Classification and recognition of encrypted EEG data based on neural network

Liu, Yongshuang; Huang, Haiping; Xiao, Fu; Malekian, Reza; Wang, Wenming

doi:10.1016/j.jisa.2020.102567

Cited by 9 publications

(5 citation statements)

References 17 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Unlike WD, which decomposes signals solely into low-frequency components, WPD is a more generalized form that decomposes high and low frequency modules. This was utilized in the detection of epileptic seizures [29,30]. Wavelet packet tree Si, n is nth (n = 0, 1, 2, 3,4,5,..., 2 j-1) wavelet packet at the I th scale, and the equivalent orthonormal basis is given as 𝑆 𝑖,0 𝑛 (t) ,…”

Section: Wavelet Packet Decomposition (Wpd)mentioning

confidence: 99%

See 1 more Smart Citation

Enhanced Eeg Signal Classification for Motor Imagery Based Brain Computer Interface: A Fusion Approach Using Discrete Wavelet Transform and Quad Binary Pattern

Ananthi,

Subathra,

et al. 2024

Preprint

View full text Add to dashboard Cite

Motor Imagery - Brain-Computer Interface (MI-BCI) is a technique that acts as a non-muscular channel for disabled people to communicate. Because it is a communication link between the wired brain and an external device. In this paper, we propose a comprehensive feature extraction approach for electroencephalogram (EEG) signals to achieve maximum classification accuracy by fusion techniques. The Wavelet Packet Decomposition (WPD), Discrete Wavelet Transform (DWT), and Quad Binary Pattern (QBP) methods are applied to EEG signals. The proposed method consists of five strategies out of which two are fusion techniques based on EEG signal and pattern techniques. One hybrid method involves decomposing EEG signals using DWT, and feeding it into the QBP domain to extract the histogram features. In another hybrid model, the signals are decomposed using WPD and were given to the QBP for feature extraction. Artificial Neural Network (ANN) is used as a classifier in our work. Among the used algorithm DWT based QBP approach attained maximum accuracy. The proposed method is validated using well-known Physionet EEG dataset by considering binary class (imaginary movement of left and right fist). The extracted features are to classify the movement of the right and left fist EEG signals. Using a DWT-based QBP approach, the proposed technique attained a accuracy of 99.50 percent on physionet dataset.

show abstract

Section: Wavelet Packet Decomposition (Wpd)mentioning

confidence: 99%

“…This underscores the rationale behind our selection of a feed-forward neural network. The approach proposed by the author [29],…”

Section: 5dwt Based Qbpmentioning

confidence: 99%

Enhanced Eeg Signal Classification for Motor Imagery Based Brain Computer Interface: A Fusion Approach Using Discrete Wavelet Transform and Quad Binary Pattern

Ananthi,

Subathra,

et al. 2024

Preprint

View full text Add to dashboard Cite

show abstract

“…erefore, it is necessary to encrypt the transmission channel and the storage of EEG signals to prevent attackers from stealing the EEG signals of legitimate users. For the purpose of solving the defects such as low accuracy, high time complexity, or slow processing speed, Liu et al [102] used the Paillier encryption algorithm to encrypt EEG data. e neural network is used for the classification and recognition of encrypted EEG data [102].…”

Section: Security Issuesmentioning

confidence: 99%

“…For the purpose of solving the defects such as low accuracy, high time complexity, or slow processing speed, Liu et al [102] used the Paillier encryption algorithm to encrypt EEG data. e neural network is used for the classification and recognition of encrypted EEG data [102]. Meanwhile, the EEG signals stored locally should be encrypted to prevent attackers from stealing [103].…”

Section: Security Issuesmentioning

confidence: 99%

Review on EEG‐Based Authentication Technology

Zhang

Sun

Mao

et al. 2021

Computational Intelligence and Neuroscience

View full text Add to dashboard Cite

With the rapid development of brain-computer interface technology, as a new biometric feature, EEG signal has been widely concerned in recent years. The safety of brain-computer interface and the long-term insecurity of biometric authentication have a new solution. This review analyzes the biometrics of EEG signals, and the latest research is involved in the authentication process. This review mainly introduced the method of EEG-based authentication and systematically introduced EEG-based biometric cryptosystems for authentication for the first time. In cryptography, the key is the core basis of authentication in the cryptographic system, and cryptographic technology can effectively improve the security of biometric authentication and protect biometrics. The revocability of EEG-based biometric cryptosystems is an advantage that traditional biometric authentication does not have. Finally, the existing problems and future development directions of identity authentication technology based on EEG signals are proposed, providing a reference for the related studies.

show abstract

“…Since the purpose is to secure EEG information while being stored or transmitted over an insecure channel, the encryption method does not allow for arithmetic operations. Liu et al [26] used a feed-forward neural network where the activation function is approximated with a linear function to solve an EEG-based classification problem. The training step was performed using plaintext data, while the encryption was employed only during the inference phase.…”

Section: Introductionmentioning

confidence: 99%

Privacy Preserving Classification of EEG Data Using Machine Learning and Homomorphic Encryption

et al. 2021

View full text Add to dashboard Cite

Data privacy is a major concern when accessing and processing sensitive medical data. A promising approach among privacy-preserving techniques is homomorphic encryption (HE), which allows for computations to be performed on encrypted data. Currently, HE still faces practical limitations related to high computational complexity, noise accumulation, and sole applicability the at bit or small integer values level. We propose herein an encoding method that enables typical HE schemes to operate on real-valued numbers of arbitrary precision and size. The approach is evaluated on two real-world scenarios relying on EEG signals: seizure detection and prediction of predisposition to alcoholism. A supervised machine learning-based approach is formulated, and training is performed using a direct (non-iterative) fitting method that requires a fixed and deterministic number of steps. Experiments on synthetic data of varying size and complexity are performed to determine the impact on runtime and error accumulation. The computational time for training the models increases but remains manageable, while the inference time remains in the order of milliseconds. The prediction performance of the models operating on encoded and encrypted data is comparable to that of standard models operating on plaintext data.

show abstract

Classification and recognition of encrypted EEG data based on neural network

Cited by 9 publications

References 17 publications

Enhanced Eeg Signal Classification for Motor Imagery Based Brain Computer Interface: A Fusion Approach Using Discrete Wavelet Transform and Quad Binary Pattern

Enhanced Eeg Signal Classification for Motor Imagery Based Brain Computer Interface: A Fusion Approach Using Discrete Wavelet Transform and Quad Binary Pattern

Review on EEG‐Based Authentication Technology

Privacy Preserving Classification of EEG Data Using Machine Learning and Homomorphic Encryption

Contact Info

Product

Resources

About