ENHANCING WOMEN'S SAFETY: UNLEASHING THE POWER OF EEG PRE-PROCESSING TO DECIPHER DEEP LEARNING INSIGHTS INTO EMOTIONS

In the 21st century, women have made significant strides in contributing to society and participating alongside men in various fields. However, gender-based violence continues to negatively impact women's lives and limit their involvement in decision- making processes. Wearable technologies and mobile applications have emerged as potential solutions for enhancing women's safety, but their effectiveness remains limited due to inadequate social action. This study focuses on the classification of women's emotions using EEG signals, employing a four-stage preprocessing approach. The acquisition of accurate EEG signals necessitates a lengthy observation period and multiple activity processes to capture all brain waves. To improve signal quality, Zero-timeframe windowing (ZTFW) is applied for signal segmentation, Discrete Wavelet Transform (DWT) is utilized for filtering to eliminate erroneous brain activity analysis, Empirical Mode Decomposition (EMD) is employed for noise removal, and Fuzzy Elliptical Filter (FEF) is used to reduce power line noise and electrode interference. The proposed FEF filter demonstrates significant improvements in the denoising process, with a remarkably low mean error rate of 0.03 compared to other filters.