An incremental learning method for spoof fingerprint detection
作者:
Highlights:
• A new spoof fingerprint detection framework to learn new fakes incrementally.
• An expert-construction method is proposed, based on boosting and SVMs.
• A dynamic weight-update rule is proposed to combine multiple experts.
• The proposed method provides high incremental learning ability.
摘要
•A new spoof fingerprint detection framework to learn new fakes incrementally.•An expert-construction method is proposed, based on boosting and SVMs.•A dynamic weight-update rule is proposed to combine multiple experts.•The proposed method provides high incremental learning ability.
论文关键词:Spoof fingerprint detection,Static feature,Incremental learning,Catastrophic forgetting
论文评审过程:Received 24 August 2017, Revised 23 August 2018, Accepted 31 August 2018, Available online 3 September 2018, Version of Record 13 September 2018.
论文官网地址:https://doi.org/10.1016/j.eswa.2018.08.055