The installation of the Muon Telescope Detector (MTD) at STAR allows a measurement of the dimuon (μ+μ−) production in heavy-ion collisions over a large invariant mass range for the first time. Data has been collected with the MTD from Au+Au collisions at sNN=200GeV and from p+p collisions at s=200GeV. These two datasets allow for new opportunities to measure the dimuon invariant mass spectra at STAR. Before any dimuon measurements can be made, muons must be identified. This contribution presents muon identification employing deep neural networks (DNN) and compares it with other multi-variate techniques. Applications of the DNN technique for data-driven purity measurements are discussed.