pygad.kerasga Module

This section of the PyGAD’s library documentation discusses the pygad.kerasga module.

The pygad.kerarsga module has helper a class and 2 functions to train Keras models using the genetic algorithm (PyGAD). The Keras model can be built either using the Sequential Model or the Functional API.

The contents of this module are:

  1. KerasGA: A class for creating an initial population of all parameters in the Keras model.

  2. model_weights_as_vector(): A function to reshape the Keras model weights to a single vector.

  3. model_weights_as_matrix(): A function to restore the Keras model weights from a vector.

  4. predict(): A function to make predictions based on the Keras model and a solution.

More details are given in the next sections.

Steps Summary

The summary of the steps used to train a Keras model using PyGAD is as follows:

  1. Create a Keras model.

  2. Create an instance of the pygad.kerasga.KerasGA class.

  3. Prepare the training data.

  4. Build the fitness function.

  5. Create an instance of the pygad.GA class.

  6. Run the genetic algorithm.

Create Keras Model

Before discussing training a Keras model using PyGAD, the first thing to do is to create the Keras model.

According to the Keras library documentation, there are 3 ways to build a Keras model:

  1. Sequential Model

  2. Functional API

  3. Model Subclassing

PyGAD supports training the models created either using the Sequential Model or the Functional API.

Here is an example of a model created using the Sequential Model.

import tensorflow.keras

input_layer  = tensorflow.keras.layers.Input(3)
dense_layer1 = tensorflow.keras.layers.Dense(5, activation="relu")
output_layer = tensorflow.keras.layers.Dense(1, activation="linear")

model = tensorflow.keras.Sequential()
model.add(input_layer)
model.add(dense_layer1)
model.add(output_layer)

This is the same model created using the Functional API.

input_layer  = tensorflow.keras.layers.Input(3)
dense_layer1 = tensorflow.keras.layers.Dense(5, activation="relu")(input_layer)
output_layer = tensorflow.keras.layers.Dense(1, activation="linear")(dense_layer1)

model = tensorflow.keras.Model(inputs=input_layer, outputs=output_layer)

Feel free to add the layers of your choice.

pygad.kerasga.KerasGA Class

The pygad.kerasga module has a class named KerasGA for creating an initial population for the genetic algorithm based on a Keras model. The constructor, methods, and attributes within the class are discussed in this section.

__init__()

The pygad.kerasga.KerasGA class constructor accepts the following parameters:

  • model: An instance of the Keras model.

  • num_solutions: Number of solutions in the population. Each solution has different parameters of the model.

Instance Attributes

All parameters in the pygad.kerasga.KerasGA class constructor are used as instance attributes in addition to adding a new attribute called population_weights.

Here is a list of all instance attributes:

  • model

  • num_solutions

  • population_weights: A nested list holding the weights of all solutions in the population.

Methods in the KerasGA Class

This section discusses the methods available for instances of the pygad.kerasga.KerasGA class.

create_population()

The create_population() method creates the initial population of the genetic algorithm as a list of solutions where each solution represents different model parameters. The list of networks is assigned to the population_weights attribute of the instance.

Functions in the pygad.kerasga Module

This section discusses the functions in the pygad.kerasga module.

pygad.kerasga.model_weights_as_vector()

The model_weights_as_vector() function accepts a single parameter named model representing the Keras model. It returns a vector holding all model weights. The reason for representing the model weights as a vector is that the genetic algorithm expects all parameters of any solution to be in a 1D vector form.

This function filters the layers based on the trainable attribute to see whether the layer weights are trained or not. For each layer, if its trainable=False, then its weights will not be evolved using the genetic algorithm. Otherwise, it will be represented in the chromosome and evolved.

The function accepts the following parameters:

  • model: The Keras model.

It returns a 1D vector holding the model weights.

pygad.kerasga.model_weights_as_matrix()

The model_weights_as_matrix() function accepts the following parameters:

  1. model: The Keras model.

  2. weights_vector: The model parameters as a vector.

