A machine learning approach to axion-like particle searches in CTAO observations of blazars

Axion-like particles (ALPs) are a common prediction of several extensions of the Standard Model of particle physics and could be detected through their coupling to photons, which enables ALP-photon conversions in external magnetic fields. This conversion could lead to two distinct signatures in gamma-ray spectra of blazars: a superimposition of energy-dependent "wiggles" on the spectral shape, and a hardening at high (multi-TeV) energies, due to the ALP beam eluding absorption by the extragalactic background light (EBL). The enhanced energy resolution of the Cherenkov Telescope Array Observatory (CTAO) with respect to present ground-based gamma-ray telescopes makes it an ideal instrument to probe such phenomena. In this contribution, we explore a different approach based on the use of machine learning (ML) classifiers and compare it to the standard method. Our preliminary results suggest that both techniques yield consistent results, with the ML-based method offering comparable or even slightly broader coverage, potentially extending the CTAO sensitivity beyond existing constraints.