.png){kind=icon}
.png){kind=icon}
Return
TotalPLS: Local Dimension Reduction for Multicategory Microarray Data
.png){kind=icon}
PRE
AI
DOI：10.1109/THMS.2013.2288777.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
Dimension reduction is an important topic in data mining, which is widely used in the areas of genetics, medicine, and bioinformatics. We propose a new local dimension reduction algorithm TotalPLS that operates in a unified partial least squares (PLS) framework and implement an information fusion of PLSbased feature selection and feature extraction. This paper focuses on extracting the potential structure hidden in high-dimensional multicategory microarray data, and interpreting and understanding the results provided by the potential structure information. First, we propose using PLS-based recursive feature elimination (PLSRFE) in multicategory problems. Then, we perform feature importance analysis based on PLSRFE for high-dimensional microarray data to determine the information feature (biomarkers) subset, which relates to the studied tumor subtypes problem. Finally, PLS-based supervised feature extraction is conducted on the selected specific genes subset to extract comprehensive features that best reflect the nature of classification to have a discriminating ability. The proposed algorithm is compared with several state-of-the-art methods using multiple high-dimensional multicategory microarray datasets. Our comparison is performed in terms of recognition accuracy, relevance, and redundancy. Experimental results show that the algorithm proposed by us can improve the recognition rate and computational efficiency. Furthermore, mining potential structure information improves the interpretability and understandability of recognition results. The proposed algorithm can be effectively applied tomicroarray data analysis for the discovery of gene coexpression and coregulation.
Keywords:
Dimension reduction
feature extraction
feature selection
microarray data analysis
partial least squares (PLS)
AI Summary
Key information extracted from the uploaded paper, including a brief overview, abstract, background, key highlights, visual analysis, and future outlook.
Journal
IF:
4.4
Papers:
1.1K
Citations:
3.5K