|University / Institute||Universiti Teknikal Malaysia Melaka (UTeM)|
|Research Center/Lab (optional)||Computational and Technologies Research Lab|
|Project Title||Enhancing Adaptability of Continuous Brainwaves Authentication using Fuzzy-Rough Approximated Clustering Model|
|Start Date||Feb 01, 2018|
|End Date||Aug 31, 2019|
|Funding Agency||Ministry of Higher Education (MOHE)|
|Contact Person / Email(s)||Yun-Huoy CHOO / email@example.com|
|Contact Phone(s)||+6012 635 2292|
|Nationality Requirements on Candidate(s)||Applicants must be Malaysian citizens|
|Description||Project Brief Description:
Behavioural biometrics emerge to reduce in-session spoofing by introducing continuous informed monitoring. Despite promising evidence, there exists no strong empirical work on adaptable continuous brainprints modelling in handling the dynamic situation. This is essential because EEG dynamics is easily influenced by emotional states, perception, and reactions. Thus, the gap is on identifying an invariant feature extraction scheme, an adaptive learning model to sanction the client from the imposter. Also, many learning models are striving to include more personal information to increase model performance through personalization, which is against the ethics and human rights. A balance of information masking on brainprint needs to be carefully considered while designing new brainprint model. This research aims to propose a continuous EEG biometrics model by enhancing tolerant features estimator in the Incremental Fuzzy-rough Nearest Neighbour (IncFRNN) classifier from our past research. It is expected to conform adaptively to dynamic condition to manage continuous authentication within a client. This research follows the typical experimental methodology includes phases from designing feature extraction scheme, formulate the continuous authentication model, composing the tolerant estimator into the current IncFRNN model, experimentation, performance analysis, to results validation. The proposed model will be benchmarked with public datasets and collected data from 30 subjects. The primary contribution includes an invariant behavioural feature extraction scheme, a continuous monitoring approach, a continuous brainprints authentication model, a novel fuzzy-rough tolerant estimator, a modified IncFRNN algorithm, and the quantitative analysis of continuous brainprints authentication.
The monthly allowance is RM 1,500 (negotiable up to RM1,800 depending on experience and skills) for the duration of 18 months. All are welcome to apply.