Conference PaperPDF Available

A Survey of EEG Based User Authentication Schemes

Authors:
  • Independent Researcher

Abstract and Figures

Electroencephalography (EEG) is the recording of electrical activity occurring in the brain, which is recorded from the scalp through placement of voltage sensitive electrodes. It has been repeatedly demonstrated that the brain emits voltage fluctuations on a continuous basis. These fluctuations are a reflection of the on-going brain dynamics, which present as a series of fluctuations that have characteristic waveforms and amplitude patterns, depending on the cognitive state of the subject. A number of published reports have indicated that there is enough depth in the EEG recording, rendering it suitable as a tool for person authentication. This idea has a solid underpinning in that recent evidence suggests much of the on-going EEG recordable activity within brains has a genetic component. This study presents the common steps for developing a human identification systems based on EEG signals. It will also present some of the important techniques used.
Content may be subject to copyright.
The 8th International Conference on INFOrmatics and Systems (INFOS2012) 14-16 May
Bio-inspired Optimization Algorithms and Their Applications Track
Faculty of Computers and Information - Cairo University BIO-55
A Survey of EEG Based User Authentication
Schemes
W. Khalifa , A. Salem, M. Roushdy
Computer Science Department,
Faculty of Computer and Information Sciences,
Ain Shams University,
Cairo, Egypt
abmsalem@yahoo.com
K. Revett
Faculty of Informatics and Computer Science
The British University in Cairo,
Cairo, Egypt
Ken.revett@bue.edu.eg
Abstract Electroencephalography (EEG) is the recording of
electrical activity occurring in the brain, which is recorded from
the scalp through placement of voltage sensitive electrodes. It has
been repeatedly demonstrated that the brain emits voltage
fluctuations on a continuous basis. These fluctuations are a
reflection of the on-going brain dynamics, which present as a
series of fluctuations that have characteristic waveforms and
amplitude patterns, depending on the cognitive state of the
subject. A number of published reports have indicated that there
is enough depth in the EEG recording, rendering it suitable as a
tool for person authentication. This idea has a solid
underpinning in that recent evidence suggests much of the on-
going EEG recordable activity within brains has a genetic
component. This study presents the common steps for developing
a human identification systems based on EEG signals. It will also
present some of the important techniques used.
Keywords-component; artificial intelligence; behavioral
biometrics; cognitive biometrics; EEG; user identification;signal
processing
I. INTRODUCTION
Biometrics is the process of uniquely identifying
individuals based on one or more physical or behavioral
characteristics. Physiological biometrics is related to the shape
of the body such as finder print, face, and DNA; while
behavioral biometrics is related to the person’s behavior such
as typing rhythm, gait and signature. There are several
techniques of recording brain activities such as
magnetoencephalography (MEG) and functional magnetic
resonance imaging (fMRI), electroencephalography and
Electroencephalography (EEG). EEG signals are brain
activities recorded from electrodes mounted on the scalp. EEG
is the most practical capturing method that can be used in
biometrics due to the advances in its hardware devices; there
are some a EEG signal capturing device that are equal in size to
a mobile phone or computer headset. EEG is one of the
physiological unique characteristics of an individual [1].
There are several requirements that need to be covered by
any biometric system: [2]
a) Changeability: If the user’s authentication information is
compromised, we must be able to replace this information (and
revoke any old password or access credential).
b) Shoulder-surfing resistanc: The scheme must not be
vulnerable to shoulder-surfing, particularly in the presence of
ubiquitous visual recording devices.
c) Theft protection: This includes physical theft and the
computational infeasibility of guessing attacks. If we must rely
on the entropy of an authentication scheme for protection
against off-line dictionary attack, we require an authentication
method whose entropy can scale with processor speeds
d) Protection from user non-compliance: To discourage
unintended transfer to other parties, the user should not be able
to write down (in a manner useful to an attacker) or share their
authentication information “too easily”.
The discovery of electrical currents in the brain was
discovered in the 19th century, but understanding the meaning
of such currents advanced in the past years more rapidly as the
technology improved allowing researchers to capture more data
accurately. Moreover, the advancement in the signal processing
and data classification techniques helped researches to use the
data captured in disease diagnosis, brain computer interface
and finally user identification. Several classifiers were
experimented for the use of EEG signals in user identification
such as neural networks, fisher’s classifier and linear classifier.
