# run data through autoencoder (so that it can be pulled into classifier) ae_predictions = ae.predict_mse(x_train) #print 'ae_predictions.shape: ', ae_predictions.shape # format the output vectors y_train_vector = np.array([ 1 if y_train[np.where(item >= 1)].size else 0 for item in y_train ]) y_test_vector = np.array( [1 if y_test[np.where(item >= 1)].size else 0 for item in y_test]) #print 'y_train_vector shape :', y_train_vector.shape from classifier import Classifier cf = Classifier('classical', conf) #print 'building %s classifier...' % cf.get_model_type() cf.add_dense() cf.train_classifier(ae_predictions, y_train_vector) cf_model = cf.get_model() # predict nom nom predictions = np.array([ 1 if item >= 0.5 else 0 for item in cf_model.predict(ae.predict_mse(x_test)) ]) predictions_train = np.array([ 1 if item >= 0.5 else 0 for item in cf_model.predict(ae.predict_mse(x_train)) ]) p.plot_wave(predictions, '[test] Classifier Predictions') p.plot_wave(predictions_train, '[train] Classifier predicitions')