Detecting intermediate mass black holes in globular clusters with machine learning

Mergers of stellar-mass black holes were recently observed in the gravitational wave window opened by LIGO. This puts the spotlight on dense stellar systems and their ability to create intermediate-mass black holes (IMBHs) through repeated merging. Unfortunately, attempts at direct and indirect IMBH detection in star clusters in the nearby universe have proven inconclusive as of now. Indirect detection methods attempt to constrain IMBHs through their effect on star cluster photometric and kinematic observables. They are usually based on looking for a specific, physically motivated signature. While this approach is justified, it may be suboptimal in its usage of the available data. Here I present a new indirect detection method, based on machine learning, that is unaffected by these restrictions. I reduce the scientific question whether a star cluster hosts an IMBH to a classification problem in the machine learning framework. I present preliminary results to illustrate how machine learning models are trained on simulated dataset and measure their performance on previously unseen, simulated data.