In Each experiment the subjects were asked to do one or more
mental tasks such as solving a mathematical problem, mental
visual counting, composing a letter or even just resting.
Depending on the tasks and the classifiers used several
experiments were conducted that showed very promising
results in the use of this characteristic as a behavioral biometric
feature.
The need of a new behavioral biometric is derived from the
need of securing important facilities and important information.
Most of the market available secure systems can be penetrated
by hacking or by a mistake of one the authorized personnel.
The good thing about using EEG is that it covers all the above
The 8th International Conference on INFOrmatics and Systems (INFOS2012) 14-16 May
Bio-inspired Optimization Algorithms and Their Applications Track
Faculty of Computers and Information - Cairo University BIO-56
mentioned requirements unlike other techniques. Users can
change their password by selecting a different mental tasks, its
prone to shoulder surfing no one can view your thoughts, Users
have to do the authentication their selves, they can’t give a
copy of the password and they have to be alive. One of the
most common used authentication techniques is finger print
recognition; if the user fingerprint is captured by an intruder,
when the system administrators discover this breach the user
and the intruder will be prevented from entering the system,
The benefits of EEG based authentication system is enormous,
it have all the benefits of different authentication methods.
In this paper we investigate the various techniques and
experiments that were developed for using EEG signals as a
user identification characteristic. In the next section we present
a brief summary on the medical aspects of the EEG signals In
section three we describe the methods of capturing EEG signals
and user identification methodology. In section four we discuss
the techniques used to identify the users. Finally, we present a
brief discussion.
II. MEDICAL ASPECTS OF EEG
The electrical currents in the brain was discovered in 1875
by an English physician Richard Caton. He observed the EEG
from the exposed brains of rabbits and monkeys. In 1924 Hans
Berger, a German neurologist, used his ordinary radio
equipment to amplify the brain's electrical activity measured on
the human scalp. He announced that weak electric currents
generated in the brain can be recorded without opening the
skull, and depicted graphically on a strip of paper. The activity
that he observed changed according to the functional status of
the brain, such as in sleep, anesthesia, lack of oxygen and in
certain neural diseases, such as in epilepsy. [3]
EEG signals are generated from activities in the neurons.
When the neurons are activated, local current flows are
produced. EEG measures mostly the currents that flow during
synaptic excitations of the dendrites of many pyramidal
neurons in the cerebral cortex. Differences of electrical
potentials are caused by summed postsynaptic graded
potentials from pyramidal cells that create electrical dipoles
between soma (body of neuron) and apical dendrites (neural
branches). See figure 1.
Brain electrical current consists mostly of Na+, K+, Ca++,
and Cl- ions that are pumped through channels in neuron
membranes in the direction governed by membrane potential
[4]. The detailed microscopic picture is more sophisticated,
including different types of synapses involving variety of
neurotransmitters. Only large populations of active neurons can
generate electrical activity recordable on the head surface.
Between electrode and neuronal layers current penetrates
through skin, skull and several other layers. Weak electrical
signals detected by the scalp electrodes are massively
amplified, and then displayed on paper or stored to computer
memory [5]. The human brain electric activity starts around the
17-23 week of prenatal development. It is assumed that at birth
the full number of neural cells is already developed, roughly
1011 neurons [6]. This makes an average density of 104
neurons per cubic mm. Neurons are mutually connected into
neural nets through synapses. Adults have about 500 trillion
(5.1014) synapses. The number of synapses per one neuron
with age increases, however the number of neurons with age
decreases, thus the total number of synapses decreases with age
too. From the anatomical point of view, the brain can be
divided into three sections: cerebrum, cerebellum, and brain
stem (see figure 2). The cerebrum consists of left and right
hemisphere with highly convoluted surface layer called
cerebral cortex. The cortex is a dominant part of the central
nervous system. The cerebrum obtains centres for movement
initiation, conscious awareness of sensation, complex analysis,
and expression of emotions and behaviour. The cerebellum
coordinates voluntary movements of muscles and balance
maintaining. The brain stem controls respiration, heart
regulation, biorythms, neurohormone and hormone secretion,
etc.[5]. The highest influence to EEG comes from electric
activity of cerebral cortex due to its surface position.
Figure 1. Basic structural features of a neuron, highlighting the three principal
functions elements.
Figure 2. Regions of Human Brain, highlighting the major anatomical
divisions (the Lobes).
(http://msnowe.files.wordpress.com/2009/06/brain_witelson1.jpg).