It returns the restored model weights after reshaping the vector.

pygad.kerasga.predict()

The predict() function makes a prediction based on a solution. It accepts the following parameters:

  1. model: The Keras model.

  2. solution: The solution evolved.

  3. data: The test data inputs.

  4. batch_size=None: The batch size (i.e. number of samples per step or batch).

  5. verbose=None: Verbosity mode.

  6. steps=None: The total number of steps (batches of samples).

Check documentation of the Keras Model.predict() method for more information about the batch_size, verbose, and steps parameters.

It returns the predictions of the data samples.

Examples

This section gives the complete code of some examples that build and train a Keras model using PyGAD. Each subsection builds a different network.

Example 1: Regression Example

The next code builds a simple Keras model for regression. The next subsections discuss each part in the code.

import tensorflow.keras
import pygad.kerasga
import numpy
import pygad

def fitness_func(ga_instance, solution, sol_idx):
    global data_inputs, data_outputs, keras_ga, model

    predictions = pygad.kerasga.predict(model=model,
                                        solution=solution,
                                        data=data_inputs)

    mae = tensorflow.keras.losses.MeanAbsoluteError()
    abs_error = mae(data_outputs, predictions).numpy() + 0.00000001
    solution_fitness = 1.0/abs_error

    return solution_fitness

def on_generation(ga_instance):
    print(f"Generation = {ga_instance.generations_completed}")
    print(f"Fitness    = {ga_instance.best_solution()[1]}")

input_layer  = tensorflow.keras.layers.Input(3)
dense_layer1 = tensorflow.keras.layers.Dense(5, activation="relu")(input_layer)
output_layer = tensorflow.keras.layers.Dense(1, activation="linear")(dense_layer1)

model = tensorflow.keras.Model(inputs=input_layer, outputs=output_layer)

keras_ga = pygad.kerasga.KerasGA(model=model,
                                 num_solutions=10)

# Data inputs
data_inputs = numpy.array([[0.02, 0.1, 0.15],
                           [0.7, 0.6, 0.8],
                           [1.5, 1.2, 1.7],
                           [3.2, 2.9, 3.1]])

# Data outputs
data_outputs = numpy.array([[0.1],
                            [0.6],
                            [1.3],
                            [2.5]])

# Prepare the PyGAD parameters. Check the documentation for more information: https://pygad.readthedocs.io/en/latest/pygad.html#pygad-ga-class
num_generations = 250 # Number of generations.
num_parents_mating = 5 # Number of solutions to be selected as parents in the mating pool.
initial_population = keras_ga.population_weights # Initial population of network weights

ga_instance = pygad.GA(num_generations=num_generations,
                       num_parents_mating=num_parents_mating,
                       initial_population=initial_population,
                       fitness_func=fitness_func,
                       on_generation=on_generation)

ga_instance.run()

# After the generations complete, some plots are showed that summarize how the outputs/fitness values evolve over generations.
ga_instance.plot_fitness(title="PyGAD & Keras - Iteration vs. Fitness", linewidth=4)

# Returning the details of the best solution.
solution, solution_fitness, solution_idx = ga_instance.best_solution()
print(f"Fitness value of the best solution = {solution_fitness}")
print(f"Index of the best solution : {solution_idx}")

# Make prediction based on the best solution.
predictions = pygad.kerasga.predict(model=model,
                                    solution=solution,
                                    data=data_inputs)
print(f"Predictions : \n{predictions}")

mae = tensorflow.keras.losses.MeanAbsoluteError()
abs_error = mae(data_outputs, predictions).numpy()
print(f"Absolute Error : {abs_error}")

Create a Keras Model

According to the steps mentioned previously, the first step is to create a Keras model. Here is the code that builds the model using the Functional API.

import tensorflow.keras

input_layer  = tensorflow.keras.layers.Input(3)
dense_layer1 = tensorflow.keras.layers.Dense(5, activation="relu")(input_layer)
output_layer = tensorflow.keras.layers.Dense(1, activation="linear")(dense_layer1)

model = tensorflow.keras.Model(inputs=input_layer, outputs=output_layer)

