diff --git a/main/experiments/scripts/november_2018/keras_end_to_end/deepstrom_classif_end_to_end.py b/main/experiments/scripts/november_2018/keras_end_to_end/deepstrom_classif_end_to_end.py
index be1d65708991f93703a9495b339b5960dd52f1b7..03d3196c40f7834c26f1ea8bdd3f0cf030731711 100644
--- a/main/experiments/scripts/november_2018/keras_end_to_end/deepstrom_classif_end_to_end.py
+++ b/main/experiments/scripts/november_2018/keras_end_to_end/deepstrom_classif_end_to_end.py
@@ -62,7 +62,7 @@ from keras.optimizers import Adam
 from keras.preprocessing.image import ImageDataGenerator
 
 import skluc.main.data.mldatasets as dataset
-from skluc.main.keras_.kernel import map_kernel_name_function
+from skluc.main.keras_.kernel import keras_chi_square_CPD, map_kernel_name_function
 # from skluc.main.keras_.kernel_approximation.nystrom_layer import DeepstromLayerEndToEnd
 from skluc.main.keras_.kernel_approximation.fastfood_layer import FastFoodLayer
 from skluc.main.keras_.models import build_lenet_model, build_vgg19_model_glorot
@@ -81,9 +81,9 @@ def evaluation_function(x_data, y_data, model, list_subsample_bases, datagen_eva
         if X_batch.shape[0] != paraman["--batch-size"]:
             break
         if paraman["network"] == "deepstrom":
-            loss, acc = model.evaluate([X_batch] + list_subsample_bases, [Y_batch])
+            loss, acc = model.evaluate([X_batch] + list_subsample_bases, [Y_batch], verbose=0)
         else:
-            loss, acc = model.evaluate([X_batch], [Y_batch])
+            loss, acc = model.evaluate([X_batch], [Y_batch], verbose=0)
 
         accuracies_val += [acc]
         i += 1
@@ -249,6 +249,7 @@ def main(paraman: ParameterManagerMain, resman, printman):
 
     logger.debug(paraman["kernel_dict"])
 
+    list_subsample_bases = []
     if paraman["network"] == "deepstrom":
         input_subsample = [Input(batch_shape=(paraman["--batch-size"], *input_dim)) for _ in range(paraman["nb_subsample_bases"])]
         if paraman["nb_subsample_bases"] > 1: