rmontanana/Platform
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

RocAuc refactor to speed up binary classif. problems

Browse Source
This commit is contained in:
Ricardo Montañana Gómez
2024-07-13 16:54:34 +02:00
parent 028522f180
commit 28f6a0d7a7
1 changed files with 6 additions and 3 deletions
Download Patch File Download Diff File Expand all files Collapse all files

9
src/main/RocAuc.cpp
View File

@@ -1,9 +1,7 @@
#include <sstream> #include <sstream>
#include <vector>
#include <algorithm> #include <algorithm>
#include <numeric> #include <numeric>
#include <utility> #include <utility>
#include "common/Colors.h"
#include "RocAuc.h" #include "RocAuc.h"
namespace platform { namespace platform {
std::vector<int> tensorToVector(const torch::Tensor& tensor) std::vector<int> tensorToVector(const torch::Tensor& tensor)
@@ -30,8 +28,10 @@ namespace platform {
double RocAuc::compute(const torch::Tensor& y_proba, const torch::Tensor& labels) double RocAuc::compute(const torch::Tensor& y_proba, const torch::Tensor& labels)
{ {
size_t nClasses = y_proba.size(1); size_t nClasses = y_proba.size(1);
// In binary classification problem there's no need to calculate the average of the AUCs
if (nClasses == 2)
nClasses = 1;
size_t nSamples = y_proba.size(0); size_t nSamples = y_proba.size(0);
assert(nSamples = y_test.size(0));
y_test = tensorToVector(labels); y_test = tensorToVector(labels);
std::vector<double> aucScores(nClasses, 0.0); std::vector<double> aucScores(nClasses, 0.0);
for (size_t classIdx = 0; classIdx < nClasses; ++classIdx) { for (size_t classIdx = 0; classIdx < nClasses; ++classIdx) {
@@ -47,6 +47,9 @@ namespace platform {
{ {
y_test = labels; y_test = labels;
size_t nClasses = y_proba[0].size(); size_t nClasses = y_proba[0].size();
// In binary classification problem there's no need to calculate the average of the AUCs
if (nClasses == 2)
nClasses = 1;
size_t nSamples = y_proba.size(); size_t nSamples = y_proba.size();
std::vector<double> aucScores(nClasses, 0.0); std::vector<double> aucScores(nClasses, 0.0);
for (size_t classIdx = 0; classIdx < nClasses; ++classIdx) { for (size_t classIdx = 0; classIdx < nClasses; ++classIdx) {