Related Experiment Videos
Robust High-dimensional Bioinformatics Data Streams Mining by ODR-ioVFDT.
Dantong Wang1, Simon Fong1, Raymond K Wong2
1Department of Computer and Information Science, Univeristy of Macau, SAR, Macau.
Scientific Reports
|February 24, 2017
Summary
This study introduces a new algorithm, ODR-ioVFDT, for handling outliers in bioinformatics data streams. It improves classification accuracy and efficiency in analyzing biological data for better disease diagnosis.
Area of Science:
- Bioinformatics
- Data Mining
- Machine Learning
Background:
- Outlier detection in streaming data is crucial for accurate bioinformatics analysis.
- Distinguishing noise from exceptions and deciding how to handle them presents a challenge.
Purpose of the Study:
- To propose a novel algorithm, ODR-ioVFDT, for effective outlier management in continuous bioinformatics data learning.
- To integrate outlier detection directly into the data learning process, unlike traditional separate steps.
Main Methods:
- Developed ODR-ioVFDT, an algorithm combining outlier detection with an incrementally Optimized Very Fast Decision Tree.
- Utilized an adaptive interquartile-range based method to set a tolerance threshold for identifying exceptional data points.
- Integrated outlier handling within the training process to adaptively include or exclude data.
Main Results:
- ODR-ioVFDT demonstrated superior performance compared to models without outlier detection.
- Significant improvements were observed in classification accuracy and kappa statistics.
- The algorithm showed reduced time consumption in processing bioinformatics datasets.
Conclusions:
- ODR-ioVFDT offers an effective, integrated approach to outlier detection in bioinformatics streaming data.
- The algorithm enhances the analysis of biological data, potentially leading to more effective disease diagnosis and treatment.