Before going into the various methods of the mission of sensing a few shots, we highlight in this section some evaluation measures that are more suitable for the task of learning. Data distribution will show a degree of defect between the training group and the test set for the size of the small sample, unlike the usual tasks of learning, and therefore the appropriate measures that deal with such a defect needs to be invoked. In Table 1, we show different metrics along with a brief overview. The metrics are the confusion matrix, resolution, summons, F1 degree, overall resolution (OA), average accuracy (AA), Kappa κ, and PR curve. (5)-(9) Mathematical describes some standards as shown in the relevant equations.
TP, FP, TN and FN variables in previous equations represent positive, negative, negative and wrong groups, respectively. In the equation (6), NCLASSES indicates the total number of categories taken into account.
Authors:
(1) Gao Yu Lee, College of Electrical and Electronic Engineering, Nanyang University Technology, 50 Nanyang Ave, 639798, Singapore ([email protected]);
(2) Tanmoy Dam, College of Mechanical and Space Engineering, Nanyang University Technology, 65 Nanyang Drive, 637460, Singapore and Computer Science Department, New Orleans University, New Orleans, 2000 Lixur Drive, Los Angeles 70148, USA (USA (USA (USA)[email protected]);
(3) MD Meftahul Ferdaus, College of Electrical and Electronic Engineering, Nanyang University, 50 Nanyang Ave, 639798, Singapore ([email protected]);
(4) Daniel Buyu Buenar, College of Electrical and Electronic Engineering, Nianiang Technological University, 50 Nanish Avi, 639798, Singapore ([email protected]);
(5) Vu N. Duong, College of Mechanical and Space Engineering, Nanyang University, 65 Nanyang Drive, 637460, Singapore ([email protected]).
