Struct SoftMax

pub struct SoftMax;

Softmax function (normalized exponential function).

Converts a vector of K real numbers into a probability distribution of K possible outcomes. The output values are in the range [0, 1] and sum to 1.

Mathematical Definition

For an input vector x, the Softmax function is defined as:

\text{Softmax}(x_i) = \frac{e^{x_i}}{\sum_{j=1}^{K} e^{x_j}}

Examples

use tensorrs::activation::{Function, SoftMax};
use tensorrs::linalg::Matrix;
use tensorrs::matrix;

let softmax = SoftMax::new();
let input = matrix![[1.0, 2.0, 3.0]];
let output = softmax.call(input);
println!("Softmax output: {}", output);
//[{0.09003057 0.24472848 0.66524094},
// {0.090030566 0.24472846 0.66524094}]

See Also

• Wikipedia: Softmax function

Implementations

impl SoftMax

pub fn new() -> Self

Trait Implementations

impl<T: Float> Function<T> for SoftMax

fn derivative(&self, matrix: Matrix<T>) -> Matrix<T>

$Softmax'(x_i)= Softmax(x_i) * (δ_{ij} - Softmax(x_j))$

$δ_{ij}$ - the Kronecker symbol, which is 1 when i = j, and 0 otherwise

WARNING UNTESTED WELL AND DO NOT WORK NORMAL

fn name(&self) -> String

fn call(&self, matrix: Matrix<T>) -> Matrix<T>

Applies the function to the input matrix.

fn is_linear(&self) -> bool

fn get_data(&self) -> Option<Matrix<T>>

fn set_data(&mut self, _data: Matrix<T>)

fn get_weights(&self) -> Option<Matrix<T>>

fn get_bias(&self) -> Option<Matrix<T>>

fn is_bias(&self) -> bool