TarantulaNN

🕷🕸 Implementation of a Deep Neural Network with flexibles architectures and activation functions

Layers:

• FullyConnectedLayer

Activation functions:

• ReLU (*, Rectified Linear Unit)
• Sigmoid
• Tanh

Training:

• mini-batch gradient descent
• stochastic gradient descent

Cost function:

• Sum of squares

License

This project is licensed under the terms of the MIT license, see LICENSE.