The model can also be build using the Keras Sequential Model API.

input_layer  = tensorflow.keras.layers.Input(3)
dense_layer1 = tensorflow.keras.layers.Dense(5, activation="relu")
output_layer = tensorflow.keras.layers.Dense(1, activation="linear")

model = tensorflow.keras.Sequential()
model.add(input_layer)
model.add(dense_layer1)
model.add(output_layer)

Create an Instance of the pygad.kerasga.KerasGA Class

The second step is to create an instance of the pygad.kerasga.KerasGA class. There are 10 solutions per population. Change this number according to your needs.

import pygad.kerasga

keras_ga = pygad.kerasga.KerasGA(model=model,
                                 num_solutions=10)

Prepare the Training Data

The third step is to prepare the training data inputs and outputs. Here is an example where there are 4 samples. Each sample has 3 inputs and 1 output.

import numpy

# Data inputs
data_inputs = numpy.array([[0.02, 0.1, 0.15],
                           [0.7, 0.6, 0.8],
                           [1.5, 1.2, 1.7],
                           [3.2, 2.9, 3.1]])

# Data outputs
data_outputs = numpy.array([[0.1],
                            [0.6],
                            [1.3],
                            [2.5]])

Build the Fitness Function

The fourth step is to build the fitness function. This function must accept 2 parameters representing the solution and its index within the population.

The next fitness function returns the model predictions based on the current solution using the predict() function. Then, it calculates the mean absolute error (MAE) of the Keras model based on the parameters in the solution. The reciprocal of the MAE is used as the fitness value. Feel free to use any other loss function to calculate the fitness value.

def fitness_func(ga_instance, solution, sol_idx):
    global data_inputs, data_outputs, keras_ga, model

    predictions = pygad.kerasga.predict(model=model,
                                        solution=solution,
                                        data=data_inputs)

    mae = tensorflow.keras.losses.MeanAbsoluteError()
    abs_error = mae(data_outputs, predictions).numpy() + 0.00000001
    solution_fitness = 1.0/abs_error

    return solution_fitness

Create an Instance of the pygad.GA Class

The fifth step is to instantiate the pygad.GA class. Note how the initial_population parameter is assigned to the initial weights of the Keras models.

For more information, please check the parameters this class accepts.

# Prepare the PyGAD parameters. Check the documentation for more information: https://pygad.readthedocs.io/en/latest/pygad.html#pygad-ga-class
num_generations = 250 # Number of generations.
num_parents_mating = 5 # Number of solutions to be selected as parents in the mating pool.
initial_population = keras_ga.population_weights # Initial population of network weights

ga_instance = pygad.GA(num_generations=num_generations,
                       num_parents_mating=num_parents_mating,
                       initial_population=initial_population,
                       fitness_func=fitness_func,
                       on_generation=on_generation)

Run the Genetic Algorithm

The sixth and last step is to run the genetic algorithm by calling the run() method.

ga_instance.run()

After the PyGAD completes its execution, then there is a figure that shows how the fitness value changes by generation. Call the plot_fitness() method to show the figure.

ga_instance.plot_fitness(title="PyGAD & Keras - Iteration vs. Fitness", linewidth=4)

Here is the figure.

To get information about the best solution found by PyGAD, use the best_solution() method.

# Returning the details of the best solution.
solution, solution_fitness, solution_idx = ga_instance.best_solution()
print(f"Fitness value of the best solution = {solution_fitness}")
print(f"Index of the best solution : {solution_idx}")

Fitness value of the best solution = 72.77768757825352
Index of the best solution : 0

The next code makes prediction using the predict() function to return the model predictions based on the best solution.

