neural_network 0.1.3

Modular neural network with an implementation of back propagation learning algorithm.
Documentation 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18

use rand::{Rng, thread_rng};
use fast_io::CustomIO;
use std::fmt::Debug;
use super::connection::Connection;

pub trait Neuron: Clone+Debug+PartialEq+CustomIO {
    type C: Connection+Sized;
    fn new(prev_layer_size: usize) -> Self;

    fn bias(&self) -> f64;
    fn output(&self) -> f64;
    fn set_output(&mut self, output: f64);
    fn connections(&mut self) -> &mut Vec<Self::C>;
}

pub fn rand_weight() -> f64 {
    thread_rng().gen_range(-0.33, 0.33)
}