The accuracy of the recognition has been evaluated by; the Receiver Operating Characteristic (ROC) curve. The ROC curve is a plot of the False Non Matching Rate (FNMR) against the False Matching Rate (FMR). Also The EER refers to the point where the FNMR and FMR are equal can be used
It depends on the nature of the problem itself ( classification , regression , or clustering ).
To measure the accuracy of a classification problem, the confusion matrix should be calculated first. Then, you will get the main parameters which are ( True Positive (TP), True Negative (TN), False Positive (FP), and False Negative (FN)).
Acc = All the correctly classified samples / the total number of samples
(or)
Acc = (TP + TN) / ( TP + TN + FP + FN)
Note: the number of Ts and Fs depend on the number of classes, if you have two classes (binary classification problem ), then there are two True values, and two False values. Because the confusion matrix is equal to 2 * 2. In case you are working on a multi-class problem, then the confusion matrix is equal to ( No. of Classes * No. of Classes )
For regression problems, you can't measure the accuracy just like the method above due two the nature of the class label ( numeric values ). So, there are several Statistical measurements could be used, such as (Mean Absolute Error "MAE", Mean Square Error "MSE" .. etc). The comparison stage depends on something called "R value" which depends on the above mentioned methods.
The accuracy of the recognition has been evaluated by; the Receiver Operating Characteristic (ROC) curve. The ROC curve is a plot of the False Non Matching Rate (FNMR) against the False Matching Rate (FMR). Also The EER refers to the point where the FNMR and FMR are equal can be used