An attempt to create a neural network framework
Clone the repository:
git clone https://github.com/nguyenphuminh/catbrain.git
Install the dependencies:
npm i
Create a Javascript file like this:
const { CatBrain } = require("./index");
// Create a neural network
const neuralNetwork = new CatBrain({
inputAmount: 2, // Amount of input nodes
hiddenAmounts: [3], // Amount of nodes for each hidden layer
outputAmount: 1, // Amount of output nodes
learningRate: 0.02 // Learning rate
});
// Train
neuralNetwork.train(
// Amount of iterations
100000,
// Dataset as an array
[
// A data object with expected output of an input
{ inputs: [0, 0], outputs: [0] },
{ inputs: [0, 1], outputs: [1] },
{ inputs: [1, 0], outputs: [1] },
{ inputs: [1, 1], outputs: [0] }
]
);
// Run the neural net with our own input
console.log(neuralNetwork.feedForward([1, 0]));
There are several examples available in ./examples
:
Current what I have in my mind are:
- Refactor code.
- More activation functions.
- Train with GPU.
- More neural network models?
Copyrights © 2024 Nguyen Phu Minh.
This project is licensed under the GPL 3.0 License.