One approach to a search for the artifacts within the Solar system is to look for the objects (e.g., probes, defunct or active) with anomalous orbits that are significantly different from those of the asteroids. To this effect, we used the data on the orbital parameters of 524,214 asteroids from AstDys-2. Approximately 24% of the asteroids belong to the known families in the orbital parameter space. The unclassified ones are labeled as the ‘background’, produced mainly by the dynamical scattering in the course of the evolution of the solar system. We apply Machine Learning tools to identify objectively defined outliers in the feature space of orbital parameters. Various techniques can be used for this task, including DBSCAN, which use distance measures, and Isolation Forest, which use decision trees, and many others.