# Fetch the parameters of the best solution.
predictions = pygad.kerasga.predict(model=model,
                                    solution=solution,
                                    data=data_inputs)
print(f"Predictions : \n{predictions}")

Predictions :
[[0.09935353]
 [0.63082725]
 [1.2765523 ]
 [2.4999595 ]]

The next code measures the trained model error.

mae = tensorflow.keras.losses.MeanAbsoluteError()
abs_error = mae(data_outputs, predictions).numpy()
print(f"Absolute Error : {abs_error}")

Absolute Error :  0.013740465

Example 2: XOR Binary Classification

The next code creates a Keras model to build the XOR binary classification problem. Let’s highlight the changes compared to the previous example.

import tensorflow.keras
import pygad.kerasga
import numpy
import pygad

def fitness_func(ga_instance, solution, sol_idx):
    global data_inputs, data_outputs, keras_ga, model

    predictions = pygad.kerasga.predict(model=model,
                                        solution=solution,
                                        data=data_inputs)

    bce = tensorflow.keras.losses.BinaryCrossentropy()
    solution_fitness = 1.0 / (bce(data_outputs, predictions).numpy() + 0.00000001)

    return solution_fitness

def on_generation(ga_instance):
    print(f"Generation = {ga_instance.generations_completed}")
    print(f"Fitness    = {ga_instance.best_solution()[1]}")

# Build the keras model using the functional API.
input_layer  = tensorflow.keras.layers.Input(2)
dense_layer = tensorflow.keras.layers.Dense(4, activation="relu")(input_layer)
output_layer = tensorflow.keras.layers.Dense(2, activation="softmax")(dense_layer)

model = tensorflow.keras.Model(inputs=input_layer, outputs=output_layer)

# Create an instance of the pygad.kerasga.KerasGA class to build the initial population.
keras_ga = pygad.kerasga.KerasGA(model=model,
                                 num_solutions=10)

# XOR problem inputs
data_inputs = numpy.array([[0, 0],
                           [0, 1],
                           [1, 0],
                           [1, 1]])

# XOR problem outputs
data_outputs = numpy.array([[1, 0],
                            [0, 1],
                            [0, 1],
                            [1, 0]])

# Prepare the PyGAD parameters. Check the documentation for more information: https://pygad.readthedocs.io/en/latest/pygad.html#pygad-ga-class
num_generations = 250 # Number of generations.
num_parents_mating = 5 # Number of solutions to be selected as parents in the mating pool.
initial_population = keras_ga.population_weights # Initial population of network weights.

# Create an instance of the pygad.GA class
ga_instance = pygad.GA(num_generations=num_generations,
                       num_parents_mating=num_parents_mating,
                       initial_population=initial_population,
                       fitness_func=fitness_func,
                       on_generation=on_generation)

# Start the genetic algorithm evolution.
ga_instance.run()

# After the generations complete, some plots are showed that summarize how the outputs/fitness values evolve over generations.
ga_instance.plot_fitness(title="PyGAD & Keras - Iteration vs. Fitness", linewidth=4)

# Returning the details of the best solution.
solution, solution_fitness, solution_idx = ga_instance.best_solution()
print(f"Fitness value of the best solution = {solution_fitness}")
print(f"Index of the best solution : {solution_idx}")

# Make predictions based on the best solution.
predictions = pygad.kerasga.predict(model=model,
                                    solution=solution,
                                    data=data_inputs)
print(f"Predictions : \n{predictions}")

# Calculate the binary crossentropy for the trained model.
bce = tensorflow.keras.losses.BinaryCrossentropy()
print("Binary Crossentropy : ", bce(data_outputs, predictions).numpy())

# Calculate the classification accuracy for the trained model.
ba = tensorflow.keras.metrics.BinaryAccuracy()
ba.update_state(data_outputs, predictions)
accuracy = ba.result().numpy()
print(f"Accuracy : {accuracy}")

Compared to the previous regression example, here are the changes:

  • The Keras model is changed according to the nature of the problem. Now, it has 2 inputs and 2 outputs with an in-between hidden layer of 4 neurons.

# Build the keras model using the functional API.
input_layer  = tensorflow.keras.layers.Input(2)
dense_layer = tensorflow.keras.layers.Dense(4, activation="relu")(input_layer)
output_layer = tensorflow.keras.layers.Dense(2, activation="softmax")(dense_layer)

model = tensorflow.keras.Model(inputs=input_layer, outputs=output_layer)

  • The train data is changed. Note that the output of each sample is a 1D vector of 2 values, 1 for each class.

