functional.cpp File Reference

#include <utility>
import cpp_matrix.neural_network:tensor;

Include dependency graph for functional.cpp:

Functions
template<Backend Matrix>
Tensor< Matrix >	cpp_matrix::neural_network::functional::MeanSquaredErrorLoss (const Tensor< Matrix > &input, const Tensor< Matrix > &target)
	Measure the element-wise Mean Squared Error.
auto	cpp_matrix::neural_network::functional::MseLoss (auto &&... args)
	Alias of MeanSquaredErrorLoss().
template<Backend Matrix>
Tensor< Matrix >	cpp_matrix::neural_network::functional::Sigmoid (const Tensor< Matrix > &tensor)
	Applies the element-wise function \( Sigmoid(x) \).

Function Documentation

MeanSquaredErrorLoss()

template<Backend Matrix>

Tensor< Matrix > cpp_matrix::neural_network::functional::MeanSquaredErrorLoss ( const Tensor< Matrix > & input, const Tensor< Matrix > & target )

export

Measure the element-wise Mean Squared Error.

If \( input \) is the matrix of predicted values and \( target \) is the matrix of ground truth values, then:

\[ MSE(input, target) = {1 \over n} \sum_{i=1}^n (input_i - target_i)^2 \]

Parameters

input	Predicted values.
target	Ground truth values.

Returns

Mean Squared Error loss, should be a 1x1 Tensor.

Sigmoid()

template<Backend Matrix>

Tensor< Matrix > cpp_matrix::neural_network::functional::Sigmoid ( const Tensor< Matrix > & tensor )

export

Applies the element-wise function \( Sigmoid(x) \).

\( Sigmoid(x) \) is defined as:

\[ Sigmoid(x) = {1 \over 1 + e^{-x}} \]