Description: Building a neural network using TensorFlow 2.0 and comparing the accuracy of the outputs based off of the models training.
In developing Treazure Map AI Agents, I developed additional functionalities to enhance the Cartpole simulation using Deep Q-Learning to solve a game where the agent acts as a pirate and trains itself using RL to find the treasure.
TreasureMaze.pydefines the maze environment andGameExperience.pyis used for implementing an experience replay.
The qtrain function was developed for optimal Deep Q-Learning, tuning hyperparameters, and enhancing visualizations to track the AI's progress after its training.
The Deep Q-Learning project showed me how AI fits into what computer scientists do every day and how can be used in solving problems, advancing technology, and handling ethical challenges. This class taught me to look at problems closer and more clearly, to design and test solutions differently and more efficiently. This class also reminded me how important it is to respect user privacy, follow rules, and think about the bigger picture and to solve issues while keeping ethics in mind.
---
To install TensorFlow 2.0 using pip:
pip install tensorflow==2.0.0
pip install --upgrade tensorflowimport tensorflow as tf
import keras
print("TensorFlow version:", tf.__version__)
print("Keras version:", keras.__version__)
print("Num GPUs Available: ", len(tf.config.list_physical_devices('GPU')))Output:
TensorFlow version: 2.18.0
Keras version: 3.6.0
Num GPUs Available: 0
from tensorflow.keras import backend as K
K.clear_session()Warning:
Use tf.compat.v1.reset_default_graph instead of tf.reset_default_graph.
Improving the Simple Net in Keras with Hidden Layers
Adding additional layers to improve accuracy.
import numpy as np
from tensorflow.keras.datasets import mnist
from tensorflow.keras.models import Sequential
from tensorflow.keras.layers import Dense, Activation
from tensorflow.keras.optimizers import SGD
from tensorflow.keras.utils import to_categorical
np.random.seed(1671)
NB_EPOCH = 20
BATCH_SIZE = 128
VERBOSE = 1
NB_CLASSES = 10
OPTIMIZER = SGD()
N_HIDDEN = 128
VALIDATION_SPLIT = 0.2
(X_train, y_train), (X_test, y_test) = mnist.load_data()
RESHAPED = 784
X_train = X_train.reshape(60000, RESHAPED).astype('float32') / 255
X_test = X_test.reshape(10000, RESHAPED).astype('float32') / 255
Y_train = to_categorical(y_train, NB_CLASSES)
Y_test = to_categorical(y_test, NB_CLASSES)
