Journal: Scientific Reports
Article Title: An adaptive approach to machine learning for compact particle accelerators
doi: 10.1038/s41598-021-98785-0
Figure Lengend Snippet: ( A – E ) Comparing input distributions to the CNN predictions reconstructed from the predicted PCA component coefficients. 5 examples were chosen from the 1000 test distributions and show the CNN’s best prediction ( A ), worst prediction ( E ), and a uniformly spaced range between ( B – D ). ( F ) The CNN’s prediction is shown for an experimentally measured beam output and compared to the experimentally measured beam input. ( G ) Results of the prediction from ( F ) fine tuned via ES.
Article Snippet: Recently encoder-decoder CNNs have also been demonstrated with measured beam data at the European XFEL to provide extremely high accuracy (768 × 1064 pixel images) predictions of the beam’s LPS and have also demonstrated an innovative method in which once the decoder half is trained and fixed, multiple different encoders can be used for various working points without having to re-train the decoder .
Techniques: