neural_network_get_accuracy.cu File Reference

Implementation of the NeuralNetwork::get_accuracy method. More...

#include "neural_network.h"
#include <cuda_runtime.h>
#include <iostream>

Include dependency graph for neural_network_get_accuracy.cu:

Go to the source code of this file.

Functions
__global__ void	calculate_accuracy_kernel (const double *predictions, const double *Y, int size, int *correct_count)

Detailed Description

Implementation of the NeuralNetwork::get_accuracy method.

Definition in file neural_network_get_accuracy.cu.

Function Documentation

calculate_accuracy_kernel()

__global__ void calculate_accuracy_kernel	(	const double *	predictions,
		const double *	Y,
		int	size,
		int *	correct_count
	)

Definition at line 9 of file neural_network_get_accuracy.cu.

                                                                                                                    {
    // Calculate global thread index
    int idx = blockIdx.x * blockDim.x + threadIdx.x;
 
    // Check if thread is within bounds
    if (idx < size) {
        // Increment correct_count if prediction matches true label
        if (predictions[idx] == Y[idx]) {
            atomicAdd(correct_count, 1);
        }
    }
}