Module connectome.visualization.viz_nn

visualization helpers and functions for convolutional neural networks

Functions

def model_modifier_function(cloned_model)

def nn_feature_visualization(model, X, y, method='saliency', average=False, ordered=False)

Returns plots for different feature attribution methods: Vanilla Saliency and Smooth Saliency. If you want to view e.g. the first 3 visualizations and not the average, but have structural data, the input should follow the format X=[X_test[0][0:3],X_test[1][0:3]], y=y_test[0:3]

Args

model

A fitted neural network model

X

A test dataframe to visualize

y

labels

method

'saliency' or 'saliency_smooth'

average

Return the average or individual feature maps

ordered

Boolean, whether to reorder the matrices based on the yeo7 network. (True is recommended when training with Brainnetome data. Only applicable to data based on the brainnetome atlas.

Returns

Returns a list feature attribution plots for the desired method

def score_function(output)