Multiwavelength classification of X-ray selected galaxy cluster candidates using convolutional neural networks Matej Kosiba CEA, Saclay 1 September 2020
2/14
3/14
6/14
7/14
8/14
9/14
10/14
11/14
12/14
13/14
Conclusion ● 62% agreement between Zooniverse volunteers and experts, however an extremely pure sample of clusters (99% agreement). ● Our hand-made convolutional neural network achieved the best average accuracy, 90%. ● Future plans: - Galaxy cluster catalogue. - Object detection approach. - Utilisation of other wavelengths, radio, infrared and/or SZ effect. - Conducting cosmological studies. 14/14
Bonus Slides
Convolutional filters
Recommend
More recommend