model = Sequential()
model.add(Dense(N_HIDDEN, input_shape=(RESHAPED,)))
model.add(Activation('relu'))
model.add(Dense(N_HIDDEN))
model.add(Activation('relu'))
model.add(Dense(NB_CLASSES))
model.add(Activation('softmax'))
model.summary()
model.compile(loss='categorical_crossentropy', optimizer=OPTIMIZER, metrics=['accuracy'])
history = model.fit(X_train, Y_train, batch_size=BATCH_SIZE, epochs=NB_EPOCH, verbose=VERBOSE, validation_split=VALIDATION_SPLIT)
score = model.evaluate(X_test, Y_test, verbose=VERBOSE)
print("Test score:", score[0])
print("Test accuracy:", score[1])60000 train samples
10000 test samples
Model: "sequential"
┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓
┃ Layer (type) ┃ Output Shape ┃ Param # ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩
│ dense (Dense) │ (None, 128) │ 100,480 │
├─────────────────────────────────┼────────────────────────┼───────────────┤
│ activation (Activation) │ (None, 128) │ 0 │
├─────────────────────────────────┼────────────────────────┼───────────────┤
│ dense_1 (Dense) │ (None, 128) │ 16,512 │
├─────────────────────────────────┼────────────────────────┼───────────────┤
│ activation_1 (Activation) │ (None, 128) │ 0 │
├─────────────────────────────────┼────────────────────────┼───────────────┤
│ dense_2 (Dense) │ (None, 10) │ 1,290 │
├─────────────────────────────────┼────────────────────────┼───────────────┤
│ activation_2 (Activation) │ (None, 10) │ 0 │
└─────────────────────────────────┴────────────────────────┴───────────────┘
Total params: 118,282 (462.04 KB)
Trainable params: 118,282 (462.04 KB)
Non-trainable params: 0 (0.00 B)
Epoch 1/20
375/375 ━━━━━━━━━━━━━━━━━━━━ 2s 4ms/step - accuracy: 0.4416 - loss: 1.9155 - val_accuracy: 0.8271 - val_loss: 0.7860
Epoch 2/20
375/375 ━━━━━━━━━━━━━━━━━━━━ 1s 3ms/step - accuracy: 0.8250 - loss: 0.7064 - val_accuracy: 0.8775 - val_loss: 0.4707
Epoch 3/20
375/375 ━━━━━━━━━━━━━━━━━━━━ 1s 3ms/step - accuracy: 0.8677 - loss: 0.4762 - val_accuracy: 0.8944 - val_loss: 0.3829
Epoch 4/20
375/375 ━━━━━━━━━━━━━━━━━━━━ 1s 2ms/step - accuracy: 0.8854 - loss: 0.4056 - val_accuracy: 0.9039 - val_loss: 0.3426
Epoch 5/20
375/375 ━━━━━━━━━━━━━━━━━━━━ 1s 3ms/step - accuracy: 0.8985 - loss: 0.3614 - val_accuracy: 0.9096 - val_loss: 0.3168
Epoch 6/20
375/375 ━━━━━━━━━━━━━━━━━━━━ 1s 3ms/step - accuracy: 0.9067 - loss: 0.3279 - val_accuracy: 0.9158 - val_loss: 0.2979
Epoch 7/20