# XOR problem inputs
data_inputs = numpy.array([[0, 0],
                           [0, 1],
                           [1, 0],
                           [1, 1]])

# XOR problem outputs
data_outputs = numpy.array([[1, 0],
                            [0, 1],
                            [0, 1],
                            [1, 0]])

  • The fitness value is calculated based on the binary cross entropy.

bce = tensorflow.keras.losses.BinaryCrossentropy()
solution_fitness = 1.0 / (bce(data_outputs, predictions).numpy() + 0.00000001)

After the previous code completes, the next figure shows how the fitness value change by generation.

Here is some information about the trained model. Its fitness value is 739.24, loss is 0.0013527311 and accuracy is 100%.

Fitness value of the best solution = 739.2397344644013
Index of the best solution : 7

Predictions :
[[9.9694413e-01 3.0558957e-03]
 [5.0176249e-04 9.9949825e-01]
 [1.8470541e-03 9.9815291e-01]
 [9.9999976e-01 2.0538971e-07]]

Binary Crossentropy :  0.0013527311

Accuracy :  1.0

Example 3: Image Multi-Class Classification (Dense Layers)

Here is the code.

import tensorflow.keras
import pygad.kerasga
import numpy
import pygad

def fitness_func(ga_instance, solution, sol_idx):
    global data_inputs, data_outputs, keras_ga, model

    predictions = pygad.kerasga.predict(model=model,
                                        solution=solution,
                                        data=data_inputs)

    cce = tensorflow.keras.losses.CategoricalCrossentropy()
    solution_fitness = 1.0 / (cce(data_outputs, predictions).numpy() + 0.00000001)

    return solution_fitness

def on_generation(ga_instance):
    print(f"Generation = {ga_instance.generations_completed}")
    print(f"Fitness    = {ga_instance.best_solution()[1]}")

# Build the keras model using the functional API.
input_layer  = tensorflow.keras.layers.Input(360)
dense_layer = tensorflow.keras.layers.Dense(50, activation="relu")(input_layer)
output_layer = tensorflow.keras.layers.Dense(4, activation="softmax")(dense_layer)

model = tensorflow.keras.Model(inputs=input_layer, outputs=output_layer)

# Create an instance of the pygad.kerasga.KerasGA class to build the initial population.
keras_ga = pygad.kerasga.KerasGA(model=model,
                                   num_solutions=10)

# Data inputs
data_inputs = numpy.load("../data/dataset_features.npy")

# Data outputs
data_outputs = numpy.load("../data/outputs.npy")
data_outputs = tensorflow.keras.utils.to_categorical(data_outputs)

# Prepare the PyGAD parameters. Check the documentation for more information: https://pygad.readthedocs.io/en/latest/pygad.html#pygad-ga-class
num_generations = 100 # Number of generations.
num_parents_mating = 5 # Number of solutions to be selected as parents in the mating pool.
initial_population = keras_ga.population_weights # Initial population of network weights.

# Create an instance of the pygad.GA class
ga_instance = pygad.GA(num_generations=num_generations,
                       num_parents_mating=num_parents_mating,
                       initial_population=initial_population,
                       fitness_func=fitness_func,
                       on_generation=on_generation)

# Start the genetic algorithm evolution.
ga_instance.run()

# After the generations complete, some plots are showed that summarize how the outputs/fitness values evolve over generations.
ga_instance.plot_fitness(title="PyGAD & Keras - Iteration vs. Fitness", linewidth=4)

# Returning the details of the best solution.
solution, solution_fitness, solution_idx = ga_instance.best_solution()
print(f"Fitness value of the best solution = {solution_fitness}")
print(f"Index of the best solution : {solution_idx}")

# Make predictions based on the best solution.
predictions = pygad.kerasga.predict(model=model,
                                    solution=solution,
                                    data=data_inputs)
# print(f"Predictions : \n{predictions}")

