Surrogate Signal

Surrogate signals are synthetic representations of complex, often noisy, data, aiming to improve efficiency and accuracy in various scientific applications. Current research focuses on developing sophisticated machine learning models, particularly deep neural networks (including convolutional and fully connected architectures), to generate these surrogates, often by learning the mapping between a readily available, imperfect signal and a desired, high-quality target. This approach is proving valuable in fields like medical imaging (e.g., improving motion correction in SPECT scans) and gravitational wave astronomy (e.g., accelerating the creation of waveform models), where accurate surrogate signals significantly enhance data analysis and reduce computational burdens. The ultimate goal is to create highly accurate surrogates that effectively replace more difficult-to-obtain or computationally expensive data.

Papers

July 20, 2022

Learning to estimate a surrogate respiratory signal from cardiac motion by signal-to-signal translation
Akshay Iyer, Clifford Lindsay, Hendrik Pretorius, Michael King
LeArning Abstract Respiratory Motion Cardiac Motion Signal to Signal Translation Surrogate Signal

March 16, 2022

Deep Residual Error and Bag-of-Tricks Learning for Gravitational Wave Surrogate Modeling
Styliani-Christina Fragkouli, Paraskevi Nousi, Nikolaos Passalis, Panagiotis Iosif, Nikolaos Stergioulas, Anastasios Tefas
Surrogate Model Gravitational Wave Bag of Word Binary Black Hole Merger Deep Residual Learning Surrogate Signal

Surrogate Signal

Papers

Learning to estimate a surrogate respiratory signal from cardiac motion by signal-to-signal translation

Deep Residual Error and Bag-of-Tricks Learning for Gravitational Wave Surrogate Modeling