The 8th International Conference on INFOrmatics and Systems (INFOS2012) 14-16 May
Bio-inspired Optimization Algorithms and Their Applications Track
Faculty of Computers and Information - Cairo University BIO-57
. III. EEG signals are sinusoidal waves, their amplitude is
normally between 0.5 and 100 μV. After applying a Fourier
transform to the row signals and the power spectrum is
generated, we have four groups of waves [7] see figure 3:
Alpha 8-13 Hz, appears during relaxation without
attention and concentration.
Theta 4-7.5 Hz, theta waves in adults while awake is
abnormal, they are generated by access to unconscious
material, deep inspiration and meditation
Beta 14-26 Hz Usual working rhythm
Delta 0.5-4, usually happens during deep sleep
Figure 3.Samples of EEG demonstrating the major frequency bands that
are presented at various stages of cognitive activity.
III. EXPERIMENT ASPECTS
A. Task Definition
Each part of the brain is responsible for a certain mental or
physical activity. The identification technique will require that
the user perform a certain mental task. Accordingly this task
will trigger neurons in a certain parts of the brain that is
responsible for handling such task. Different tasks have been
used; the most used task was just letting the user stay still in a
quite environment and relax; then capture the signal for a
period of time to identify the user. Several tasks were later
introduced; R. Palaniappan used the following five different
tasks in his experiment [8]:
Baseline task. The subjects were asked to relax and
think of nothing in particular. This task was used as a
control and as a baseline measure of the EEG signals.
Geometric figure rotation task. The subjects were
given 30s to study a particular three-dimensional
block object, after which the drawing was removed
and the subjects were asked to visualize the object
being rotated about an axis. The EEG signals were
recorded during the mental rotation period.
Math task. The subjects were given nontrivial
multiplication problems, such as 79 times 56 and
were asked to solve them without vocalizing or
making any other physical movements. The tasks
were non-repeating and designed so that an
immediate answer was not apparent. The subjects
verified at the end of the task whether or not he/she
arrived at the solution and no subject completed the
task before the end of the 10s recording session.
Mental letter composing task. The subjects were
asked to mentally compose a letter to a friend or a
relative without vocalizing. Since the task was
repeated for several times the subjects were told to
continue with the letter from where they left off.
Visual counting task. The subjects were asked to
imagine a blackboard and to visualize numbers being
written on the board sequentially, with the previous
number being erased before the next number was
written. The subjects were instructed not to verbalize
the numbers but to visualize them. They were also
told to resume counting from the previous task rather
than starting over each time.
B. Signal Capturing
A typical EEG Signal capturing device consists of
electrodes with conductive media, filters and amplifiers and
analogue/digital converters. Devices have up to 256
electrodes; nowadays exists commercial devices with much
less electrodes, devices area available in with 4 or even 2
electrodes. Of course they capture much less information but
data analyzing techniques are enhanced to be able to cope with
such devices.
Electrodes are usually placed on the scalp using the 10-20
standards (see Figure 4). This method was developed to ensure
standardized reproducibility so that a subject's studies could be
compared over time and subjects could be compared to each
other. This system is based on the relationship between the
location of an electrode and the underlying area of cerebral
cortex. The "10" and "20" refer to the fact that the actual
distances between adjacent electrodes are either 10% or 20%
of the total front-back or right-left distance of the skull.
Each site has a letter to identify the lobe and a number to
identify the hemisphere location. The letters F, T, C, P and O
stand for Frontal, Temporal, Central, Parietal and Occipital
respectively. Note that there exists no central lobe, the "C"
letter is only used for identification purposes only. A "z"
(zero) refers to an electrode placed on the midline. Even
numbers (2,4,6,8) refer to electrode positions on the right
hemisphere, whereas odd numbers (1,3,5,7) refer to those on
the left hemisphere. [3]
The 8th International Conference on INFOrmatics and Systems (INFOS2012) 14-16 May
Bio-inspired Optimization Algorithms and Their Applications Track
Faculty of Computers and Information - Cairo University BIO-58
Figure 4. Standard electrode map, illustrating the commonly deployed 10-20
System (http://www.immrama.org/eeg/electrode.html)
Values at each channel is calculated by finding the
difference in the reading of two electrodes based on one of the
below methods
Bipolar: calculating the difference related to a nearest
electrode.
Referential: calculating the difference with reference
to the ear lobe so nodes at the right side of the head
use reference of the right ear, and the left with the left
ear.
