The KM3NeT research infrastructure is currently under construction at two locations in the Mediterranean Sea. The KM3NeT/ORCA water-Cherenkov neutrino detector off the French coast will instrument several megatons of seawater with photosensors. Its main objective is the determination of the neutrino mass ordering. This work aims at demonstrating the general applicability of deep convolutional neural networks to neutrino telescopes, using simulated datasets for the KM3NeT/ORCA detector as an example. To this end, the networks are employed to achieve reconstruction and classification tasks that constitute an alternative to the analysis pipeline presented for KM3NeT/ORCA in the KM3NeT Letter of Intent. They are used to infer event reconstruction estimates for the energy, the direction, and the interaction point of incident neutrinos. The spatial distribution of Cherenkov light generated by charged particles induced in neutrino interactions is classified as shower- or track-like, and the main background processes associated with the detection of atmospheric neutrinos are recognized. Performance comparisons to machine-learning classification and maximum-likelihood reconstruction algorithms previously developed for KM3NeT/ORCA are provided. It is shown that this application of deep convolutional neural networks to simulated datasets for a large-volume neutrino telescope yields competitive reconstruction results and performance improvements with respect to classical approaches.

MeV-GeV dark matter (DM) is theoretically well motivated but remarkably unexplored. This proposal presents the MeV-GeV DM discovery potential for a $\sim$1 m$^3$ segmented CsI(Tl) scintillator detector placed downstream of the Hall A beam-dump at Jefferson Lab, receiving up to 10$^{22}$ electrons-on-target (EOT) in 285 days. This experiment (Beam-Dump eXperiment or BDX) would be sensitive to elastic DM-electron and to inelastic DM scattering at the level of 10 counts per year, reaching the limit of the neutrino irreducible background. The distinct signature of a DM interaction will be an electromagnetic shower of few hundreds of MeV, together with a reduced activity in the surrounding active veto counters. A detailed description of the DM particle $\chi$ production in the dump and subsequent interaction in the detector has been performed by means of Monte Carlo simulations. Different approaches have been used to evaluate the expected backgrounds: the cosmogenic background has been extrapolated from the results obtained with a prototype detector running at INFN-LNS (Italy), while the beam-related background has been evaluated by GEANT4 Monte Carlo simulations. The proposed experiment will be sensitive to large regions of DM parameter space, exceeding the discovery potential of existing and planned experiments in the MeV-GeV DM mass range by up to two orders of magnitude.