view article

Figure 3
The performance of the CNN model and χ2 error in testing. (a) Training and validation loss as a function of the training epochs for the CNN model. (b) Histogram of χ2 error for the modulus of the coherent X-ray diffraction intensity and its corresponding fitting. Furthermore, histograms of χ2 error for the corresponding amplitude (c) and phase (d) of the particle in real space and their corresponding fitting results.

ISSN: 2052-2525