Enhancing the reliability of machine learning for gravitational wave parameter estimation with attention-based models

Osaka Metropolitan University researchers introduced a technique to enhance the reliability of gravitational wave parameter estimation results produced by machine learning. The team developed two independent machine learning models based on the Vision Transformer to estimate effective spin and chirp mass from spectrograms of gravitational wave signals from binary black hole mergers.  

To enhance the reliability of these models, the researchers utilized attention maps to visualize the areas the models focus on when making predictions. This approach enabled the team to demonstrate that both models perform parameter estimation based on physically meaningful information. Furthermore, by leveraging these attention maps, the team demonstrated a method to quantify the impact of glitches on parameter estimation. They showed that as the models focus more on glitches, the parameter estimation results become more strongly biased. This suggests that attention maps could potentially be used to distinguish between cases where the results produced by the machine learning model are reliable and cases where they are not.

### 

About OMU 

Established in Osaka as one of the largest public universities in Japan, Osaka Metropolitan University is committed to shaping the future of society through “Convergence of Knowledge” and the promotion of world-class research. For more research news, visit https://www.omu.ac.jp/en/ and follow us on social media: X, Facebook, Instagram, LinkedIn.