thomasnield

    thomasnield

    3 years ago
    A neural network built completely from scratch in Kotlin. Instead of using nasty calculus and backpropagation, it uses simple random-based hill climbing or simulated annealing to optimize the node weights/biases.

    https://www.youtube.com/watch?v=WkPQgigeAdc

    These implementations were added to the "Kotlin Machine Learning Demos" project https://github.com/thomasnield/kotlin-machine-learning-demos/blob/master/README.md