Distraction descriptor for brainprint authentication modelling using probability-based incremental fuzzy-rough nearest neighbour technique

Liew, Siaw Hong (2021) Distraction descriptor for brainprint authentication modelling using probability-based incremental fuzzy-rough nearest neighbour technique. Doctoral thesis, Universiti Teknikal Malaysia Melaka.

[img] Text (24 Pages)
Distraction descriptor for brainprint authentication modelling using probability-based incremental fuzzy-rough nearest neighbour technique.pdf - Submitted Version

Download (1MB)
[img] Text (Full Text)
Distraction descriptor for brainprint authentication modelling using probability-based incremental fuzzy-rough nearest neighbour technique.pdf - Submitted Version
Restricted to Registered users only

Download (2MB)

Abstract

The characteristics of uniqueness and proof of aliveness have driven the research in Brainprint as a biometric modality. Brainprint measuring by Electroencephalogram (EEG) suffers from low signal-to-noise ratio and are varied across time. Most of the brainprint authentication models were tested in well-controlled environments to minimize the influence of ambient disturbance on the EEG signals. These settings significantly contradict the real- world situations. Thus, making use of the distraction is wiser than eliminating it. This research aims to design a distraction descriptor, elicited through the object variation, to refine the granular knowledge incrementally, using the proposed probability-based update strategy in Incremental Fuzzy-Rough Nearest Neighbor (IncFRNN) technique. The research follows the experimental methodology, starting from data acquisition to data imputation, EEG distraction descriptor, probability-based IncFRNN and model analysis. The EEG of 45 volunteer human subjects were collected using visual stimuli in three levels of auditory ambient distraction, which are in quiet, low, and high distraction conditions. An artefact rejection with amplitude greater than 100 µV was applied for data cleaning. Occasionally, missing values occurred after removing the noisy trials. A similarity matching imputation method is proposed for EEG data imputation. The power spectral density, wavelet phase stability, and coherence were used as feature extraction methods. The probability-based IncFRNN technique was used to construct the learning model. The proposed probability- based incremental update strategy is benchmarked with the ground truth (actual class) incremental update strategy. Besides, the proposed technique is also benchmarked with First- In-First-Out (FIFO) incremental update strategy in K-Nearest Neighbour (KNN). The authentication accuracy, area under receiver operating characteristic curve, recall, precision, and the F-measure were used to evaluate the proposed technique. The experimental results have shown equivalence discriminatory performance in both high distraction and quiet conditions. This has proven that the proposed distraction descriptor is able to utilize the unique EEG response towards ambient distraction to complement person authentication modelling in the uncontrolled environment. However, the authentication results in low distraction condition are significantly worse than both the quiet and high distraction conditions. This might because the distraction is too mild to elicit the cognitive measures representing individual characteristics. The probability-based IncFRNN technique has significantly outperformed the KNN technique for both with and without defining the window size threshold. Nevertheless, its performance is slightly worse than the actual class incremental update strategy since the ground truth represents the gold standard. In overall, this study demonstrated a more practical brainprint authentication model with the proposed distraction descriptor and the probability-based incremental update strategy. However, the data acquisition was carried out in a single session. The EEG distraction descriptor may vary due to intersession variability. Future research should focus on the intersession variability to improve the robustness of the brainprint authentication model.

Item Type: Thesis (Doctoral)
Uncontrolled Keywords: Electroencephalography, Analysis
Subjects: Q Science > Q Science (General)
Q Science > QP Physiology
Divisions: Library > Tesis > FTMK
Depositing User: F Haslinda Harun
Date Deposited: 13 Jan 2023 15:38
Last Modified: 13 Jan 2023 15:38
URI: http://eprints.utem.edu.my/id/eprint/26097
Statistic Details: View Download Statistic

Actions (login required)

View Item View Item