ai4materials.models.cnn_architectures module

ai4materials.models.cnn_architectures.cnn_architecture_polycrystals(learning_rate=0.0003, conv2d_filters=[32, 16, 8, 8, 16, 32], kernel_sizes=[3, 3, 3, 3, 3, 3], hidden_layer_size=64, n_rows=50, n_columns=32, nb_classes=5, dropout=0.125, img_channels=1)

Deep convolutional neural network model for crystal structure recognition.

This neural network architecture was used to classify crystal structures - represented by the three-dimensional diffraction fingerprint - in Ref. [1].

[1]	A. Ziletti et al., “Automatic structure identification in polycrystals via Bayesian deep learning”, in preparation (2018)

Code author: Angelo Ziletti <angelo.ziletti@gmail.com>

ai4materials.models.cnn_architectures.cnn_nature_comm_ziletti2018(conv2d_filters, kernel_sizes, max_pool_strides, hidden_layer_size, n_rows, n_columns, img_channels, nb_classes)

Deep convolutional neural network model for crystal structure recognition.

This neural network architecture was used to classify crystal structures - represented by the two-dimensional diffraction fingerprint - in Ref. [2]

[2]	A. Ziletti, D. Kumar, M. Scheffler, and L. M. Ghiringhelli, “Insightful classification of crystal structures using deep learning”, Nature Communications, vol. 9, pp. 2775 (2018)

Code author: Angelo Ziletti <angelo.ziletti@gmail.com>

ai4materials.models.cnn_architectures.model_architecture_3d(dim1, dim2, dim3, img_channels, nb_classes)