Common reference, calculating the difference to a
single electrode reference for all electrodes.
Based on the mental tasks that the user will perform,
activities should be expected at certain channels which reflect
the part of the brain responsible for this activity.
IV. PROCESSING TECHNIQUES
In this section we will discuss various techniques used in
human identification based on EEG Signals
A. Fisher’s Discriminant Analysis
A. Riera et al. [9] have developed a multimodal
authentication algorithm based on EEG and ECG signals.
They conducted the test on 40 healthy subjects. Each subject
was required to sit in a comfortable armchair, to relax, be quiet
and close their eyes. Then three 3 minute takes are recorded to
32 subjects and four 3minutes takes are recorded to the 8
subjects. The 32 subject set are used as reference subject in the
classification stage and the 8 subjects are the ones that are
enrolled into the systems. Then several 1 minute takes are
recorded afterwards to these enrolled subjects, in order to use
them as authentication tests. Two electrodes were used to
capture the EEG signals and 2 for the ECG. The data was
divided to four seconds epochs. Two types of Features were
extracted from the four seconds epochs, one channel features
(Auto regression, Fourier Transform) and Synchronicity
features. Three features were selected from the Synchronicity
features namely; Mutual information (measures the
dependency degree between two random variables given in
bits, when logarithms of base 2 are used in its computation) ,
Coherence (quantizes the correlation between two time series
at different frequencies ), Correlation measures (measure of
the similarity of two signals,). The classifier used in the
authentication process is the classical Fisher’s Discriminant
Analysis, Four different discriminant functions were used
(Linear, Diagonal Linear, quadratic, diagonal quadratic). The
five best classifiers from the original 28 classifiers generated
for each subject are selected during the enrollment and
authentication of each subject.
The False Acceptance Rate (FAR) is computed taking into
account both the intruder and the impostor cases (21.8%). The
True Acceptance Rate (TAR) only takes into account the legal
cases.(71.9%)
After combining the 2 signals (EEG and ECG) the TAR is
97.9% and the FAR is 0.82%.
B. Linear Discriminant Classifier
R Palaniappan[8] proposed a multiple mental thought
identification modal. The experiment was conducted on four
subjects The subjects were seated in an Industrial Acoustics
Company sound controlled booth with dim lighting and noise-
less fan (for ventilation). An Electro-Cap elastic electrode cap
was used to record EEG signals from positions C3, C4, P3, P4,
O1 and O2 defined by the 10-20 system of electrode
placement. Each subject was requested to do up to five mental
tasks. Signals were recorded for 10 seconds during each task
and each task was repeated 10 times. Each recording was
segmented into 20 segments, each 0.5s length. The five
mental tasks performed by the subjects are baseline task
(relaxed state), geometric figure rotation, math task (2 digit
multiplication), mental letter composing task, and visual
mental counting. The captured data features were extracted
using AR modeling. Six AR coefficients were obtained for
each channel, giving a total of 36 feature vector for each EEG
segment for a mental thought. When two mental thoughts were
used, the size of the feature vector was 72 and so forth when
more mental thoughts were used.
Linear Discriminant Classifier was used to classify the
EEG feature vectors, LDC is a linear classification method
that is computationally attractive as compared to other
classifiers like artificial neural network. Various results were
presented showing the error rate using 1,2…5 five
combination of the mental tasks. Using 1 task an average of
error rate is 2.6%, while using the 5 mental tasks, the error rate
was 0.1%.
The 8th International Conference on INFOrmatics and Systems (INFOS2012) 14-16 May
Bio-inspired Optimization Algorithms and Their Applications Track
Faculty of Computers and Information - Cairo University BIO-59
C. LVQ Nerual Net
Cempírek et al. [10], proposed neural network
classification technique for user identification. The algorithm
was conducted on a datasets of 8 subjects. The subject was sat
is a dim and silent room, eyes kept closed. Then the EEG
recordings were segmented (segment length 180 sec, step 22.5
sec); the single segments were centered. Linear magnitude
spectra of the single segments were computed by Fast Fourier
transform (Hamming window was used).
The LVQ neural network is a selforganizing neural
network, with added second layer for vectors classification
intended to be used with unlabelled training data. The first
network layer detects subclasses. The second layer combines
these subclasses into one single class. Actually, the first layer
computes distance between input and stored patterns; the
winning neuron is the one with minimum distance. Hence
LVQ network is a kind of nearest-neighbour classifier; it does
not make clusters, but the algorithm search through the
weights of connections between input layer neurons and
output map neurons. These represent classes. The best
classification rate was around 80% .
