We present a novel technique, called DSVP (Discrimination through Singular Vectors Projections), to discriminate spurious events within a dataset. The purpose of this paper is to lay down a general procedure which can be tailored for a broad variety of applications. After describing the general concept, we apply the algorithm to the problem of identifying nearly coincident events in low temperature microcalorimeters in order to push the time resolution close to its intrinsic limit. In fact, from simulated datasets it was possible to achieve an effective time resolution even shorter than the sampling time of the system considered. The obtained results are contextualized in the framework of the HOLMES experiment, which aims at directly measuring the neutrino mass with the calorimetric approach, allowing to significally improve its statistical sensitivity.