Tensorflow built-in evaluation

Sometimes the error function , It can also be used as an evaluation function
Commonly used built-in evaluation indicators
MeanSquaredError（ Square error , For regression , I could just write it as MSE, The function form is mse）

MeanAbsoluteError ( Absolute error , For regression , I could just write it as MAE, The function form is mae)

MeanAbsolutePercentageError ( Average percentage error , For regression , I could just write it as MAPE, The function form is mape)

RootMeanSquaredError ( Root mean square error , For regression )

Accuracy ( Accuracy rate , Used for classification , You can use strings "Accuracy" Express ,Accuracy=(TP+TN)/(TP+TN+FP+FN), requirement y_true and y_pred All are category serial numbers )

Precision ( Accuracy , Used for dichotomies ,Precision = TP/(TP+FP))

Recall ( Recall rate , Used for dichotomies ,Recall = TP/(TP+FN))

TruePositives ( Real examples , Used for dichotomies )

TrueNegatives ( What a negative example , Used for dichotomies )

FalsePositives ( False positive example , Used for dichotomies )

FalseNegatives ( False negative example , Used for dichotomies )

AUC(ROC curve (TPR vs FPR) The area under , Used for dichotomies , The intuitive explanation is to randomly select a positive sample and a negative sample , The probability that the predicted value of positive samples is greater than that of negative samples )

CategoricalAccuracy（ Classification accuracy , And Accuracy The meaning is the same , requirement y_true(label) by onehot Coding form ）

SparseCategoricalAccuracy ( Sparse classification accuracy , And Accuracy The meaning is the same , requirement y_true(label) It is in the form of serial number )

MeanIoU (Intersection-Over-Union, Commonly used in image segmentation )

TopKCategoricalAccuracy ( Many classification TopK Accuracy rate , requirement y_true(label) by onehot Coding form )

SparseTopKCategoricalAccuracy ( Sparse multi classification TopK Accuracy rate , requirement y_true(label) It is in the form of serial number )

Mean ( Average )

Sum ( Sum up )