375/375 ━━━━━━━━━━━━━━━━━━━━ 1s 3ms/step - accuracy: 0.9108 - loss: 0.3154 - val_accuracy: 0.9204 - val_loss: 0.2830
Epoch 8/20
375/375 ━━━━━━━━━━━━━━━━━━━━ 1s 2ms/step - accuracy: 0.9154 - loss: 0.3001 - val_accuracy: 0.9217 - val_loss: 0.2716
Epoch 9/20
375/375 ━━━━━━━━━━━━━━━━━━━━ 1s 3ms/step - accuracy: 0.9211 - loss: 0.2822 - val_accuracy: 0.9271 - val_loss: 0.2601
Epoch 10/20
375/375 ━━━━━━━━━━━━━━━━━━━━ 1s 3ms/step - accuracy: 0.9233 - loss: 0.2759 - val_accuracy: 0.9289 - val_loss: 0.2501
Epoch 11/20
375/375 ━━━━━━━━━━━━━━━━━━━━ 1s 3ms/step - accuracy: 0.9265 - loss: 0.2607 - val_accuracy: 0.9306 - val_loss: 0.2428
Epoch 12/20
375/375 ━━━━━━━━━━━━━━━━━━━━ 1s 3ms/step - accuracy: 0.9307 - loss: 0.2451 - val_accuracy: 0.9333 - val_loss: 0.2359
Epoch 13/20
375/375 ━━━━━━━━━━━━━━━━━━━━ 1s 3ms/step - accuracy: 0.9320 - loss: 0.2386 - val_accuracy: 0.9352 - val_loss: 0.2280
Epoch 14/20
375/375 ━━━━━━━━━━━━━━━━━━━━ 1s 3ms/step - accuracy: 0.9360 - loss: 0.2307 - val_accuracy: 0.9366 - val_loss: 0.2230
Epoch 15/20
375/375 ━━━━━━━━━━━━━━━━━━━━ 1s 3ms/step - accuracy: 0.9361 - loss: 0.2260 - val_accuracy: 0.9400 - val_loss: 0.2143
Epoch 16/20
375/375 ━━━━━━━━━━━━━━━━━━━━ 1s 3ms/step - accuracy: 0.9395 - loss: 0.2138 - val_accuracy: 0.9421 - val_loss: 0.2095
Epoch 17/20
375/375 ━━━━━━━━━━━━━━━━━━━━ 1s 3ms/step - accuracy: 0.9388 - loss: 0.2153 - val_accuracy: 0.9437 - val_loss: 0.2031
Epoch 18/20
375/375 ━━━━━━━━━━━━━━━━━━━━ 1s 3ms/step - accuracy: 0.9424 - loss: 0.2021 - val_accuracy: 0.9448 - val_loss: 0.1985
Epoch 19/20
375/375 ━━━━━━━━━━━━━━━━━━━━ 1s 3ms/step - accuracy: 0.9452 - loss: 0.1937 - val_accuracy: 0.9474 - val_loss: 0.1951
Epoch 20/20
375/375 ━━━━━━━━━━━━━━━━━━━━ 1s 3ms/step - accuracy: 0.9439 - loss: 0.1940 - val_accuracy: 0.9482 - val_loss: 0.1894
313/313 ━━━━━━━━━━━━━━━━━━━━ 1s 2ms/step - accuracy: 0.9368 - loss: 0.2182
Test score: 0.18993066251277924
Test accuracy: 0.9442999958992004
NB_EPOCH = 30
BATCH_SIZE = 128
N_HIDDEN = 128
VALIDATION_SPLIT = 0.2
K.clear_session()
model_exp1 = Sequential()
model_exp1.add(Dense(N_HIDDEN, input_shape=(RESHAPED,)))
model_exp1.add(Activation('relu'))
model_exp1.add(Dense(N_HIDDEN))
model_exp1.add(Activation('relu'))
model_exp1.add(Dense(NB_CLASSES))
model_exp1.add(Activation('softmax'))
optimizer_exp1 = SGD()
model_exp1.compile(loss='categorical_crossentropy', optimizer=optimizer_exp1, metrics=['accuracy'])