# Calculate the categorical crossentropy for the trained model.
cce = tensorflow.keras.losses.CategoricalCrossentropy()
print(f"Categorical Crossentropy : {cce(data_outputs, predictions).numpy()}")

# Calculate the classification accuracy for the trained model.
ca = tensorflow.keras.metrics.CategoricalAccuracy()
ca.update_state(data_outputs, predictions)
accuracy = ca.result().numpy()
print(f"Accuracy : {accuracy}")

Compared to the previous binary classification example, this example has multiple classes (4) and thus the loss is measured using categorical cross entropy.

cce = tensorflow.keras.losses.CategoricalCrossentropy()
solution_fitness = 1.0 / (cce(data_outputs, predictions).numpy() + 0.00000001)

Prepare the Training Data

Before building and training neural networks, the training data (input and output) needs to be prepared. The inputs and the outputs of the training data are NumPy arrays.

The data used in this example is available as 2 files:

  1. dataset_features.npy: Data inputs. https://github.com/ahmedfgad/NumPyANN/blob/master/dataset_features.npy

  2. outputs.npy: Class labels. https://github.com/ahmedfgad/NumPyANN/blob/master/outputs.npy

The data consists of 4 classes of images. The image shape is (100, 100, 3). The number of training samples is 1962. The feature vector extracted from each image has a length 360.

Simply download these 2 files and read them according to the next code. Note that the class labels are one-hot encoded using the tensorflow.keras.utils.to_categorical() function.

import numpy

data_inputs = numpy.load("../data/dataset_features.npy")

data_outputs = numpy.load("../data/outputs.npy")
data_outputs = tensorflow.keras.utils.to_categorical(data_outputs)

The next figure shows how the fitness value changes.

Here are some statistics about the trained model.

Fitness value of the best solution = 4.197464252185969
Index of the best solution : 0
Categorical Crossentropy :  0.23823906
Accuracy :  0.9852192

Example 4: Image Multi-Class Classification (Conv Layers)

Compared to the previous example that uses only dense layers, this example uses convolutional layers to classify the same dataset.

Here is the complete code.

import tensorflow.keras
import pygad.kerasga
import numpy
import pygad

def fitness_func(ga_instance, solution, sol_idx):
    global data_inputs, data_outputs, keras_ga, model

    predictions = pygad.kerasga.predict(model=model,
                                        solution=solution,
                                        data=data_inputs)

    cce = tensorflow.keras.losses.CategoricalCrossentropy()
    solution_fitness = 1.0 / (cce(data_outputs, predictions).numpy() + 0.00000001)

    return solution_fitness

def on_generation(ga_instance):
    print(f"Generation = {ga_instance.generations_completed}")
    print(f"Fitness    = {ga_instance.best_solution()[1]}")

# Build the keras model using the functional API.
input_layer = tensorflow.keras.layers.Input(shape=(100, 100, 3))
conv_layer1 = tensorflow.keras.layers.Conv2D(filters=5,
                                             kernel_size=7,
                                             activation="relu")(input_layer)
max_pool1 = tensorflow.keras.layers.MaxPooling2D(pool_size=(5,5),
                                                 strides=5)(conv_layer1)
conv_layer2 = tensorflow.keras.layers.Conv2D(filters=3,
                                             kernel_size=3,
                                             activation="relu")(max_pool1)
flatten_layer  = tensorflow.keras.layers.Flatten()(conv_layer2)
dense_layer = tensorflow.keras.layers.Dense(15, activation="relu")(flatten_layer)
output_layer = tensorflow.keras.layers.Dense(4, activation="softmax")(dense_layer)

model = tensorflow.keras.Model(inputs=input_layer, outputs=output_layer)

# Create an instance of the pygad.kerasga.KerasGA class to build the initial population.
keras_ga = pygad.kerasga.KerasGA(model=model,
                                 num_solutions=10)

# Data inputs
data_inputs = numpy.load("../data/dataset_inputs.npy")

# Data outputs
data_outputs = numpy.load("../data/dataset_outputs.npy")
data_outputs = tensorflow.keras.utils.to_categorical(data_outputs)

# Prepare the PyGAD parameters. Check the documentation for more information: https://pygad.readthedocs.io/en/latest/pygad.html#pygad-ga-class
num_generations = 200 # Number of generations.
num_parents_mating = 5 # Number of solutions to be selected as parents in the mating pool.
initial_population = keras_ga.population_weights # Initial population of network weights.

