Abstract:We present a novel machine learning framework tailored to detect massive black hole binaries observed by spaceborne gravitational wave detectors like the Laser Interferometer Space Antenna (LISA) and predict their future merger times. The detection is performed via convolutional neural networks that analyze time-evolving Time-Delay Interferometry (TDI) spectrograms and utilize variations in signal power to trigger alerts. The prediction of future merger times is accomplished with reinforcement learning. Here, the proposed algorithm dynamically refines time-to-merger predictions by assimilating new data as it becomes available. Deep Q-learning serves as the core technique of the approach, utilizing a neural network to estimate Q-values throughout the observational state space. To enhance robust learning in a noisy environment, we integrate an actor-critic mechanism that segregates action proposals from their evaluation, thus harnessing the advantages of policy-based and value-based learning paradigms. We leverage merger estimation obtained via template matching with truncated waveforms to generate rewards for the reinforcement learning agent. These estimations come with uncertainties, which magnify as the merger event stretches further into the future. The reinforcement learning setup is shown to adapt to these uncertainties by employing a policy fine-tuning approach, ensuring the reliability of predictions despite the varying degrees of template-matching precision. The algorithm denotes a first step toward a low-latency model that provides early warnings for impending transient phenomena. By delivering timely and accurate forecasts of merger events, the framework supports the coordination of gravitational wave observations with accompanied electromagnetic counterparts, thus enhancing the prospects of multi-messenger astronomy. We use the LISA Spritz data challenge for validation.

Dictionary learning: a novel approach to detecting binary black holes in the presence of Galactic noise with LISA

Global Analysis of the Gravitational Wave Signal from Galactic Binaries

Detection and Prediction of Future Massive Black Hole Mergers with Machine Learning and Truncated Waveforms

Rapid detection of gravitational waves from binary black hole mergers using sparse dictionary learning

Premerger observation and characterization of massive black hole binaries

A Novel Deep Learning Approach to Detecting Binary Black Hole Mergers

Low-dimensional signal representations for massive black hole binary signals analysis from LISA data

Global Analysis of LISA Data with Galactic Binaries and Massive Black Hole Binaries

Fast resolving Galactic binaries in LISA data and its ability to study the Milky Way

A hierarchical search for gravitational waves from supermassive black hole binary mergers

Properties and merger signatures of galaxies hosting LISA coalescing massive black hole binaries

Neutron Star - White Dwarf Binaries: Probing Formation Pathways and Natal Kicks with LISA

Merging Black Holes in Dwarf Galaxies: Calculating Binary Black Hole Coalescence Timescales from Simulations for LISA Detection

Exploring gravitational-wave detection and parameter inference using deep learning methods

Identifying the hosts of binary black hole and neutron star-black hole mergers with next-generation gravitational-wave detectors

The LISA Data Challenge Radler Analysis and Time-dependent Ultra-compact Binary Catalogues

Adapting a novel framework for rapid inference of massive black hole binaries for LISA

Identifying Disappearance of a White Dwarf Binary with LISA

Reconstructing the LISA massive black hole binary population via iterative kernel density estimation