history_exp1 = model_exp1.fit(X_train, Y_train, batch_size=BATCH_SIZE, epochs=NB_EPOCH, verbose=VERBOSE, validation_split=VALIDATION_SPLIT)
score_exp1 = model_exp1.evaluate(X_test, Y_test, verbose=VERBOSE)
print("Experiment 1 - Test score:", score_exp1[0])
print("Experiment 1 - Test accuracy:", score_exp1[1])Epoch 1/30
�[1m375/375�[0m �[32m━━━━━━━━━━━━━━━━━━━━�[0m�[37m�[0m �[1m2s�[0m 3ms/step - accuracy: 0.4659 - loss: 1.8417 - val_accuracy: 0.8301 - val_loss: 0.7231
Epoch 2/30
�[1m375/375�[0m �[32m━━━━━━━━━━━━━━━━━━━━�[0m�[37m�[0m �[1m1s�[0m 3ms/step - accuracy: 0.8315 - loss: 0.6646 - val_accuracy: 0.8858 - val_loss: 0.4531
Epoch 3/30
�[1m375/375�[0m �[32m━━━━━━━━━━━━━━━━━━━━�[0m�[37m�[0m �[1m1s�[0m 3ms/step - accuracy: 0.8782 - loss: 0.4614 - val_accuracy: 0.8992 - val_loss: 0.3711
Epoch 4/30
�[1m375/375�[0m �[32m━━━━━━━━━━━━━━━━━━━━�[0m�[37m�[0m �[1m1s�[0m 3ms/step - accuracy: 0.8940 - loss: 0.3883 - val_accuracy: 0.9052 - val_loss: 0.3319
Epoch 5/30
�[1m375/375�[0m �[32m━━━━━━━━━━━━━━━━━━━━�[0m�[37m�[0m �[1m1s�[0m 3ms/step - accuracy: 0.9025 - loss: 0.3476 - val_accuracy: 0.9134 - val_loss: 0.3071
Epoch 6/30
�[1m375/375�[0m �[32m━━━━━━━━━━━━━━━━━━━━�[0m�[37m�[0m �[1m1s�[0m 3ms/step - accuracy: 0.9107 - loss: 0.3190 - val_accuracy: 0.9175 - val_loss: 0.2908
Epoch 7/30
�[1m375/375�[0m �[32m━━━━━━━━━━━━━━━━━━━━�[0m�[37m�[0m �[1m1s�[0m 3ms/step - accuracy: 0.9149 - loss: 0.2980 - val_accuracy: 0.9208 - val_loss: 0.2771
Epoch 8/30
�[1m375/375�[0m �[32m━━━━━━━━━━━━━━━━━━━━�[0m�[37m�[0m �[1m1s�[0m 3ms/step - accuracy: 0.9160 - loss: 0.2930 - val_accuracy: 0.9236 - val_loss: 0.2662
Epoch 9/30
�[1m375/375�[0m �[32m━━━━━━━━━━━━━━━━━━━━�[0m�[37m�[0m �[1m1s�[0m 3ms/step - accuracy: 0.9204 - loss: 0.2760 - val_accuracy: 0.9268 - val_loss: 0.2559
Epoch 10/30
�[1m375/375�[0m �[32m━━━━━━━━━━━━━━━━━━━━�[0m�[37m�[0m �[1m1s�[0m 3ms/step - accuracy: 0.9256 - loss: 0.2636 - val_accuracy: 0.9284 - val_loss: 0.2479
Epoch 11/30
�[1m375/375�[0m �[32m━━━━━━━━━━━━━━━━━━━━�[0m�[37m�[0m �[1m1s�[0m 3ms/step - accuracy: 0.9280 - loss: 0.2518 - val_accuracy: 0.9316 - val_loss: 0.2387
Epoch 12/30
�[1m375/375�[0m �[32m━━━━━━━━━━━━━━━━━━━━�[0m�[37m�[0m �[1m1s�[0m 3ms/step - accuracy: 0.9280 - loss: 0.2505 - val_accuracy: 0.9342 - val_loss: 0.2322
Epoch 13/30
�[1m375/375�[0m �[32m━━━━━━━━━━━━━━━━━━━━�[0m�[37m�[0m �[1m1s�[0m 3ms/step - accuracy: 0.9309 - loss: 0.2434 - val_accuracy: 0.9355 - val_loss: 0.2252
Epoch 14/30
