libdt 2.0.3

Library for training neural networks.
Documentation 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173

use nalgebra as na;

use na::DMatrix;
use na::DVector;
use na::Matrix;
use na::Vector;
use na::MatrixView;
use na::base::dimension as dim;
use na::base::VecStorage;

use super::Layer;

pub struct LinLayer<const NEURONS_IN: usize, const NEURONS_OUT: usize> {
    pub signal: Vector::<f64, dim::Dyn,
                         VecStorage::<f64, dim::Dyn, dim::U1>>,
    chain_element: Matrix::<f64, dim::Dyn, dim::Dyn,
                            VecStorage::<f64, dim::Dyn, dim::Dyn>>,
}

impl<const NEURONS_IN: usize,
     const NEURONS_OUT: usize> LinLayer<NEURONS_IN, NEURONS_OUT>
{
    pub fn new() -> Self {
        Self {
            signal: Vector::from_element_generic(dim::Dyn(Self::NEURONS_OUT),
                                                 dim::U1, 0f64),
            chain_element: Matrix::from_element_generic(dim::Dyn(Self::NEURONS_OUT),
                                                        dim::Dyn(Self::NEURONS_IN), 0f64),
        }
    }
}

impl<const NEURONS_IN: usize, const NEURONS_OUT: usize> Layer
    for LinLayer<NEURONS_IN, NEURONS_OUT>
{
    const PARAMS_CNT: usize = NEURONS_IN * NEURONS_OUT + NEURONS_OUT;
    const NEURONS_IN: usize = NEURONS_IN;
    const NEURONS_OUT: usize = NEURONS_OUT;

    unsafe fn eval_unchecked(p: &[f64], x: DVector<f64>) -> DVector<f64> {
        let m = MatrixView::from_slice_generic_unchecked(
            p,
            0,
            dim::Dyn(NEURONS_OUT),
            dim::Dyn(NEURONS_IN),
        );
        let v = MatrixView::from_slice_generic_unchecked(
            p,
            NEURONS_OUT * NEURONS_IN,
            dim::Dyn(NEURONS_OUT),
            dim::Dyn(1),
        );

        &(&m * x) + v
    }

    fn eval(p: &[f64], x: DVector<f64>) -> DVector<f64> {
        assert_eq!(p.len(), Self::PARAMS_CNT);
        assert_eq!(x.len(), Self::NEURONS_IN);

        let m = MatrixView::from_slice_generic(
            p, dim::Dyn(NEURONS_OUT), dim::Dyn(NEURONS_IN));
        let v = MatrixView::from_slice_generic(
            &p[NEURONS_OUT * NEURONS_IN..],
            dim::Dyn(NEURONS_OUT),
            dim::Dyn(1),
        );

        &(&m * x) + v
    }

    fn forward(&mut self, p: &[f64], x: DVector<f64>) -> DVector<f64> {
        assert_eq!(p.len(), Self::PARAMS_CNT);
        assert_eq!(x.len(), Self::NEURONS_IN);

        self.signal = Self::eval(p, x);
        self.signal.clone()
    }

    fn backward(&mut self, p: &[f64]) {
        self.chain_element = MatrixView::from_slice_generic(
            p, dim::Dyn(NEURONS_OUT), dim::Dyn(NEURONS_IN)).into();
    }

    fn chain_element(&self) -> &DMatrix<f64> {
        &self.chain_element
    }

    fn chain_end(&self, x: &DVector<f64>) -> DMatrix<f64>
    {
        let mut matrix: DMatrix<f64> =
            DMatrix::from_element_generic(
            dim::Dyn(Self::NEURONS_OUT),
            dim::Dyn(Self::PARAMS_CNT), 0f64);

        for i in 0..NEURONS_IN {
            for j in 0..NEURONS_OUT {
                matrix[(j, i*NEURONS_OUT+j)] = x[i];
            }
        }
        for i in NEURONS_IN*NEURONS_OUT..Self::PARAMS_CNT {
            matrix[(i - NEURONS_IN*NEURONS_OUT, i)] = 1.;           
        }

        matrix
    }

    fn default_initial_params() -> Vec<f64> {
        use rand::Rng;

        let mut rng = rand::thread_rng();
        let mut p: Vec<f64> =
            Vec::with_capacity(Self::PARAMS_CNT);
        for _ in 0..Self::PARAMS_CNT {
            p.push(rng.gen_range(-1.0..1.0));
        }

        p
    }
}

#[cfg(test)]
mod tests {
    use super::*;

    #[test]
    fn test_eval() {
        let p: [f64; 6] = [1., 1., 1., 0., 7., 7.];
        let x = DVector::from_column_slice(
            na::vector![1f64, 2f64].as_slice());
        
        let y = LinLayer::<2, 2>::eval(&p, x);
        assert_eq!(y, na::vector![10., 8.]);
    }

    #[test]
    fn test_forward() {
        let p: [f64; 6] = [1., 1., 1., 0., 7., 7.];
        let x = DVector::from_column_slice(
            na::vector![1f64, 2f64].as_slice());
        let mut layer = LinLayer::<2, 2>::new();
        
        let y = layer.forward(&p, x);
        assert_eq!(y, na::vector![10., 8.]);
    }

    #[test]
    fn test_backward() {
        let p: [f64; 6] = [1., 1., 1., 0., 7., 7.];
        let x = DVector::from_column_slice(
            na::vector![1f64, 2f64].as_slice());
        let mut layer = LinLayer::<2, 2>::new();
        
        let _ = layer.forward(&p, x);
        layer.backward(&p);

        assert_eq!(*layer.chain_element(),
                   na::matrix![1., 1.;
                               1., 0.]);
    }

    #[test]
    fn test_chain_end() {
        let x = DVector::from_column_slice(
            na::vector![1f64, 2f64].as_slice());
        let layer = LinLayer::<2, 3>::new();

        assert_eq!(layer.chain_end(&x),
                   na::matrix![1., 0., 0., 2., 0., 0., 1., 0., 0.;
                               0., 1., 0., 0., 2., 0., 0., 1., 0.;
                               0., 0., 1., 0., 0., 2., 0., 0., 1.]);
    }
}