ES1D: A Deep Network for EEG-Based Subject Identification

Arnau-González, Pablo; Katsigiannis, Stamos; Ramzan, Naeem; Tolson, Debbie; Arevalillo-Herrez, Miguel

doi:10.1109/bibe.2017.00-74

Cited by 23 publications

(17 citation statements)

References 22 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The proposed CNN architecture achieved 0.19% Equal Error Rate (EER), however, in practice, collecting 12s EEG signals could take too long for authentication purposes. P. Arnau-Gonzalez et al [6] has also proposed a network architecture named EEG-based Subject Identification (ES1D). This network was more of a fine structural modification of the conventional CNN by implementing a series of 1-D CNN layers and an inception layer using Welch's power spectral density estimation of EEG signal collected from a public database DREAMER with 23 individuals' EEG data.…”

Section: B Deep Learning For User Identificationmentioning

confidence: 99%

“…Despite the increasing on EEG-based biometric authentication and identification, the state-of-the-art EEG authentication approaches are still mainly relies on manually designed features and processed using conventional classification techniques such as k-Nearest Neighbors (k-NN) and Extended Nearest Neighbour (ENN). Deep learning approaches used for EEG-based identity classification has been considered only recently [4]- [6].…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

EEG-based user identification system using 1D-convolutional long short-term memory neural networks

Sun

2019

Expert Systems with Applications

148

View full text Add to dashboard Cite

Electroencephalographic signals (EEG) have widely used in medical applications, yet the use of the signals as biometric identifier has only gained interests in the last few years. The advantage of EEG-based biometric recognition lie in its dynamic property and uniqueness among different individuals. However, it is for this reason that manually designed features are not always adapted to the needs. A novel approach based on 1D Convolutional Long Short-term Memory Neural Network (1D-Convolutional LSTM) for EEG-based biometric identification is proposed in this paper. The complementarity of CNNs and LSTMs largely improves the performance of the identification by removing variation of the input, facilitating the temporal modeling and transforming the features into the high dimensional space which makes the classification more effective respectively. We investigate the architecture to determine the appropriate topology to improve the efficacy of the 1D-Convolutional LSTMs for the authentication. The performance of the proposed approach was validated with a public database consists of EEG data of 109 subjects. The experimental results showed that the proposed network has a very high recognition rate of 99.58%, when using only 16 channel EEG signals, which outperforms the state-of-the-art EEG-based biometric identification methods.

show abstract

Section: B Deep Learning For User Identificationmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

EEG-based user identification system using 1D-convolutional long short-term memory neural networks

Sun

2019

Expert Systems with Applications

148

View full text Add to dashboard Cite

show abstract

“…However in recent years and with the use of machine learning technologies, EEG signals have been used for an ever-increasing pool of applications, mainly focused on Brain-Computer Interfaces (BCI). More recently, researchers showed interest in the uniqueness of EEG for each individual and attempted to create biometric systems based on EEG signals [19,28,20]. Available works have typically disregarded the influence of the so called template ageing, an effect that reflects how a given biometric trait changes over a period of time.…”

Section: Eeg In Biometricsmentioning

confidence: 99%

“…Attempts to identify individuals by only recording one session of data are commonly found in the literature [14,19,20,21,22]. These approaches typically have disregarded the issue of the permanence in EEG signals.…”

Section: Introductionmentioning

confidence: 99%

EEG-based biometrics: effects of template ageing

Arnau-González

Katsigiannis

Arevalillo‐Herráez

et al. 2020

AI for Emerging Verticals: Human-Robot Computing, Sensing and Networking

View full text Add to dashboard Cite

This chapter discusses the effects of template ageing in EEG-based biometrics. The chapter also serves as an introduction to general biometrics and its main tasks: Identification and verification. To do so, we investigate different characterisations of EEG signals and examine the difference of performance in subject identification between single session and cross-session identification experiments. In order to do this, EEG signals are characterised with common state-of-the-art features, i.e. Mel Frequency Cepstral Coefficients (MFCC), Autoregression Coefficients, and Power Spectral Density-derived features. The samples were later classified using various classifiers, including Support Vector Machines and k-Nearest Neighbours with different parametrisations. Results show that performance tends to be worse for crosssession identification compared to single session identification. This finding suggests that temporal permanence of EEG signals is limited and thus more sophisticated methods are needed in order to characterise EEG signals for the task of subject identification.

show abstract

“…Regardless the application, it is clear that AI techniques, such as supervised classification, are essential for detecting different emotions from the acquired brain signals [2,4,5], while affect is key to improve the user experience in many different areas. Efficient affect detection from brain signals is currently an open problem with numerous research works being conducted every year.…”

Section: Introductionmentioning

confidence: 99%