�[1m375/375�[0m �[32m━━━━━━━━━━━━━━━━━━━━�[0m�[37m�[0m �[1m1s�[0m 3ms/step - accuracy: 0.9332 - loss: 0.2348 - val_accuracy: 0.9382 - val_loss: 0.2184
Epoch 15/30
�[1m375/375�[0m �[32m━━━━━━━━━━━━━━━━━━━━�[0m�[37m�[0m �[1m1s�[0m 3ms/step - accuracy: 0.9375 - loss: 0.2205 - val_accuracy: 0.9392 - val_loss: 0.2146
Epoch 16/30
�[1m375/375�[0m �[32m━━━━━━━━━━━━━━━━━━━━�[0m�[37m�[0m �[1m1s�[0m 3ms/step - accuracy: 0.9399 - loss: 0.2144 - val_accuracy: 0.9422 - val_loss: 0.2081
Epoch 17/30
�[1m375/375�[0m �[32m━━━━━━━━━━━━━━━━━━━━�[0m�[37m�[0m �[1m1s�[0m 3ms/step - accuracy: 0.9384 - loss: 0.2154 - val_accuracy: 0.9440 - val_loss: 0.2024
Epoch 18/30
�[1m375/375�[0m �[32m━━━━━━━━━━━━━━━━━━━━�[0m�[37m�[0m �[1m1s�[0m 3ms/step - accuracy: 0.9422 - loss: 0.2020 - val_accuracy: 0.9457 - val_loss: 0.1983
Epoch 19/30
�[1m375/375�[0m �[32m━━━━━━━━━━━━━━━━━━━━�[0m�[37m�[0m �[1m1s�[0m 3ms/step - accuracy: 0.9429 - loss: 0.1951 - val_accuracy: 0.9448 - val_loss: 0.1962
Epoch 20/30
�[1m375/375�[0m �[32m━━━━━━━━━━━━━━━━━━━━�[0m�[37m�[0m �[1m1s�[0m 3ms/step - accuracy: 0.9454 - loss: 0.1904 - val_accuracy: 0.9467 - val_loss: 0.1896
Epoch 21/30
�[1m375/375�[0m �[32m━━━━━━━━━━━━━━━━━━━━�[0m�[37m�[0m �[1m1s�[0m 3ms/step - accuracy: 0.9471 - loss: 0.1853 - val_accuracy: 0.9469 - val_loss: 0.1857
Epoch 22/30
�[1m375/375�[0m �[32m━━━━━━━━━━━━━━━━━━━━�[0m�[37m�[0m �[1m1s�[0m 3ms/step - accuracy: 0.9461 - loss: 0.1850 - val_accuracy: 0.9498 - val_loss: 0.1819
Epoch 23/30
�[1m375/375�[0m �[32m━━━━━━━━━━━━━━━━━━━━�[0m�[37m�[0m �[1m1s�[0m 3ms/step - accuracy: 0.9494 - loss: 0.1762 - val_accuracy: 0.9507 - val_loss: 0.1780
Epoch 24/30
�[1m375/375�[0m �[32m━━━━━━━━━━━━━━━━━━━━�[0m�[37m�[0m �[1m1s�[0m 3ms/step - accuracy: 0.9496 - loss: 0.1755 - val_accuracy: 0.9489 - val_loss: 0.1753
Epoch 25/30
�[1m375/375�[0m �[32m━━━━━━━━━━━━━━━━━━━━�[0m�[37m�[0m �[1m1s�[0m 3ms/step - accuracy: 0.9526 - loss: 0.1675 - val_accuracy: 0.9509 - val_loss: 0.1714
Epoch 26/30
�[1m375/375�[0m �[32m━━━━━━━━━━━━━━━━━━━━�[0m�[37m�[0m �[1m1s�[0m 2ms/step - accuracy: 0.9539 - loss: 0.1635 - val_accuracy: 0.9521 - val_loss: 0.1682
Epoch 27/30
�[1m375/375�[0m �[32m━━━━━━━━━━━━━━━━━━━━�[0m�[37m�[0m �[1m1s�[0m 3ms/step - accuracy: 0.9544 - loss: 0.1632 - val_accuracy: 0.9541 - val_loss: 0.1662
Epoch 28/30
�[1m375/375�[0m �[32m━━━━━━━━━━━━━━━━━━━━�[0m�[37m�[0m �[1m1s�[0m 3ms/step - accuracy: 0.9548 - loss: 0.1571 - val_accuracy: 0.9539 - val_loss: 0.1634
Epoch 29/30
