util.visualization package

Submodules

util.visualization.confusion_matrix_heatmap module

util.visualization.confusion_matrix_heatmap.make_heatmap(confusion_matrix, class_names)[source]

This function prints and plots the confusion matrix.

Adapted from https://gist.github.com/shaypal5/94c53d765083101efc0240d776a23823

Parameters

  • confusion_matrix (numpy.ndarray) – Array containing the confusion matrix to be plotted

  • class_names (list of strings) – Names of the different classes

Returns

data – Contains an RGB image of the plotted confusion matrix

Return type

numpy.ndarray

util.visualization.decision_boundaries module

util.visualization.decision_boundaries.plot_decision_boundaries(output_winners, output_confidence, grid_x, grid_y, point_x, point_y, point_class, num_classes, step, writer, epochs, **kwargs)[source]

Plots the decision boundaries as a 2D image onto Tensorboard.

Parameters

  • output_winners (numpy.ndarray) – which class is the ‘winner’ of the network at each location

  • output_confidence (numpy.ndarray) – confidence value of the network for the ‘winner’ class

  • grid_x (numpy.ndarray) – X axis locations of the decision grid

  • grid_y (numpy.ndarray) – Y axis locations of the decision grid

  • point_x (numpy.ndarray) – X axis locations of the real points to be plotted

  • point_y (numpy.ndarray) – Y axis locations of the real points to be plotted

  • point_class (numpy.ndarray) – class of the real points at each location

  • writer (tensorboardX SummaryWriter) – Tensorboard summarywriter object

  • num_classes (int) – number of unique classes

  • step (int) – global training step

  • epochs (int) – total number of training epochs

Returns

Return type

None

util.visualization.embedding module

This script generates embedding visualization for features produced by the apply_model script.

util.visualization.embedding.isomap(features, n_components=2)[source]

Returns the embedded points for Isomap. :Parameters: * features (numpy.ndarray) – contains the input feature vectors.

  • n_components (int) – number of components to transform the features into

Returns

embedding – x,y(z) points that the feature vectors have been transformed into

Return type

numpy.ndarray

util.visualization.embedding.mds(features, n_components=2)[source]

Returns the embedded points for MDS. :Parameters: * features (numpy.ndarray) – contains the input feature vectors.

  • n_components (int) – number of components to transform the features into

Returns

embedding – x,y(z) points that the feature vectors have been transformed into

Return type

numpy.ndarray

util.visualization.embedding.pca(features, n_components=2)[source]

Returns the embedded points for PCA. :Parameters: * features (numpy.ndarray) – contains the input feature vectors.

  • n_components (int) – number of components to transform the features into

Returns

embedding – x,y(z) points that the feature vectors have been transformed into

Return type

numpy.ndarray

util.visualization.embedding.tsne(features, n_components=2)[source]

Returns the embedded points for TSNE. :Parameters: * features (numpy.ndarray) – contains the input feature vectors.

  • n_components (int) – number of components to transform the features into

Returns

embedding – x,y(z) points that the feature vectors have been transformed into

Return type

numpy.ndarray

util.visualization.mean_std_plot module

util.visualization.mean_std_plot.plot_mean_std(x=None, arr=None, suptitle='', title='', xlabel='X', ylabel='Y', xlim=None, ylim=None)[source]

Plots the accuracy/loss curve over several runs with standard deviation and mean. :Parameters: * x (numpy.ndarray) – contains the ticks on the x-axis

  • arr (numpy.ndarray) – contains the accuracy values for each epoch per run

  • suptitle (str) – title for the plot

  • title (str) – sub-title for the plot

  • xlabel (str) – label for the x-axis

  • ylabel (str) – label for the y-axis

  • xlim (float or None) – optionally specify a upper limit on the x-axis

  • ylim (float or None) – optionally specify a upper limit on the y-axis

Returns

data – Contains an RGB image of the plotted accuracy curves

Return type

numpy.ndarray

util.visualization.visualize_activations module

This script generates visualizations of the activation of intermediate layers of CNNs.

util.visualization.visualize_activations.main(args)[source]

Main routine of script to generate activation heatmaps. :Parameters: args (argparse.Namespace) – contains all arguments parsed from input

Returns

Return type

None

util.visualization.visualize_activations.make_grid_own(activations)[source]

Plots all activations of a layer in a grid format. :Parameters: activations (numpy.ndarray) – array of activation values for each filter in a layer

Returns

large_fig – image array containing all activation heatmaps of a layer

Return type

numpy.ndarray

Module contents