# Create an instance of the pygad.GA class
ga_instance = pygad.GA(num_generations=num_generations,
                       num_parents_mating=num_parents_mating,
                       initial_population=initial_population,
                       fitness_func=fitness_func,
                       on_generation=on_generation)

# Start the genetic algorithm evolution.
ga_instance.run()

# After the generations complete, some plots are showed that summarize how the outputs/fitness values evolve over generations.
ga_instance.plot_fitness(title="PyGAD & Keras - Iteration vs. Fitness", linewidth=4)

# Returning the details of the best solution.
solution, solution_fitness, solution_idx = ga_instance.best_solution()
print(f"Fitness value of the best solution = {solution_fitness}")
print(f"Index of the best solution : {solution_idx}")

# Make predictions based on the best solution.
predictions = pygad.kerasga.predict(model=model,
                                    solution=solution,
                                    data=data_inputs)
# print(f"Predictions : \n{predictions}")

# Calculate the categorical crossentropy for the trained model.
cce = tensorflow.keras.losses.CategoricalCrossentropy()
print(f"Categorical Crossentropy : {cce(data_outputs, predictions).numpy()}")

# Calculate the classification accuracy for the trained model.
ca = tensorflow.keras.metrics.CategoricalAccuracy()
ca.update_state(data_outputs, predictions)
accuracy = ca.result().numpy()
print(f"Accuracy : {accuracy}")

Compared to the previous example, the only change is that the architecture uses convolutional and max-pooling layers. The shape of each input sample is 100x100x3.

# Build the keras model using the functional API.
input_layer = tensorflow.keras.layers.Input(shape=(100, 100, 3))
conv_layer1 = tensorflow.keras.layers.Conv2D(filters=5,
                                             kernel_size=7,
                                             activation="relu")(input_layer)
max_pool1 = tensorflow.keras.layers.MaxPooling2D(pool_size=(5,5),
                                                 strides=5)(conv_layer1)
conv_layer2 = tensorflow.keras.layers.Conv2D(filters=3,
                                             kernel_size=3,
                                             activation="relu")(max_pool1)
flatten_layer  = tensorflow.keras.layers.Flatten()(conv_layer2)
dense_layer = tensorflow.keras.layers.Dense(15, activation="relu")(flatten_layer)
output_layer = tensorflow.keras.layers.Dense(4, activation="softmax")(dense_layer)

model = tensorflow.keras.Model(inputs=input_layer, outputs=output_layer)

Prepare the Training Data

The data used in this example is available as 2 files:

  1. dataset_inputs.npy: Data inputs. https://github.com/ahmedfgad/NumPyCNN/blob/master/dataset_inputs.npy

  2. dataset_outputs.npy: Class labels. https://github.com/ahmedfgad/NumPyCNN/blob/master/dataset_outputs.npy

The data consists of 4 classes of images. The image shape is (100, 100, 3) and there are 20 images per class for a total of 80 training samples. For more information about the dataset, check the Reading the Data section of the pygad.cnn module.

Simply download these 2 files and read them according to the next code. Note that the class labels are one-hot encoded using the tensorflow.keras.utils.to_categorical() function.

import numpy

data_inputs = numpy.load("../data/dataset_inputs.npy")

data_outputs = numpy.load("../data/dataset_outputs.npy")
data_outputs = tensorflow.keras.utils.to_categorical(data_outputs)

The next figure shows how the fitness value changes.

Here are some statistics about the trained model. The model accuracy is 75% after the 200 generations. Note that just running the code again may give different results.

Fitness value of the best solution = 2.7462310258668805
Index of the best solution : 0
Categorical Crossentropy :  0.3641354
Accuracy :  0.75

To improve the model performance, you can do the following:

  • Add more layers

  • Modify the existing layers.

