Extraction of Features for Time Series Classification Using Noise Injection

Area of Science:

• Data Science
• Machine Learning
• Signal Processing

Background:

• Time series data classification faces challenges due to variability, noise, and imbalance.
• Traditional methods often struggle with generalization performance on complex time series.
• There is a need for advanced techniques to improve data quality and diversity for classification.

Purpose of the Study:

• To introduce a novel feature extraction method for time series classification.
• To enhance data diversity and quality using noise injection and digital signal processing (DSP).
• To improve the generalization performance of time series classification models.

Main Methods:

• Data augmentation via noise injection to increase training data diversity.
• Feature extraction using digital signal processing (DSP) techniques including sampling, quantization, and Fourier transformation.
• Comparison of the proposed method against existing time series classification models.

Main Results:

• The proposed method demonstrates superior performance compared to existing time series classification models.
• Experimental results validate the effectiveness of data augmentation and DSP in time series classification.
• The approach successfully enhances data quality and maximizes model generalization performance.

Conclusions:

• Noise injection and DSP are effective tools for improving time series data classification.
• The developed methodology offers a robust approach for time series data analysis and classification.
• This research has potential applications across various data analysis problems.