As a kind of physical signals that could be easily acquired in daily life, photoplethysmography (PPG)
signal becomes a promising solution to biometric identification for daily access management system (AMS). State-
of-the-art PPG-based identification systems are susceptible to the form of motions and physical conditions of the
subjects. In this work, to exploit the advantage of deep learning, we developed an improved deep convolutional
neural network (CNN) architecture by using the Gram matrix (GM) technique to convert time-serial PPG signals
to two-dimensional images with a temporal dependency to improve accuracy under different forms of motions. To
ensure a fair evaluation, we have adopted cross-validation method and “training and testing” dataset splitting
method on the TROIKA dataset collected in ambulatory conditions. As a result, the proposed GM-CNN method
achieved accuracy improvement from 69.5% to 92.4%, which is the best result in terms of multi-class classification
compared with state-of-the-art models. Based on average five-fold cross-validation, we achieved an accuracy of
99.2%, improved the accuracy by 3.3% compared with the best existing method for the binary-class.
WU Caiyu, (吴彩钰), SABOR Nabil, ZHOU Shihong, (周世鸿), WANG Min, (王 敏), YING Liang (应 亮), WANG Guoxing∗ (王国兴)
. Gram Matrix-Based Convolutional Neural Network for Biometric Identification Using Photoplethysmography Signal[J]. Journal of Shanghai Jiaotong University(Science), 2022
, 27(4)
: 463
-472
.
DOI: 10.1007/s12204-022-2426-5
[1] BONISSI A, LABATI R D, PERICO L, et al. A pre-liminary study on continuous authentication methods for photo plethy smographic biometrics [C]//2013 IEEE Workshop on Biometric Measurements and Systems for Security and Medical Applications. Napoli, Italy:IEEE, 2013: 28-33.
[2] BRUNELLI R, POGGIO T. Face recognition: Features versus templates [J]. IEEE Transactions on Pat-tern Analysis and Machine Intelligence, 1993, 15(10):1042-1052.
[3] WINCHELL S D, RAKVIC R N. Feasibility of iris identification algorithm optimization by fractional template matching [C]//IET Conference on Image Processing. London, UK: IET, 2012: 1-5.
[4] HIGGINS A L, BAHLER L G, PORTER J E. Voice identification using nearest-neighbor distance measure[C]//1993 IEEE International Conference on Acoustics, Speech, and Signal Processing. Minneapolis, MN,USA: IEEE, 1993: 375-378.
[5] YAMAGUCHI O, FUKUI K, MAEDA K I. Face recognition using temporal image sequence [C]//Proceedings Third IEEE International Conference on Automatic Face and Gesture Recognition. Nara, Japan: IEEE,1998: 318-323.
[6] PALANIAPPAN R, KRISHNAN S M. Identifying individuals using ECG beats [C]//2004 Internationa lConference on Signal Processing and Communications.Bangalore, India: IEEE, 2004: 569-572.
[7] POULOS M, RANGOUSSI M, ALEXANDRIS N, etal. Person identification from the EEG using nonlin-ear signal classification [J]. Methods of Information in Medicine, 2002, 41(1): 64-75.
[8] GU Y Y, ZHANG Y T. Photoplethysmographic authentication through fuzzy logic [C]//IEEE EMBSAsian-Pacific Conference on Biomedical Engineering.Kyoto, Japan: IEEE, 2003: 136-137.
[9] NOBUNAGA T, W ATANABE T, TANAKA H. Identification of exercising individuals based on feature sextracted from ECG frequency spectrums [J]. IEICE Transactions on Fundamentals of Electronics, Communications and Computer Sciences, 2018, E101.A(7):1151-1155.
[10] JOSEPH G, JOSEPH A, TITUS G, et al. Photo-plethysmogram (PPG) signal analysis and wavelet de-noising [C]//2014 Annual International Conference on Emerging Research Areas: Magnetics, Machines and Drives (AICERA/iCMMD). Kottayam, India: IEEE,2014: 1-5.
[11] GU Y Y, ZHANG Y, ZHANG Y T. A novel biomet-ric approach in human verification by photoplethys-mographic signals [C]//4th International IEEE EMBSSpecial Topic Conference on Information TechnologyApplications in Biomedicine. Birmingham, UK: IEEE,2003: 13-14.
[12] SPACHOS P, GAO J X, HATZINAKOS D. Feasibilitystudy of photoplethysmographic signals for biometricidentification [C]//2011 17th International Conference o n D i g i t a l S i g n a l P r o c e s s i n g (DSP). Corfu, Greece:IEEE, 2011: 1-5.
