pykitml (Python Kit for Machine Learning)

Machine Learning library written in Python and NumPy.

Installation

python3 -m pip install pykitml

Documentation

https://pykitml.readthedocs.io/en/latest/

Demo (MNIST)

Training

importos.pathimportnumpyasnpimportpykitmlaspkfrompykitml.datasetsimportmnist# Download datasetif(notos.path.exists('mnist.pkl')): mnist.get()
# Load datasettraining_data, training_targets, testing_data, testing_targets=mnist.load()
# Create a new neural networkdigit_classifier=pk.NeuralNetwork([784, 100, 10])
# Train itdigit_classifier.train(
training_data=training_data,
targets=training_targets, batch_size=50, epochs=1200, optimizer=pk.Adam(learning_rate=0.012, decay_rate=0.95), testing_data=testing_data, testing_targets=testing_targets,
testing_freq=30,
decay_freq=15
)
# Save itpk.save(digit_classifier, 'digit_classifier_network.pkl')
# Show performanceaccuracy=digit_classifier.accuracy(training_data, training_targets)
print('Train Accuracy:', accuracy) accuracy=digit_classifier.accuracy(testing_data, testing_targets)
print('Test Accuracy:', accuracy)
# Plot performance graphdigit_classifier.plot_performance()
# Show confusion matrixdigit_classifier.confusion_matrix(training_data, training_targets)

Trying the model

importrandomimportnumpyasnpimportmatplotlib.pyplotaspltimportpykitmlaspkfrompykitml.datasetsimportmnist# Load datasettraining_data, training_targets, testing_data, testing_targets=mnist.load()
# Load the trained networkdigit_classifier=pk.load('digit_classifier_network.pkl')
# Pick a random example from testing dataindex=random.randint(0, 9999)
# Show the test data and the labelplt.imshow(training_data[index].reshape(28, 28))
plt.show()
print('Label: ', training_targets[index])
# Show predictiondigit_classifier.feed(training_data[index])
model_output=digit_classifier.get_output_onehot()
print('Predicted: ', model_output)

Performance Graph

Confusion Matrix