Gravitational wave detectors suffer severely from (non-linear) noise. A new branch of research in computational physics and gravitational wave physics proposes the use of recent models of Artificial Intelligence and Deep Learning in detector pipelines to improve GW search performance.
In this project, I decided to implement and reproduce the results of a scientific paper:
[ 1 ] George, Daniel, Hongyu Shen, and E. A. Huerta. "Deep Transfer Learning: A new deep learning glitch classification method for advanced LIGO." arXiv preprint arXiv:1706.07446 (2017).
Nevertheless the limited computational power, my tests confirmed the [ 1 ] conclusions. I showed, yet confirmed [ 1 ], how transfer learning is effective with a clustering approach.
In the presention, there are all the results. In the notebook, you can find the code and all the necessary information to reproduce the results. Actaullly the notebook is almost ready to be executed.
The framework used in this project is TensorFlow and Spark.