  • Use different parameters for the layers.

  • Use different parameters for the genetic algorithm (e.g. number of solution, number of generations, etc)

Example 5: Image Classification using Data Generator

This example uses the image data generator tensorflow.keras.preprocessing.image.ImageDataGenerator to feed data to the model. Instead of reading all the data in the memory, the data generator generates the data needed by the model and only save it in the memory instead of saving all the data. This frees the memory but adds more computational time.

import tensorflow as tf
import tensorflow.keras
import pygad.kerasga
import pygad

def fitness_func(ga_instanse, solution, sol_idx):
    global train_generator, data_outputs, keras_ga, model

    predictions = pygad.kerasga.predict(model=model,
                                        solution=solution,
                                        data=train_generator)

    cce = tensorflow.keras.losses.CategoricalCrossentropy()
    solution_fitness = 1.0 / (cce(data_outputs, predictions).numpy() + 0.00000001)

    return solution_fitness

def on_generation(ga_instance):
    print("Generation = {ga_instance.generations_completed}")
    print("Fitness    = {ga_instance.best_solution(ga_instance.last_generation_fitness)[1]}")

# The dataset path.
dataset_path = r'../data/Skin_Cancer_Dataset'

num_classes = 2
img_size = 224

# Create a simple CNN. This does not gurantee high classification accuracy.
model = tf.keras.models.Sequential()
model.add(tf.keras.layers.Input(shape=(img_size, img_size, 3)))
model.add(tf.keras.layers.Conv2D(32, (3,3), activation="relu", padding="same"))
model.add(tf.keras.layers.MaxPooling2D((2, 2)))
model.add(tf.keras.layers.Flatten())
model.add(tf.keras.layers.Dropout(rate=0.2))
model.add(tf.keras.layers.Dense(num_classes, activation="softmax"))

# Create an instance of the pygad.kerasga.KerasGA class to build the initial population.
keras_ga = pygad.kerasga.KerasGA(model=model,
                                 num_solutions=10)

data_generator = tf.keras.preprocessing.image.ImageDataGenerator()
train_generator = data_generator.flow_from_directory(dataset_path,
                                                     class_mode='categorical',
                                                     target_size=(224, 224),
                                                     batch_size=32,
                                                     shuffle=False)
# train_generator.class_indices
data_outputs = tf.keras.utils.to_categorical(train_generator.labels)

# Check the documentation for more information about the parameters: https://pygad.readthedocs.io/en/latest/pygad.html#pygad-ga-class
initial_population = keras_ga.population_weights # Initial population of network weights.

# Create an instance of the pygad.GA class
ga_instance = pygad.GA(num_generations=10,
                       num_parents_mating=5,
                       initial_population=initial_population,
                       fitness_func=fitness_func,
                       on_generation=on_generation)

# Start the genetic algorithm evolution.
ga_instance.run()

# After the generations complete, some plots are showed that summarize how the outputs/fitness values evolve over generations.
ga_instance.plot_fitness(title="PyGAD & Keras - Iteration vs. Fitness", linewidth=4)

# Returning the details of the best solution.
solution, solution_fitness, solution_idx = ga_instance.best_solution(ga_instance.last_generation_fitness)
print(f"Fitness value of the best solution = {solution_fitness}")
print(f"Index of the best solution : {solution_idx}")

predictions = pygad.kerasga.predict(model=model,
                                    solution=solution,
                                    data=train_generator)
# print(f"Predictions : \n{predictions}")

# Calculate the categorical crossentropy for the trained model.
cce = tensorflow.keras.losses.CategoricalCrossentropy()
print(f"Categorical Crossentropy : {cce(data_outputs, predictions).numpy()}")

# Calculate the classification accuracy for the trained model.
ca = tensorflow.keras.metrics.CategoricalAccuracy()
ca.update_state(data_outputs, predictions)
accuracy = ca.result().numpy()
print(f"Accuracy : {accuracy}")