D. Neural Network
Sun [11] has developed a user identification system based
on Neural Networks. The system was tested on 9 subjects. The
task was to imagine moving his or her left or right index finger
in response to a highly predictable visual cue. EEG signals
were recorded with 59 electrodes mounted according to the
international 10-10 system. Only Signals from 15 electrodes
were used in the system. Totally 180 trials were recorded for
each subject. Ninety trials with half labeled left and the other
half right were used for training, and the other 90 trials were
for testing. Each trial lasted six seconds with two important
cues. The preparation cue appeared at 3.75 s indicating which
hand movement should be imagined, and the execution cue
appeared at 5.0 s indicating it was time to carry out the
assigned response. The common spatial patterns (CSP) is
employed to carry out energy feature extraction. As a result,
each trial is modeled by an 8-dimensional vector (4 sources
from each kind of mental task is assumed in this paper). Based
on these features, neural network classifiers can be learned.
Neural networks of one hidden layer and one output layer for
experiments. The results showed that imagining left index
finger movements is more appropriate for personal
identification. Left index movement gave a classification
accuracy of 95.6% and right index accuracy gave 94.81%. To
summarize the above mentioned techniques’, Table I presents
a summary of these techniques.
V. CONCLUSION
This paper has summarised several techniques and
implementations that have been published with respect to the
deployment of EEG for human authentication and/or
identification. The results are quite promising with upwards
of 95% accuracy. One of the major obstacles for deploying
the EEG as a biometric is the signal acquisition process. In
the clinical arena, large electrode caps with 128 or 256
electrodes are deployed. These caps require a significant
effort to put on and setup. The impedance must be at a
certain level, the skin requires preparation, conductive gels are
often deployed to enhance the SNR. These steps preclude the
deployment of EEG based approaches to biometrics for
obvious reasons. Furthermore, the electrode wires are
attached to cumbersome ADC boards, attached to a dedicated
computer system. What is required is a small footprint,
portable device and these are being developed at a very rapid
pace. As indicated in table I, authentication can be quite
successful with a small number of electrodes( in fact 2 have
provided significant success in several reported cases).
Furthermore, dry electrodes are now becoming common place
these do not require significant skin preparation and obviate
the need for messy conductive gels. Furthermore, an electrode
cap is not required the electrodes can be placed in a
headband or a baseball cap. In addition, these devices
transmit the data wirelessly (BlueTooth of WiFi) making
data collection a common place task. So the technology is
developing and will continue to do so, provided there is a
market.
The studies selected for presentation in this survey are
typical and one notices that most studies deploy a small
cohort of subjects. This is a significant design flaw though
it is not specific to biometrics. These studies should be
reproduced on much larger scales, utilizing 100’s of subjects
within a study, in order to investigate the scalability of the
classification ability of the EEG. If these studies provide
results consistent with small scale studies, then scalability will
not be a significant factor in the deployment of EEG based
biometrics (this is a major assumption in Cognitive biometrics
[12]). Other salient issues are the temporal stability of the
EEG, and the effect of variations in mental state such as
stress and general arousal levels. These are the principal
topics of the cognitive biometrics domain which serves to
investigate how cognitive and emotional states, in conjunction
with basic genetic variability can provide unique signatures
that will serve as an authentication vehicle. It would be
interesting to simply put on a cap, and begin interacting with
the device, without the need to remember which login ID and
password is required for the current system. The system
would present challenges to the user, to which they would
respond naturally to and based on the stimulus-response
protocol, the user would be authenticated. Lastly, this
approach provides a suitable mechanism for static and
continuous authentication.
The 8th International Conference on INFOrmatics and Systems (INFOS2012) 14-16 May
Bio-inspired Optimization Algorithms and Their Applications Track
Faculty of Computers and Information - Cairo University BIO-60
TABLE I. A SUMMARY OF SELECTED STUDIES DEPLOYING EEG FOR PERSON AUTHENTICATION, INCLUDING TASK MEASURE AND THE RESULTING CLASSIFICATION
ACCURACY (TAR = TRUE ACCEPTANCE RATE AND FAR = FALSE ACCEPTANCE RATE). THE REFERENCES ARE LISTED IN THE RIGHT-MOST COLUMN.
