jiro_nn 0.8.0

Neural Networks framework with model building & data preprocessing features.
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

use serde::{Deserialize, Serialize};

use crate::{
    learning_rate::{default_learning_rate, LearningRateSchedule},
    linalg::{Matrix, MatrixTrait, Scalar},
};

#[derive(Clone, Debug, Serialize, Deserialize)]
pub struct SGD {
    #[serde(default = "default_learning_rate")]
    learning_rate: LearningRateSchedule,
}

impl SGD {
    pub fn default() -> Self {
        Self {
            learning_rate: default_learning_rate(),
        }
    }

    pub fn with_const_lr(learning_rate: Scalar) -> Self {
        Self {
            learning_rate: LearningRateSchedule::Constant(learning_rate),
        }
    }

    pub fn new(learning_rate: LearningRateSchedule) -> Self {
        Self { learning_rate }
    }

    pub fn update_parameters(
        &mut self,
        epoch: usize,
        parameters: &Matrix,
        parameters_gradient: &Matrix,
    ) -> Matrix {
        let lr = self.learning_rate.get_learning_rate(epoch);
        parameters.component_sub(&parameters_gradient.scalar_mul(lr))
    }
}