The fast growing observation rate of the restless Universe, including the recent boom of small-size satellites at the low Earth orbit and sky monitors with large field of view, bring new challenges in the search for transients in the data, especially in the moderate to high background environment. The goal of the thesis would be to study and investigate new approaches in flare detection using machine learning techniques. The new methods will be tested on simulated data and on the real examples of the photometric data from existing space missions. If calibrated well the method has capabilities to become an important tool in transient searches in upcoming fleets of space missions. Cosmic gamma-ray bursts, alone or as a part of the search for gravitational wave counterparts is just one case of many possible fields of usage.