenkov Light Images of EAS Produced by Primary Gamma Rays and by Nuclei.](https://ntrs.nasa.gov/api/citations/19850026666/downloads/19850026666.pdf) In: 19th International Cosmic Ray Conference (ICRC19), Volume 3. Vol. 3. International Cosmic Ray Conference. Aug. 1985, p. 445. H. Abe et al. [Observations of the Crab Nebula and Pulsar with the Large-sized Telescope Prototype of the Cherenkov Telescope Array](https://arxiv.org/abs/2306.12960) In: Astrophys. J. 956.2 (2023), p. 80

$\gamma$-PhysNet

* CNN-based (interpolated inputs) * Backbone with dual attention * Multi-task architecture * Pros * No prior hypothesis * Less preprocessing (e.g. no cleaning) * Cons * Tricky to optimize * Black-box nature * Expectations * Best performances * Fast inference

---

Mikaël Jacquemont. [Cherenkov Image Analysis with Deep Multi-Task Learning from Single-Telescope Data.](https://hal.archives-ouvertes.fr/hal-03043188) Theses. Université Savoie Mont Blanc, Nov. 2020.

ImPACT and Model++

* Simulation of templates * Matching templates to real data using per-pixel likelihood * Background rejection using Boosted Decision Tree on discriminant parameters * Pros * Best overall performance * Cons * Computationally expensive

---

R.D. Parsons and J.A. Hinton. [A Monte Carlo template based analysis for airCherenkov arrays.](https://arxiv.org/abs/0907.2610) In: Astroparticle Physics 56 (Apr. 2014), pp. 26–34 Mathieu de Naurois and Loïc Rolland. [A high performance likelihood reconstruction of gamma-rays for imaging atmospheric Cherenkov telescopes](https://arxiv.org/abs/0907.2610). In: Astroparticle Physics 32.5 (Dec. 2009), pp. 231–252.

FreePACT

* Neural network to estimate the likelihood-to-evidence ratio * Training the neural network (classifier trained to differentiate between samples drawn from the joint PDF and from the product of the marginal) * Pros * Faster than ImPACT * Yields similar or improved results

---

Georg Schwefer, Robert Parsons, and Jim Hinton. [A Hybrid Approach to Event Reconstruction for Atmospheric Cherenkov Telescopes Combining Machine Learning and Likelihood Fitting.](https://arxiv.org/abs/2406.17502) 2024

$\gamma$-PhysNet application to LST-1

Results on Monte-Carlo data

• Training on Monte-Carlo (MC) data
• Testing on MC data
• Compute metrics (IRFs) for both chains

$\rightarrow$ $\gamma$-PhysNet increases the sensitivity at all energies, with greater factors at low energies.

Thomas Vuillaume et al. [Analysis of the Cherenkov Telescope Array first LargeSized Telescope real data using convolutional neural networks](https://arxiv.org/abs/2108.04130.pdf) 2021

Generalization issue - impact of NSB

* Real labelled data are intrinsically unobtainable

* MC simulations are approximations of the reality

* The Night Sky Background (NSB) is a key difference between them

Impact of NSB on $\gamma$-PhysNet - results on MC data

Generalization issue - impact of NSB on real data

Reconstruction algorithm	Significance (higher is better)	Excess of $\gamma$	Background count
lstchain (Hillas+RF)	12.0 σ	379	308
$\gamma$-PhysNet	12.5 σ	395	302
$\gamma$-PhysNet + Background matching	14.3 σ	476	317

Detection capability of the $\gamma$-PhysNet on Crab observations

Thomas Vuillaume et al. [Analysis of the Cherenkov Telescope Array first LargeSized Telescope real data using convolutional neural networks](https://arxiv.org/abs/2108.04130.pdf) 2021

Results on Crab - Baseline

Research direction: don't change the data, change the model

$\rightarrow$ to produce a more general model
$\rightarrow$ to adapt to unknown differences between MC simulations and real data

Information fusion

Multi-modality: the $\gamma$-PhysNet-CBN architecture

Results with multi-modality on simulations

Results on Crab - $\gamma$-PhysNet-CBN

Domain adaptation

Unsupervised Domain Adaptation (UDA)

**[Domain adaptation](https://arxiv.org/abs/2009.00155): Set of algorithms and techniques to reduce domain discrepancies**

* Domain $\mathcal{D} = (\mathcal{X}, P(x))$ * Take into account unknown differences between * Source domain (labelled simulations) * Target domain (unlabelled real data) * Include unlabelled real data in the training * No target labels → Unsupervised * Selection and improvement of relevant SOTA: DANN, DeepJDOT, DeepCORAL

Yaroslav Ganin et al. [Domain-Adversarial Training of Neural Networks.](https://arxiv.org/abs/1505.07818) 2016. Bharath Bhushan Damodaran et al. [DeepJDOT: Deep Joint Distribution Optimal Transport for Unsupervised Domain Adaptation”](https://arxiv.org/abs/1803.10081) 2018. Baochen Sun and Kate Saenko. [Deep CORAL: Correlation Alignment for Deep Domain Adaptation](https://arxiv.org/abs/1607.01719) 2016.

The $\gamma$-PhysNet-DANN architecture

Results on Crab - $\gamma$-PhysNet-(C)DANN

Transformers

Transformer models

* Image contains redundances * Keep 25% of the patches * In-painting task * Allows to use the hexagonal grid of pixels * Use simulations and real data * Improved generalization

---

Kaiming He et al. [Masked Autoencoders Are Scalable Vision Learners.](https://arxiv.org/abs/2111.06377) 2021.

Event reconstruction example 1

Event reconstruction example 2

Event reconstruction example 3

The $\gamma$-PhysNet-Prime

Results on Crab - $\gamma$-PhysNet-Prime

Conclusion and perspectives

Conclusions

• Novel techniques (Information fusion, Unsupervised domain adaptation, Transformers) implemented to solve simulations vs real data discreprency
• Tested on simulations, in different settings (Light pollution and label shift)
• Tested on real data (Crab), both moonlight and no moonlight conditions
• $\gamma$-PhysNet much more affected by light pollution than Hillas+RF
• Domain adaptation, information fusion and Transformers increase the performance in degraded conditions without data adaptation
• Still, best results obtained using *data* adaptation
• In our case, NSB is dominant compared to unkown differences
• Transformers are still under exploration and yield the best results on simulations

Perspectives

• Dell'aiera 2024 submitted, Dell'aiera 2025 in prep.
• Incorporate these findings to work in producion with LST-1
• Use gammalearn in real-time $\rightarrow$ ANR grant DIRECTA
• Stereoscopic analysis $\rightarrow$ see Stereograph talk on Thursday