sweet-perceptron 1.0.1

A simple neural network library

sweet-perceptron

This is a simple neural network library.

The model of NN is configurable multilayer Perceptron (MLP).

It was built only using NumPy.

Installation

Use the package manager pip to install sweet-perceptron.

pip install sweet-perceptron

Usage

from perceptron import NeuralNetwork
from perceptron.Neuron import functions

# Initialize network
nw = NeuralNetwork((2, 4, 1))

# Change activation functions
nw.activation_funcs = functions.relu, functions.tanh

# Initialize weights
nw.initialize_weights()

# Train netwotk and get list with losses
loss = nw.train(50, ['train inputs set'], ['train outputs set'])

# Get output of forward propagation
output = nw.forward(['input'])

License

MIT License