Library for easy implementation of neural networks using Kotlin
This library was created in order to support my final graduation project at FACENS - Faculdade de Engenharia de Sorocaba.
Extend KeuralManager and implement the abstract properties defined by the class, according to your rules, to convert the objects from and to DoubleArrays.
You can also define how many layers the network will have, the activation function, and the learning rate through arguments passed to KeuralManager constructor.
There's a unit test class called KeuralManagerTest where you can find an example usage of KeuralManager.
As you can see by that example, it's possible to get a neural network trained and running in less than 25 lines.