MNIST Plot

MNIST/Dataset_Plot/main.py

+import tensorflow as tf
+import numpy as np
+
+# MNIST laden
+mnist = tf.keras.datasets.mnist
+(x_train, y_train), (x_test, y_test) = mnist.load_data()
+
+x_train, x_test = x_train / 255.0, x_test / 255.0
+
+x_train = tf.reshape(x_train, (60000, 28*28))
+input_dim = 28*28
+epochs = 100
+
+for label in range(10):
+    which_labels = y_train == label
+    x_selection = x_train[which_labels]
+    img = np.reshape(x_selection[0], (28,28))
+    filename = "results/single_number_" + str(label) + ".npy"
+    np.save(filename, img)
\ No newline at end of file

MNIST/Dataset_Plot/plot.py

+import numpy as np
+import matplotlib.pyplot as plt
+
+plt.rcParams.update({'font.size': 8})
+plt.figure(figsize=(6, 2.4), constrained_layout=True)
+
+for i in range(10):
+    filename = "results/single_number_" + str(i) + ".npy"
+    img = np.load(filename)
+    ax = plt.subplot(2, 5, i + 1)
+    ax.imshow(img, cmap='gray')
+    plt.axis('off')
+plt.show()
\ No newline at end of file

MNIST/Generate_Initializers/main.py

-
-import tensorflow as tf
-
-# MNIST laden
-mnist = tf.keras.datasets.mnist
-(x_train, y_train), (x_test, y_test) = mnist.load_data()
-
-x_train = tf.reshape(x_train, (60000, 28*28))
-y_train = tf.one_hot(y_train, depth=10)
-
-# Modell erzeugen
-size_hidden_layer = 30
-
-from keras.models import Sequential
-from keras.layers import Dense
-from keras import backend as k
-
-model = Sequential()
-model.add(Dense(size_hidden_layer, input_dim = 28*28,activation='sigmoid'))
-model.add(Dense(10, input_dim=size_hidden_layer, activation='sigmoid'))
-
-loss_fn = tf.keras.losses.MeanSquaredError()
-model.compile(optimizer="adam", loss = loss_fn)
-
-model.fit(x_train, y_train, epochs=10000)
-

MNIST/Layer_Sizes/layer_sizes.bat

+#!/bin/bash -l
+#SBATCH -p gpu 
+#SBATCH	--ntasks 1
+#SBATCH --cpus-per-task 4
+#SBATCH --gres gpu:2
+#SBATCH --constraint gpu_k20
+#SBTACH --mem-per-cpu 2000
+#SBATCH --time 30:00:00
+#SBATCH --mail-type END,FAIL,REQUEUE
+#SBATCH	--mail-user eva.fesefeldt@tuhh.de
+conda activate tf-gpu
+
+python3 main.py
+
+exit

MNIST/Layer_Sizes/main.py

+import tensorflow as tf
+import numpy as np
+
+# MNIST laden
+mnist = tf.keras.datasets.mnist
+(x_train, y_train), (x_test, y_test) = mnist.load_data()
+
+x_train = tf.reshape(x_train, (60000, 28*28))
+y_train = tf.one_hot(y_train, depth=10)
+input_dim = 28*28
+epochs = 100
+
+# Modell erzeugen
+from tf.keras.models import Sequential
+from tf.keras.layers import Dense
+from tf.keras import backend as k
+
+loss_fn = tf.keras.losses.MeanSquaredError()
+
+mini_loss = np.zeros((1000,epochs))
+
+for size_hidden_layer in range(1, 1000):
+    model = Sequential()
+    model.add(Dense(size_hidden_layer, input_dim = input_dim, activation='sigmoid'))
+    model.add(Dense(10, input_dim=size_hidden_layer, activation='sigmoid'))
+    #list_of_weights_and_biases = model.get_weights()
+    model.summary()
+
+    model.compile(optimizer="adam", loss = loss_fn)
+    history = model.fit(x_train, y_train, epochs=epochs, verbose=False)
+    mini_loss[size_hidden_layer-1,:] = history.history['loss']
+    np.save("results/mini_loss_vs_size_hidden_layer.npy", mini_loss)

MNIST/Layer_Sizes/plot.py

+import numpy as np
+import matplotlib.pyplot as plt
+
+loss_vs_hidden_layer = np.load("results/mini_loss_vs_size_hidden_layer.npy")
+
+for size_hidden_layer in range(0,51, 5):
+    plt.semilogy(loss_vs_hidden_layer[size_hidden_layer,:], label="n="+str(size_hidden_layer+1))
+
+plt.xlabel("Epoche")
+plt.ylabel("Loss")
+plt.legend()
+plt.show()