Decremental generalized discriminative common vectors applied to images classification
作者：
Highlights：
• Decremental learning as an online feature extraction process for supervised problem.
• DGDCV allows updating an existing model by removing obsolete or misleading data.
• Suited for images classification, able to work w/o the Small Sample Size case.
• Very low computational cost without compromising the accuracy of the model.
• Lower spatial complexity and no requiring access to all original training samples.
摘要
•Decremental learning as an online feature extraction process for supervised problem.•DGDCV allows updating an existing model by removing obsolete or misleading data.•Suited for images classification, able to work w/o the Small Sample Size case.•Very low computational cost without compromising the accuracy of the model.•Lower spatial complexity and no requiring access to all original training samples.
论文关键词：Decremental learning,Generalized Discriminative Common Vectors,Feature extraction,Linear subspace methods,Classification
论文评审过程：Received 27 July 2016, Revised 17 May 2017, Accepted 20 May 2017, Available online 25 May 2017, Version of Record 20 June 2017.
论文官网地址：https://doi.org/10.1016/j.knosys.2017.05.020