Technique
Channels
Subjects
Task
TAR
A
2
40
Rest
79.2%
[9]
B
6
4
Rest, Math,
Letter, Count,
Rotation
-
[8]
C
-
8
Rest
80%
[10]
D
15
9
Left/Right Hand
Movement
95.6% (left)
94.81 (Right)
[11]
REFERENCES
[1] D. J. A. Smit and D. Posthuma and D. I. Boomsma and E. J. C. De Geus,
“Heritability of background EEG across the power spectrum”
Psychophysiology Journal, vol. 42, pp. 691-697, 2005.
[2] Julie Thorpe and P.C. van Oorschot and Anil Somayaji,” Pass-thoughts:
Authenticating with Our Minds”, In Proceedings of New Security
Paradigns Workshop. Lake Arrowhead, pp.4556, 2005.
[3] Teplan, M, “Fundamentals of EEG Measurement,” in Measurement
Science Review, vol. 2, pp. 111, 2002.
[4] H. L. Atwood, W. A. MacKay., “Essentials of neurophysiology,” B.C.
Decker, Hamilton, Canada, 1989.
[5] F. S. Tyner, J. R.Knott., “Fundamentals of EEG technology Volume 1:
Basic concepts andmethods,”, Raven press, New York., 1989.
[6] Y P. L. Nunez., “Neocortical Dynamics and Human EEG Rhythms,”
Oxford University Press, New York,1995.
[7] http://emedicine.medscape.com/article/1139332-overview [20100712].
[8] Ramaswamy Palaniappan,“Multiple Mental Thought Parametric
Classification: A New Approach for Individual Identification”,
Proceeding of International Journal of Signal Processing, Volume 2, PP
222-226, 2006.
[9] Alejandro Riera, Aureli Soria-Frisch, Marco Caparrini, Ivan Cester, and
Giulio Ruffini, “Multimodal Physiological Biometrics Authentication”,
Biometrics: Theory, Methods, and Applications, Wiley Press, 2010.
[10] Cempírek, M. - Šťastný, J. “The optimization of the EEG-based biometric
classification.” Applied Electronics., pp. 25-28. 2007.
[11] Shiliang Sun. “Multitask Learning for EEG-Based Biometrics”,
Proceeding of International Conference on Patter Recognition, , PP 51-55
,2008
[12] Revett, K., Cognitive Biometrics: A Novel Approach to Person
Authentication, ch 7, in Continuous Authentication using Biometrics:
Data, Methods, and Models, (eds) I. Traore and A. Ahmed, Information
Science Reference, ISBN: 1613501293, publication date: 30 September,
2011.
... Another strategy is to link a standard consumer BCI device to the IOT [17] or to enable a person with severe disabilities to remotely operate the household appliances around them [18]. BCIs can also be used to identify users in systems through bio-signals contributing to the security of the systems as these signals are difficult to synthesize, avoiding attacks and unauthorized accesses [19,20]. The fields of neuromarketing and advertisement are other areas benefiting from the connection to BCI. ...
... The EEG signal's major frequency range is [0. 5 30] Hz, which may be separated into five major bands: Delta, Theta, Alpha, Beta and Gamma waves each with frequencies in the range of [0.5-4] Hz, [4][5][6][7][8] Hz, [8][9][10][11][12][13] Hz, [13][14][15][16][17][18][19][20][21][22][23][24][25][26][27][28][29][30] Hz and [30][31][32][33][34][35][36][37][38][39][40] Hz, respectively. This range carries information about mental states. ...
... Values eSense Description 0 unable to calculate excessive noise [1][2][3][4][5][6][7][8][9][10][11][12][13][14][15][16][17][18][19][20] "strongly lowered" levels distractions, agitation "reduced" wandering thoughts, lack of focus "neutral"/"baseline" levels normal "slightly elevated"/higher than normal levels focus, concentration "elevated"/"heightened" levels elevated attention Table 1. ...
