Calibration of One-Class SVM for MV set estimation

被引:0
|
作者
Thomas, Albert [1 ,2 ]
Feuillard, Vincent [1 ]
Gramfort, Alexandre [2 ]
机构
[1] Airbus Grp Innovat, 12 Rue Pasteur, F-92150 Suresnes, France
[2] Univ Paris Saclay, Telecom Paris Tech, CNRS, LTCI, F-75013 Paris, France
关键词
D O I
暂无
中图分类号
TP [自动化技术、计算机技术];
学科分类号
0812 ;
摘要
A general approach for anomaly detection or novelty detection consists in estimating high density regions or Minimum Volume (MV) sets. The One-Class Support Vector Machine (OCSVM) is a state-of-the-art algorithm for estimating such regions from high dimensional data. Yet it suffers from practical limitations. When applied to a limited number of samples it can lead to poor performance even when picking the best hyperparameters. Moreover the solution of OCSVM is very sensitive to the selection of hyperparameters which makes it hard to optimize in an unsupervised setting. We present a new approach to estimate MV sets using the OCSVM with a different choice of the parameter controlling the proportion of outliers. The solution function of the OCSVM is learnt on a training set and the desired probability mass is obtained by adjusting the offset on a test set to prevent overfitting. Models learnt on different train/test splits are then aggregated to reduce the variance induced by such random splits. Our approach makes it possible to tune the hyperparameters automatically and obtain nested set estimates. Experimental results show that our approach outperforms the standard OCSVM formulation while suffering less from the curse of dimensionality than kernel density estimates. Results on actual data sets are also presented.
引用
收藏
页码:75 / 83
页数:9
相关论文
共 50 条
  • [41] One-Class SVM with Negative Examples for Fingerprint Liveness Detection
    Jia, Xiaofei
    Zang, Yali
    Zhang, Ning
    Yang, Xin
    Tian, Jie
    BIOMETRIC RECOGNITION (CCBR 2014), 2014, 8833 : 216 - 224
  • [42] An Adaptive Weighted One-Class SVM for Robust Outlier Detection
    Yang, Jinhong
    Deng, Tingquan
    Sui, Ran
    PROCEEDINGS OF THE 2015 CHINESE INTELLIGENT SYSTEMS CONFERENCE, VOL 1, 2016, 359 : 475 - 484
  • [43] Face detection using one-class SVM in color images
    Jin, HL
    Liu, QS
    Lu, HQ
    Tong, XF
    2004 7TH INTERNATIONAL CONFERENCE ON SIGNAL PROCESSING PROCEEDINGS, VOLS 1-3, 2004, : 1431 - 1434
  • [44] One-class SVM with negative examples for fingerprint liveness detection
    Tian, Jie, 1600, Springer Verlag (8833):
  • [45] An automatic road segmentation algorithm using one-class SVM
    Zheng, Sheng
    Liu, Jian
    Shi, Wenzhong
    Zhu, Guangxi
    GEOINFORMATICS 2006: REMOTELY SENSED DATA AND INFORMATION, 2006, 6419
  • [46] One-Class Classification by Combining Density and Class Probability Estimation
    Hempstalk, Kathryn
    Frank, Eibe
    Witten, Ian H.
    MACHINE LEARNING AND KNOWLEDGE DISCOVERY IN DATABASES, PART I, PROCEEDINGS, 2008, 5211 : 505 - 519
  • [47] Learning in presence of class imbalance and class overlapping by using one-class SVM and undersampling technique
    Devi, Debashree
    Biswas, Saroj K.
    Purkayastha, Biswajit
    CONNECTION SCIENCE, 2019, 31 (02) : 105 - 142
  • [48] Swarm intelligent tuning of one-class v-SVM parameters
    Lei Xie
    ROUGH SETS AND KNOWLEDGE TECHNOLOGY, PROCEEDINGS, 2006, 4062 : 552 - 559
  • [49] Relative density degree induced boundary detection for one-class SVM
    Fa Zhu
    Jian Yang
    Sheng Xu
    Cong Gao
    Ning Ye
    Tongming Yin
    Soft Computing, 2016, 20 : 4473 - 4485
  • [50] Spoofing Detection with DNN and One-class SVM for the ASVspoof 2015 Challenge
    Villalba, Jesus
    Miguel, Antonio
    Ortega, Alfonso
    Lleida, Eduardo
    16TH ANNUAL CONFERENCE OF THE INTERNATIONAL SPEECH COMMUNICATION ASSOCIATION (INTERSPEECH 2015), VOLS 1-5, 2015, : 2067 - 2071