Ancez / ai-learning-py

Ai Learning Python

The NeuralNetwork class is a simple implementation of a feedforward neural network with a single hidden layer. The network can be trained using gradient descent and backpropagation to make predictions on new data.

Installation

To use the NeuralNetwork class, simply clone the repository and import the neural_network.py module:

git clone https://github.com/Ancez/ai-learning-py.git

cd ai-learning-py

from neural_network import NeuralNetwork

Usage To create a new NeuralNetwork instance, use the NeuralNetwork constructor and provide the following parameters:

• num_inputs: The number of input nodes in the network.
• num_hidden: The number of hidden nodes in the network.
• num_outputs: The number of output nodes in the network.
• learning_rate: The learning rate for gradient descent.

nn = NeuralNetwork(num_inputs=2, num_hidden=3, num_outputs=1, learning_rate=0.1)

To train the network on some data, use the train method and provide the input data and the corresponding target values:

data = [
  (np.array([[1, 2]]).T, np.array([[0]])),
  (np.array([[3, 4]]).T, np.array([[1]])),
  (np.array([[5, 6]]).T, np.array([[1]]))
]

for inputs, targets in data:
  nn.train(inputs, targets)

To use the network to make predictions on new data, use the predict method and provide the input data:

test_data = [
  (np.array([[1, 2]]).T, np.array([[0]])),
  (np.array([[3, 4]]).T, np.array([[1]])),
  (np.array([[5, 6]]).T, np.array([[1]]))
]

for inputs, targets in test_data:
  prediction = nn.predict(inputs)
  print(prediction)

Contributing

If you find any bugs or have any suggestions for improvements, please open an issue or submit a pull request. Any contributions are welcome!

License

This project is not licensed. Use at your own risk.