Speaker
Description
As the global network of gravitational wave detectors grows in both size and sensitivity, the traditional matched filtering method for detecting signals from compact object mergers becomes computationally prohibitive. Machine learning algorithms are a compelling alternative approach to this problem due to their ability to shift the computational cost to the model training process, enabling efficient use of computational resources at search time. Here, we present an end-to-end binary black hole search pipeline, aframe, capable of low-latency identification of binary black hole mergers in LIGO data. Using simulated binary black hole signals and real LIGO noise and glitches, a 1-dimensional convolutional neural network is trained to perform a binary classification of detector strain timeseries data. Further, we highlight the novel infrastructure development steps that have been taken to improve the robustness of our model and establish a paradigm for applying machine learning solutions to gravitational wave problems.
