@techreport{oai:ipsj.ixsq.nii.ac.jp:00218829,
 author = {小林, 裕季 and 中島, 宏智 and 中西, 功 and Hiroki, Kobayashi and Hirotomo, Nakashima and Isao, Nakanishi},
 issue = {24},
 month = {Jul},
 note = {本研究では，秘匿性が高く継続的に検出が可能な生体情報として脳波に注目し，知覚できない振動刺激を呈示した際の誘発脳波を用いた個人識別の実現を目指している．先行研究では，振動刺激によって発生させた誘発脳波のデータから，???? 波（4–8 Hz），???? 波（8–13 Hz），???? 波（13–43 Hz）のパワースペクトルの含有率を特徴量とし，SVM（サポートベクターマシン）にて識別性能評価を行った．その結果，EER は 28.5 %  が得られたが十分な識別結果ではなかった．そこで本稿では，識別性能の改善を目的とし，従来の含有率を重み付けしたものを新規の特徴量として識別性能を行った．その結果，EER は 16.9 % となり識別性能は改善した．また，特徴帯域を誘発脳波の成分が多く含まれる帯域に変更し，識別性能評価を行った結果，EER は 16.1 % と識別性能はさらに改善した．, In this study, we focus on the EEG as a biometrics that can be detected continuously with high confidentiality, and aim to realize the person verification using the evoked EEG when presented with an imperceptible vibration stimulus. In previous studies, the content rates of the power spectrum in theta (4–8 Hz), alpha (8–13 Hz), and beta (13–43 Hz) wavebands as individual feature were derived from the evoked EEG data generated by vibration stimulation, and the verification performance was evaluated by SVM. The results showed that the EER was 28.5 %, but this was not a sufficient verification result. In this paper, for the purpose of improving the verification performance, the weighted (normalized) content rates are used as new features and the verification performance is evaluated. As a result, the EER is improved to 16.9 %. The verification performance is futher improved by changing the feature bandwidth to (6–10 Hz), which contain many components of evoked EEG, and the EER is reduced to 16.1 %.},
 title = {知覚できない振動刺激による誘発脳波を用いた個人識別-脳波スペクトル含有率への正規化導入-},
 year = {2022}
}