Not yet reviewed by Pith; the record is open.

About a third of the $\gamma$-ray sources detected by the Fermi Large Area Telescope (Fermi-LAT) remain unidentified, and some of these could be exotic objects such as dark matter subhalos. We present a search for these sources using Bayesian neural network classification methods applied to the latest 4FGL-DR3 Fermi-LAT catalog. We first simulate the $\gamma$-ray properties of dark matter subhalos using models from N-body simulations and semi-analytical approaches to the subhalo distribution. We then assess the detectability of this sample in the 4FGL-DR3 catalog using the Fermi-LAT analysis tools. We train our Bayesian neural network to identify candidate dark matter subhalos among the unidentified sources in the 4FGL-DR3 catalog. Our results allow us to derive conservative bounds on the dark matter annihilation cross section by excluding unidentified sources classified as astrophysical-like by our networks. We estimate the number of candidate dark matter subhalos for different dark matter masses and provide a publicly available list for further investigation. Our bounds on the dark matter annihilation cross section are comparable to previous results and become particularly competitive at high dark matter masses.