MQ-KPCA: Custom Kernel PCA for Classification of Microscopic Images
By: Suresha, M.
Contributor(s): Raghukumar, D. S.
Publisher: New York Springer 2022Edition: Vol, 103(6), Dec.Description: 2025–2033p.Subject(s): Electrical EngineeringOnline resources: Click here In: Journal of the institution of engineers (India): Series BSummary: Principal component analysis (PCA) is an efficient and linear feature transformation technique. However, it limited to linear components, because it defined on the mean and covariance matrix’s eigenvectors of the data. This paper proposes a new modified method of kernel PCA (KPCA) using a Multiquadric function, called MQ-KPCA. This is more efficient for classification, compared to other KPCA techniques and performs more nonlinear at dimension reduction with keeping variation of data. The proposed dimension transformation technique tested with texture features extracted from defective steel surfaces collected from the Northeastern University dataset and cancer-affected lymph nodes and Chinese hamster ovary cells gathered from IICBU Biological Image Repository, and achieved better results compared to other KPCA methods.Item type | Current location | Call number | Status | Date due | Barcode | Item holds |
---|---|---|---|---|---|---|
Articles Abstract Database | School of Engineering & Technology Archieval Section | Not for loan | 2023-0015 |
Principal component analysis (PCA) is an efficient and linear feature transformation technique. However, it limited to linear components, because it defined on the mean and covariance matrix’s eigenvectors of the data. This paper proposes a new modified method of kernel PCA (KPCA) using a Multiquadric function, called MQ-KPCA. This is more efficient for classification, compared to other KPCA techniques and performs more nonlinear at dimension reduction with keeping variation of data. The proposed dimension transformation technique tested with texture features extracted from defective steel surfaces collected from the Northeastern University dataset and cancer-affected lymph nodes and Chinese hamster ovary cells gathered from IICBU Biological Image Repository, and achieved better results compared to other KPCA methods.
There are no comments for this item.