�[1m375/375�[0m �[32m━━━━━━━━━━━━━━━━━━━━�[0m�[37m�[0m �[1m1s�[0m 3ms/step - accuracy: 0.9562 - loss: 0.1527 - val_accuracy: 0.9550 - val_loss: 0.1596
Epoch 30/30
�[1m375/375�[0m �[32m━━━━━━━━━━━━━━━━━━━━�[0m�[37m�[0m �[1m1s�[0m 3ms/step - accuracy: 0.9577 - loss: 0.1509 - val_accuracy: 0.9563 - val_loss: 0.1572
�[1m313/313�[0m �[32m━━━━━━━━━━━━━━━━━━━━�[0m�[37m�[0m �[1m1s�[0m 2ms/step - accuracy: 0.9470 - loss: 0.1808
Experiment 1 - Test score: 0.15435431897640228
Experiment 1 - Test accuracy: 0.954800009727478
NB_EPOCH = 10
BATCH_SIZE = 128
N_HIDDEN = 128
VALIDATION_SPLIT = 0.2
K.clear_session()
model_exp2 = Sequential()
model_exp2.add(Dense(N_HIDDEN, input_shape=(RESHAPED,)))
model_exp2.add(Activation('relu'))
model_exp2.add(Dense(N_HIDDEN))
model_exp2.add(Activation('relu'))
model_exp2.add(Dense(NB_CLASSES))
model_exp2.add(Activation('softmax'))
optimizer_exp2 = SGD()
model_exp2.compile(loss='categorical_crossentropy', optimizer=optimizer_exp2, metrics=['accuracy'])
history_exp2 = model_exp2.fit(X_train, Y_train, batch_size=BATCH_SIZE, epochs=NB_EPOCH, verbose=VERBOSE, validation_split=VALIDATION_SPLIT)
score_exp2 = model_exp2.evaluate(X_test, Y_test, verbose=VERBOSE)
print("Experiment 2 - Test score:", score_exp2[0])
print("Experiment 2 - Test accuracy:", score_exp2[1])Epoch 1/10
�[1m375/375�[0m �[32m━━━━━━━━━━━━━━━━━━━━�[0m�[37m�[0m �[1m2s�[0m 3ms/step - accuracy: 0.4725 - loss: 1.8071 - val_accuracy: 0.8499 - val_loss: 0.7077
Epoch 2/10
�[1m375/375�[0m �[32m━━━━━━━━━━━━━━━━━━━━�[0m�[37m�[0m �[1m1s�[0m 3ms/step - accuracy: 0.8429 - loss: 0.6510 - val_accuracy: 0.8873 - val_loss: 0.4469
Epoch 3/10
�[1m375/375�[0m �[32m━━━━━━━━━━━━━━━━━━━━�[0m�[37m�[0m �[1m1s�[0m 3ms/step - accuracy: 0.8781 - loss: 0.4551 - val_accuracy: 0.8996 - val_loss: 0.3709
Epoch 4/10
�[1m375/375�[0m �[32m━━━━━━━━━━━━━━━━━━━━�[0m�[37m�[0m �[1m1s�[0m 2ms/step - accuracy: 0.8913 - loss: 0.3883 - val_accuracy: 0.9084 - val_loss: 0.3345
Epoch 5/10
�[1m375/375�[0m �[32m━━━━━━━━━━━━━━━━━━━━�[0m�[37m�[0m �[1m1s�[0m 3ms/step - accuracy: 0.9007 - loss: 0.3509 - val_accuracy: 0.9123 - val_loss: 0.3117
Epoch 6/10
�[1m375/375�[0m �[32m━━━━━━━━━━━━━━━━━━━━�[0m�[37m�[0m �[1m1s�[0m 3ms/step - accuracy: 0.9053 - loss: 0.3360 - val_accuracy: 0.9162 - val_loss: 0.2943
Epoch 7/10
�[1m375/375�[0m �[32m━━━━━━━━━━━━━━━━━━━━�[0m�[37m�[0m �[1m1s�[0m 3ms/step - accuracy: 0.9109 - loss: 0.3107 - val_accuracy: 0.9195 - val_loss: 0.2811
Epoch 8/10
