Soft precision and recall

被引:39
作者
Franti, Pasi [1 ]
Mariescu-Istodor, Radu [2 ]
机构
[1] Univ Eastern Finland, Sch Comp, Joensuu 80101, Finland
[2] Karel Univ Appl Sci, Joensuu 80200, Finland
关键词
Soft measures; Evaluation; Precision; Recall; F-score;
D O I
10.1016/j.patrec.2023.02.005
中图分类号
TP18 [人工智能理论];
学科分类号
081104 ; 0812 ; 0835 ; 1405 ;
摘要
Precision and recall are classical measures used in machine learning. However, they are based on exact matching. This results in binary classification where the predicted item is either a true or false positive despite inexact matching is often preferred in pattern recognition. To address this problem, we introduce soft variants of precision and recall based on application-specific similarity measure. 2022 Elsevier Ltd. All rights reserved.
引用
收藏
页码:115 / 121
页数:7
相关论文
共 16 条
[1]   Framework for syntactic string similarity measures [J].
Gali, Najlah ;
Mariescu-Istodor, Radu ;
Hostettler, Damien ;
Franti, Pasi .
EXPERT SYSTEMS WITH APPLICATIONS, 2019, 129 :169-185
[2]  
Jimenez S., 2010, P INT S STRING PROCE
[3]  
Jimenez S., 2008, THESIS NATL U COLOMB
[4]   Gesture Input for GPS Route Search [J].
Mariescu-Istodor, Radu ;
Franti, Pasi .
STRUCTURAL, SYNTACTIC, AND STATISTICAL PATTERN RECOGNITION, S+SSPR 2016, 2016, 10029 :439-449
[5]   Metrics for Polyphonic Sound Event Detection [J].
Mesaros, Annamaria ;
Heittola, Toni ;
Virtanen, Tuomas .
APPLIED SCIENCES-BASEL, 2016, 6 (06)
[6]  
Mikolov T., 2013, Advances in neural information processing systems, P3111
[7]  
Mikolov T, 2013, Arxiv, DOI [arXiv:1301.3781, 10.48550/ARXIV.1301.3781, 10.48550/arXiv.1301.3781]
[8]   Beyond Precision and Recall: Understanding Uses (and Misuses) of Similarity Hashes in Binary Analysis [J].
Pagani, Fabio ;
Dell'Amico, Matteo ;
Balzarotti, Davide .
PROCEEDINGS OF THE EIGHTH ACM CONFERENCE ON DATA AND APPLICATION SECURITY AND PRIVACY (CODASPY'18), 2018, :354-365
[9]  
Rezaei M., 2015, P FO IEEEWICACM INT
[10]   Set Matching Measures for External Cluster Validity [J].
Rezaei, Mohammad ;
Franti, Pasi .
IEEE TRANSACTIONS ON KNOWLEDGE AND DATA ENGINEERING, 2016, 28 (08) :2173-2186