tf-keras-model.py

import numpy as np
import tensorflow as tf
from tensorflow import keras
from keras.models import Sequential
import h5py
from keras.utils.io_utils import HDF5Matrix

data_dir = '...'
train_fn = data_dir + '/dataset_train.h5'
test_fn = data_dir + '/dataset_test.h5'
image_dims = (64, 64, 3)
train_epochs = 2500

X_train = HDF5Matrix(train_fn, 'train_set_x', normalizer = lambda x: x / 255.0)
y_train = HDF5Matrix(train_fn, 'train_set_y')
X_test = HDF5Matrix(test_fn, 'test_set_x', normalizer = lambda x: x / 255.0)
y_test = HDF5Matrix(test_fn, 'test_set_y')
m_train = X_train.shape[0]
m_test = X_test.shape[0]

#X_train.shape: (209, 64, 64, 3)
#y_train.shape: (209,)
model = tf.keras.Sequential([
  tf.keras.layers.Flatten(input_shape=image_dims),
  tf.keras.layers.Dense(20, activation="relu"),
  tf.keras.layers.Dense(7, activation="relu"),
  tf.keras.layers.Dense(5, activation="relu"),
  tf.keras.layers.Dense(1, activation="sigmoid")
])
from keras import optimizers

model.compile(loss=keras.losses.binary_crossentropy,
              optimizer=keras.optimizers.SGD(lr=0.01),
              metrics=['accuracy'])

model.fit(X_train, y_train, epochs=train_epochs, shuffle="batch", batch_size=m_train) # batch GD

loss_and_metrics = model.evaluate(X_test, y_test, batch_size=m_test)
print("loss and metrics: %s" % loss_and_metrics)

classes = model.predict(X_test)