�[1m375/375�[0m �[32m━━━━━━━━━━━━━━━━━━━━�[0m�[37m�[0m �[1m1s�[0m 3ms/step - accuracy: 0.9162 - loss: 0.2934 - val_accuracy: 0.9224 - val_loss: 0.2699
Epoch 9/10
�[1m375/375�[0m �[32m━━━━━━━━━━━━━━━━━━━━�[0m�[37m�[0m �[1m1s�[0m 3ms/step - accuracy: 0.9199 - loss: 0.2836 - val_accuracy: 0.9252 - val_loss: 0.2593
Epoch 10/10
�[1m375/375�[0m �[32m━━━━━━━━━━━━━━━━━━━━�[0m�[37m�[0m �[1m1s�[0m 3ms/step - accuracy: 0.9227 - loss: 0.2709 - val_accuracy: 0.9279 - val_loss: 0.2506
�[1m313/313�[0m �[32m━━━━━━━━━━━━━━━━━━━━�[0m�[37m�[0m �[1m1s�[0m 2ms/step - accuracy: 0.9157 - loss: 0.2899
Experiment 2 - Test score: 0.24992172420024872
Experiment 2 - Test accuracy: 0.9279000163078308
Experiment 3: Increased Hidden Neurons to 256
NB_EPOCH = 20
BATCH_SIZE = 128
N_HIDDEN = 256
VALIDATION_SPLIT = 0.2
K.clear_session()
model_exp3 = Sequential()
model_exp3.add(Dense(N_HIDDEN, input_shape=(RESHAPED,)))
model_exp3.add(Activation('relu'))
model_exp3.add(Dense(N_HIDDEN))
model_exp3.add(Activation('relu'))
model_exp3.add(Dense(NB_CLASSES))
model_exp3.add(Activation('softmax'))
model_exp3.compile(loss='categorical_crossentropy', optimizer=SGD(), metrics=['accuracy'])
history_exp3 = model_exp3.fit(X_train, Y_train, batch_size=BATCH_SIZE, epochs=NB_EPOCH, verbose=VERBOSE, validation_split=VALIDATION_SPLIT)
score_exp3 = model_exp3.evaluate(X_test, Y_test, verbose=VERBOSE)
print("Experiment 3 - Test score:", score_exp3[0])
print("Experiment 3 - Test accuracy:", score_exp3[1])Epoch 1/20
�[1m375/375�[0m �[32m━━━━━━━━━━━━━━━━━━━━�[0m�[37m�[0m �[1m2s�[0m 4ms/step - accuracy: 0.4952 - loss: 1.8586 - val_accuracy: 0.8568 - val_loss: 0.6780
Epoch 2/20
�[1m375/375�[0m �[32m━━━━━━━━━━━━━━━━━━━━�[0m�[37m�[0m �[1m1s�[0m 4ms/step - accuracy: 0.8578 - loss: 0.6135 - val_accuracy: 0.8911 - val_loss: 0.4237
Epoch 3/20
�[1m375/375�[0m �[32m━━━━━━━━━━━━━━━━━━━━�[0m�[37m�[0m �[1m1s�[0m 4ms/step - accuracy: 0.8863 - loss: 0.4281 - val_accuracy: 0.9025 - val_loss: 0.3557
Epoch 4/20
�[1m375/375�[0m �[32m━━━━━━━━━━━━━━━━━━━━�[0m�[37m�[0m �[1m2s�[0m 4ms/step - accuracy: 0.8989 - loss: 0.3680 - val_accuracy: 0.9113 - val_loss: 0.3196
Epoch 5/20
�[1m375/375�[0m �[32m━━━━━━━━━━━━━━━━━━━━�[0m�[37m�[0m �[1m1s�[0m 4ms/step - accuracy: 0.9084 - loss: 0.3277 - val_accuracy: 0.9172 - val_loss: 0.2969
Epoch 6/20
�[1m375/375�[0m �[32m━━━━━━━━━━━━━━━━━━━━�[0m�[37m�[0m �[1m1s�[0m 3ms/step - accuracy: 0.9140 - loss: 0.3034 - val_accuracy: 0.9211 - val_loss: 0.2798
Epoch 7/20
