EEG-based Personal Identification: from Proof-of-Concept to A Practical System

Su, Fei; Xia, Liwen; Cai, Anni; Wu, Yibing; Ma, Junshui

doi:10.1109/icpr.2010.908

Cited by 42 publications

(36 citation statements)

References 7 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…With the feature set combining auto-regression (A R) model parameter [7] and power spectrum density (PSD) [8,9], the identification system achieved an average accuracy of 97.3% on a dataset of 40 subjects at the recording duration of 3 minutes [4]. This promising performance proved the statistical commonality existed in one's EEG signals.…”

Section: Introductionmentioning

confidence: 90%

“…The subjects were required to stay calm during recording. The data collection scheme is the same as that in [4,5,6]. Each original EEG segment lasts for around 5 minutes.…”

Section: Datasetsmentioning

confidence: 99%

“…Since the PSD below 4Hz is frequently contaminated by ocular artifacts, and the PSD above 33Hz is affected by a system specific notch filter, we use PSD at the particular frequency range of 4Hz to 32Hz with a frequency band being represented by 4 points as in [4]. Furthermore, the mean ener gy and variance of each subband decomposed by 6-level db4 WPT are extracted.…”

Section: Feature Setsmentioning

confidence: 99%

“…Most existing studies [4,5,6] have tried to address this problem by seeking a distinctive and stable feature repre sentation of EEG signals and making nearest neighbor (NN) search for identity classification of EEG signals. With the feature set combining auto-regression (A R) model parameter [7] and power spectrum density (PSD) [8,9], the identification system achieved an average accuracy of 97.3% on a dataset of 40 subjects at the recording duration of 3 minutes [4].…”

Section: Introductionmentioning

confidence: 99%

See 3 more Smart Citations

Local metric learning for EEG-based personal identification

Cai

Liu

2015

2015 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)

Self Cite

View full text Add to dashboard Cite

There has been an increasing attention on Electroencephalo graph (EEG) based personal identification over the last decade. Most existing methods address this problem by Euclidean metric based Nearest Neighbor (NN) search. How ever, under various recording conditions, simple Euclidean distance cannot model the similarity relations between EEG signals precisely. To overcome this drawback, a local met ric learning based on Large Margin Nearest Neighbor (L LMNN) for EEG based personal identification is proposed in this paper. For each EEG sample, a separate local met ric is learned, making the distance between intra-class EEG samples minimized and simultaneously those of inter-class EEG samples maximized. To balance the locality and com putational efficiency, the local metrics are approximated by weighted linear combinations of a small set of anchor sam ples. Experimental results demonstrate that the proposed approach obtains competitive performance compared with state-of-the-art methods. It improves the identification ac curacy overall, especially at shorter EEG durations, which is important for improving the practicability of EEG-based personal identification system.

show abstract

Section: Introductionmentioning

confidence: 90%

“…The subjects were required to stay calm during recording. The data collection scheme is the same as that in [4,5,6]. Each original EEG segment lasts for around 5 minutes.…”

Section: Datasetsmentioning

confidence: 99%

Section: Feature Setsmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Local metric learning for EEG-based personal identification

Cai

Liu

2015

2015 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)

Self Cite

View full text Add to dashboard Cite

show abstract

“…Different feature extraction techniques have been adopted for EEG biometrics in previous works, however, the Auto-Regressive (AR) modeling technique has been shown to achieve higher recognition rates than other techniques (Liu et al (2014); Su et al (2010)). So, the AR modeling is adopted to extract features from EEG frames of relaxation and visual stimulation.…”

Section: Pre-processing and Feature Extraction Of Eeg Signalsmentioning

confidence: 99%