[13] RE\U015FIT KA VSAO\U011FLU A, POLAT K, RE-CEP BOZKURT M. A novel feature ranking algorithm for biometric recognition with PPG signals [J]. Computers in Biology and Medicine, 2014, 49: 1-14.
[14] AL SIDANI A, CHERRY A, HAJJ-HASSAN H, et al.Comparison between K-nearest neighbor and support vector machine algorithms for PPG biometric identification [C]//2019 Fifth International Conferenceon Advances in Biomedical Engineering (ICABME).Tripoli, Lebanon: IEEE, 2019: 1-4.
[15] ZHANG X, QIN Z, LYU Y Q. Biometric authentication via finger photoplethysmogram [C]// 2018 2nd International Conference on Computer Science and Artificial Intelligence. Shenzhen, China: ACM, 2018: 263-267.
[16] JINDAL V, BIRJANDTALAB J, POUYAN M B, et al. An adaptive deep learning approach for PPG-based identification [C]//2016 38th Annual International Conference of the IEEE Engineering in Medicine and Biology Society. Orlando, FL, USA: IEEE, 2016: 6401-6404.
[17] BISW AS D, EVERSON L, LIU M Q, et al. CorNET:deep learning framework for PPG-based heart rate estimation and biometric identification in ambulant environment [J]. IEEE Transactions on Biomedical Circuits and Systems, 2019, 13(2): 282-291.
[18] LEE E, HO A, W ANG Y T, et al. Cross-domain adaptation for biometric identification using photoplethys-mogram [C]//2020 IEEE International Conference on Acoustics, Speech and Signal Processing. B a r c e l o n a ,Spain: IEEE, 2020: 1289-1293.
[19] ZHANG Z L, PI Z Y, LIU B Y. TROIKA: A general framework for heart rate monitoring using wrist-type photoplethysmographic signals during intensive physical exercise [J]. IEEE Transactions on Bio-Medical Engineering, 2015, 62(2): 522-531.
[20] GONZALEZ H A, MUZAFF AR S, YOO J, et al.BioCNN: A hardware inference engine for EEG-basedemotion detection [J]. IEEE Access, 2020, 8: 140896-140914.
[21] F AUST O, HAGIW ARA Y, HONG T J, et al. Deep learning for healthcare applications based on physio- logical signals: A review [J]. Computer Methods and Programs in Biomedicine, 2018, 161: 1-13.
[22] LECUN Y, BENGIO Y. Convolutional networks for images, speech, and time-series [M]// The handbook of brain theory and neural networks. Cambridge, USA:MIT Press, 1995: 255-258.
[23] KIDO S, HIRANO Y, HASHIMOTO N. Detection and classification of lung abnormalities by use of convolutional neural network (CNN) and regions with CNN features (R-CNN) [C]//2018 International Workshop on Advanced Image Technology (IWAIT ). Chiang Mai,Thailand: IEEE, 2018: 1-4.
[24] DE ALMEIDA M C, ASADA E N, GARCIA A V. On the use of gram matrix in observability analysis [J].IEEE Transactions on Power Systems, 2008, 23(1):249-251.
[25] L E C U N Y , B O T T O U L , B E N G I O Y , e t a l . G r a d i e n t -based learning applied to document recognition [J].Proceedings of the IEEE, 1998, 86(11): 2278-2324.
[26] RAHADIAN R, SUYANTO S. Deep residual neural network for age classification with face image[C]//2019 International Seminar on Research of Information Technology and Intelligent Systems (ISRITI ).Yogyakarta, Indonesia: IEEE, 2019: 21-24.
[27] HE K, ZHANG X, REN S, et al. Identity mappings indeep residual networks [M]//Computer vision - ECCV2016. Cham: Springer, 2016: 630-645.
[28] NAIR V, HINTON G E. Rectified linear units im-prove restricted Boltzmann machines [C]//27th Inter-national Conference on Machine Learning. H a i f a , I s rael: ACM, 2010: 807-814.
[29] IOFFE S, SZEGEDY C. Batch normalization:Accelerating deep network training by reducing internal covariate shift [EB/OL]. (2015-02-11).https://arxiv.org/abs/1502.03167.
[30] NOWLAN S J, HINTON G E. Simplifying neural net-works by soft weight-sharing [J]. Neural Computation,1992, 4(4): 473-493.
[31] ASGHAR M A, F A W AD, KHAN M J, et al. EEG-based emotion recognition for multi channel fast empirical mode decomposition using VGG-16 [C]//2020International Conference on Engineering and Emerging Technologies (ICEET ). Lahore, Pakistan: IEEE,2020: 1-7.