�[1m375/375�[0m �[32m━━━━━━━━━━━━━━━━━━━━�[0m�[37m�[0m �[1m1s�[0m 3ms/step - accuracy: 0.9182 - loss: 0.2869 - val_accuracy: 0.9258 - val_loss: 0.2654
Epoch 8/20
�[1m375/375�[0m �[32m━━━━━━━━━━━━━━━━━━━━�[0m�[37m�[0m �[1m1s�[0m 3ms/step - accuracy: 0.9229 - loss: 0.2725 - val_accuracy: 0.9280 - val_loss: 0.2548
Epoch 9/20
�[1m375/375�[0m �[32m━━━━━━━━━━━━━━━━━━━━�[0m�[37m�[0m �[1m1s�[0m 3ms/step - accuracy: 0.9252 - loss: 0.2613 - val_accuracy: 0.9317 - val_loss: 0.2457
Epoch 10/20
�[1m375/375�[0m �[32m━━━━━━━━━━━━━━━━━━━━�[0m�[37m�[0m �[1m1s�[0m 3ms/step - accuracy: 0.9284 - loss: 0.2498 - val_accuracy: 0.9333 - val_loss: 0.2349
Epoch 11/20
�[1m375/375�[0m �[32m━━━━━━━━━━━━━━━━━━━━�[0m�[37m�[0m �[1m1s�[0m 4ms/step - accuracy: 0.9305 - loss: 0.2408 - val_accuracy: 0.9356 - val_loss: 0.2278
Epoch 12/20
�[1m375/375�[0m �[32m━━━━━━━━━━━━━━━━━━━━�[0m�[37m�[0m �[1m1s�[0m 3ms/step - accuracy: 0.9316 - loss: 0.2359 - val_accuracy: 0.9381 - val_loss: 0.2215
Epoch 13/20
�[1m375/375�[0m �[32m━━━━━━━━━━━━━━━━━━━━�[0m�[37m�[0m �[1m1s�[0m 4ms/step - accuracy: 0.9375 - loss: 0.2184 - val_accuracy: 0.9409 - val_loss: 0.2138
Epoch 14/20
�[1m375/375�[0m �[32m━━━━━━━━━━━━━━━━━━━━�[0m�[37m�[0m �[1m1s�[0m 4ms/step - accuracy: 0.9373 - loss: 0.2163 - val_accuracy: 0.9426 - val_loss: 0.2073
Epoch 15/20
�[1m375/375�[0m �[32m━━━━━━━━━━━━━━━━━━━━�[0m�[37m�[0m �[1m1s�[0m 4ms/step - accuracy: 0.9408 - loss: 0.2033 - val_accuracy: 0.9455 - val_loss: 0.2003
Epoch 16/20
�[1m375/375�[0m �[32m━━━━━━━━━━━━━━━━━━━━�[0m�[37m�[0m �[1m1s�[0m 4ms/step - accuracy: 0.9415 - loss: 0.1984 - val_accuracy: 0.9467 - val_loss: 0.1955
Epoch 17/20
�[1m375/375�[0m �[32m━━━━━━━━━━━━━━━━━━━━�[0m�[37m�[0m �[1m1s�[0m 4ms/step - accuracy: 0.9436 - loss: 0.1969 - val_accuracy: 0.9478 - val_loss: 0.1910
Epoch 18/20
�[1m375/375�[0m �[32m━━━━━━━━━━━━━━━━━━━━�[0m�[37m�[0m �[1m1s�[0m 4ms/step - accuracy: 0.9484 - loss: 0.1876 - val_accuracy: 0.9498 - val_loss: 0.1857
Epoch 19/20
�[1m375/375�[0m �[32m━━━━━━━━━━━━━━━━━━━━�[0m�[37m�[0m �[1m1s�[0m 4ms/step - accuracy: 0.9482 - loss: 0.1853 - val_accuracy: 0.9503 - val_loss: 0.1820
Epoch 20/20
�[1m375/375�[0m �[32m━━━━━━━━━━━━━━━━━━━━�[0m�[37m�[0m �[1m1s�[0m 4ms/step - accuracy: 0.9466 - loss: 0.1813 - val_accuracy: 0.9519 - val_loss: 0.1782
�[1m313/313�[0m �[32m━━━━━━━━━━━━━━━━━━━━�[0m�[37m�[0m �[1m1s�[0m 2ms/step - accuracy: 0.9393 - loss: 0.2069
Experiment 3 - Test score: 0.17764912545681
Experiment 3 - Test accuracy: 0.9480000138282776
