tensorrs 0.3.2

Tensors is a lightweight machine learning library in Rust
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

use crate::activation::Function;
use crate::linalg::Matrix;
use crate::Float;

/// Scaled Exponential Linear Unit (SELU).
///
/// Applies the SELU activation function element-wise.
/// SELU is defined with a default scale parameter `λ` and a default alpha parameter `α`.
///
/// # Mathematical Definition
/// For an input `x`, the SELU function is defined as:
///```math
/// SELU(x) = \left\{
/// \begin{array}{ll}
/// \lambda x & \text{if } x > 0 \\
/// \lambda \alpha \left( e^x - 1 \right) & \text{if } x \leq 0
/// \end{array}
/// \right.
/// ```
///
/// By default, the parameters are set to:
/// - α = 1.67326
/// - λ = 1.0507
///
/// # See Also
/// - [velog.io: Scaled Exponential Linear Unit](https://velog.io/@greensox284/Activation-Scaled-Exponential-Linear-Unit-SELU)
pub struct SELU<T: Float> {
    alpha: T,
    lambda: T,
}

impl<T: Float> SELU<T> {
    pub fn new(_datatype_number: T) -> Self {
        let alpha: T = T::selu_alpha(T::default());
        let lambda: T = T::selu_lambda(T::default());
        Self { alpha, lambda }
    }

    fn selu_num(&self, x: T) -> T {
        let one: T = 1.into();
        if x > T::default() {
            self.lambda * x
        } else {
            self.lambda * self.alpha * (x.exp() - one)
        }
    }

    // Maybe wrong
    fn selu_der(&self, x: T) -> T {
        if x > T::default() {
            self.lambda
        } else {
            self.lambda * self.alpha * x.exp()
        }
    }
}

impl<T: Float> From<(T, T)> for SELU<T> {
    fn from(params: (T, T)) -> Self {
        let (alpha, scale) = params;
        Self {
            alpha,
            lambda: scale,
        }
    }
}

impl<T: Float> Function<T> for SELU<T> {
    fn name(&self) -> String {
        String::from("SELU")
    }

    fn call(&self, matrix: Matrix<T>) -> Matrix<T> {
        let [row, cols] = [matrix.rows, matrix.cols];
        let mut data = Vec::with_capacity(row * cols);
        for i in matrix.data {
            let num = self.selu_num(i);
            data.push(num);
        }
        Matrix::new(data, row, cols)
    }
    ///# Derivative of SELU
    ///```math
    /// SELU'(x) = \left\{
    /// \begin{array}{ll}
    /// \lambda & \text{if } x > 0 \\
    /// \lambda \alpha e^x & \text{if } x \leq 0
    /// \end{array}
    /// \right.
    /// ```
    fn derivative(&self, matrix: Matrix<T>) -> Matrix<T> {
        let [row, cols] = [matrix.rows, matrix.cols];
        let mut data = Vec::with_capacity(row * cols);
        for i in matrix.data {
            let num = self.selu_der(i);
            data.push(num);
        }
        Matrix::new(data, row, cols)
    }
}

#[cfg(test)]
mod tests {
    use crate::activation::{Function, SELU};
    use crate::linalg::Matrix;
    use crate::{matrix, DataType};

    #[test]
    fn selu() {
        let matrix = matrix![[0.0, 1.0]];
        let a = SELU::new(DataType::f64());
        let matrix = a.call(matrix);
        println!("{}", matrix);
    }

    #[test]
    fn derivative_selu() {
        let matrix = matrix![[0.0, 1.0]];
        let a = SELU::new(DataType::f64());
        let matrix = a.derivative(matrix);
        println!("{}", matrix);
    }
}