One of the general goals of the PhD project is to study the universe through gravitational waves (GWs) and use machine learning (ML) to detect features of these signatures in current and future data. Since the first detection of GWs (2015) by LIGO-Virgo has begun a new era of GW astronomy. Recently there has also been renewed interest in primordial black holes (PBHs), formed shortly after the Big Bang by the gravitational collapse of primordial density fluctuations. In particular, PBHs have unique signatures due to their Close Hyperbolic Encounters (CHEs), which could be detected by current and future GWs detectors. However, accurate models are needed for GWs emitted by CHEs, as the signal resembles other noise sources. To do this we need to study how GWs leave a signal in the network of detectors currently present on Earth.

Despite the numerous detections of GWs made, all these observations are for black holes (BHs) and neutron stars (NSs) (or BH-NS) merger events due to the strength and duration of the signal as well as its characteristic shape. The characterization of signals different from those of binary systems is more difficult because they are too weak to be measured with current detectors, such as continuous GWs and stochastic GWs, or too short in duration to be clearly distinguishable from noise, as burst (typical signal from CHEs). However, it is very likely that these sources of GWs will lead to new discoveries about the Universe. In this context, ML algorithms are very important to study the huge amount of data and try to recognize this type of signals in the noise of detectors. These algorithms have recently been introduced into cosmology, revolutionizing data analysis, because they are suitable for interpretation of many types of current cosmological data and for trying to separate signal from noise, as well to help in the parameter estimation.

So far in my PhD, I have delved into the subject of GWs, focusing on modelling the GW signal from CHEs between compact objects, and started to study machine learning techniques, deepening technical skills to apply them in future work in the context of GWs. The objective is to continue exploring GWs phenomena using ML techniques, where possible. Currently, I am working on using ML classification algorithms in the context of stochastic GWs.