Grammar for NN

Neural networks (NNs) can be described using various notations. we have designed a textual notation for their definition, supported by a grammar we developed to instantiate the concepts of the metamodel.

A textual example of the neural network (NN) model is shown below. The model definition begins by specifying the NN’s name (my_model). Next, the layers are defined outlining three layers (l1, l2, and l3), with l1 and l3 being 2D Convolutional layers, and l2 as a Pooling layer. Then, the modules definition specifies the order of the layers. Finally, hyperparameters are defined, such as the “adam” optimiser.

my_model:
    layers:
        - l1:
            type=Conv2D
            actv_func=relu
            in_channels=3
            out_channels=32
            kernel_dim=[3, 3]
        - l2:
            type=Pooling
            pooling_type=max
            dimension=2D
            kernel_dim=[2, 2]
        - l3:
            type=Conv2D
            actv_func=relu
            in_channels=32
            out_channels=64
            kernel_dim=[3, 3]
    modules:
        - l1 - l2 - l3 - l4 - l5 - l6 - l7 - l8
    config:
        batch_size=32
        epochs=10
        learning_rate=0.001
        optimiser="adam"
        metrics=["f1-score"]
        loss_function=crossentropy
        weight_decay=0.2
        momentum=0.1

Save this model as a textual file, e.g. nn_model.txt.

Then, load and process the model using our grammar and apply the transformation to obtain the B-UML based model.

# Import methods and classes
from besser.BUML.notations.nn import buml_neural_network

# PlantUML to B-UML model
nn_buml_model = buml_neural_network("nn_model.txt")

nn_buml_model is the BUML model containing the neural network specification.