Chapter
Full-text available
Brain-computer interfaces (BCIs) have diverse applications across various research domains. In healthcare, individuals with disabilities in communication and controlling prosthetic devices are aided. Beyond healthcare, BCIs integrate seamlessly into Internet of Things (IoT) and smart environments, enabling intuitive device control and interaction, enhancing user experiences. In neuromarketing and advertising, BCIs help decipher consumers’ preferences and emotional responses to products and services, providing businesses with profound insights into consumer behavior. In education and self-regulation, BCIs monitor and regulate students’ cognitive states. BCIs use sensors and hardware to capture brain signals, with non-invasive electroencephalography (EEG) technology being a pivotal component. Preliminary studies analyzing cognitive load using EEG signals and the Mindwave device pave the way for measuring student learning outcomes, shedding light on cognitive and neurological learning processes. Our research explores these parameters, particularly the Mindwave system, aiming to understand brain function across domains. To this end, we conduct a range of diversified studies, trying to better grasp parameters such as attention, concentration, stress, immersion, and fatigue during various tasks. Ultimately, our work seeks to harness BCIs’ potential to improve our understanding of brain function and enhance various areas of knowledge.
... One main challenge is designing a procedure that effectively identifies and authenticates users through their EEG patterns [14]. Current systems employ a variety of techniques, as explored in [15], but they often face issues like overfitting due to the high number of learnable parameters relative to the available training data [16,17]. Although the effectiveness of these models has increased, overfitting remains a frequent problem since there are many more learnable parameters than there are available training data. ...
... where PSD( f ) is the PSD at frequency f , x windowed is the windowed version of the signal x, and N x is the number of EEG data samples. After calculating the PSD, the power values are summed within the predefined frequency bands: delta (1-4 Hz), theta (4-8 Hz), alpha (8-12 Hz), beta (12)(13)(14)(15)(16)(17)(18)(19)(20)(21)(22)(23)(24)(25)(26)(27)(28)(29)(30), and gamma . In total, five power spectral values are computed for each EEG channel. ...
Article
Full-text available
With recent significant advancements in artificial intelligence, the necessity for more reliable recognition systems has rapidly increased to safeguard individual assets. The use of brain signals for authentication has gained substantial interest within the scientific community over the past decade. Most previous efforts have focused on identifying distinctive information within electroencephalogram (EEG) recordings. In this study, an EEG-based user authentication scheme is presented, employing a multi-layer perceptron feedforward neural network (MLP FFNN). The scheme utilizes P300 potentials derived from EEG signals, focusing on the user’s intent to select specific characters. This approach involves two phases: user identification and user authentication. Both phases utilize EEG recordings of brain signals, data preprocessing, a database to store and manage these recordings for efficient retrieval and organization, and feature extraction using mutual information (MI) from selected EEG data segments, specifically targeting power spectral density (PSD) across five frequency bands. The user identification phase employs multi-class classifiers to predict the identity of a user from a set of enrolled users. The user authentication phase associates the predicted user identities with user labels using probability assessments, verifying the claimed identity as either genuine or an impostor. This scheme combines EEG data segments with user mapping, confidence calculations, and claimed user verification for robust authentication. It also accommodates new users by transforming EEG data into feature vectors without the need for retraining. The model extracts selected features to identify users and to classify the input based on these features to authenticate the user. The experiments show that the proposed scheme can achieve 97% accuracy in EEG-based user identification and authentication.
... • Advancing Security Protocols with Biometric Verification: The distinctiveness of eye blink patterns can be harnessed as a non-intrusive biometric identifier, bolstering the security of systems that require user verification. This use case is particularly pertinent for preventing spoofing in security frameworks [49]. ...
Preprint
Full-text available
This research endeavour introduces a state-of-the-art, assimilated approach for eye blink detection from Electroencephalography (EEG) signals. It combines the prominent strategies of wavelet analysis, autoencoding, and a Crow-Search-optimized k-Nearest Neighbors (k-NN) to enhance the performance of eye blink detection from EEG signals. This procedure is initiated by escalating the robustness of EEG data through jittering, which integrates noise into the dataset. Consequently, the wavelet transform is highly demanded during feature extraction in identifying the essential time-frequency components of the signals. These features are further distilled using an autoencoder to provide a dense, yet informative representation. Prior to introducing these features into the machine learning system, they were adjusted. Evidently, the hyperparameters of the k-Nearest Neighbors model have been fine-tuned using Crow Search Algorithm, inspired by the hunting characteristics of crows. This optimization method actively samples the search space to balance exploration and exploitation to identify the optimal configuration for the model. The k-NN model that has been optimized using the proposed method demonstrates significantly higher performance in the eye blink detection process in comparison to the deep learning models when equipped with decorous feature extraction and fine tuning. The effectiveness of the developed system has been ascertained according to the assessment indices such as accuracy, classification reports, and confusion matrix. Thus, the present work offers an optimal method of detecting the eye blink from the EEG signals assisting in the development of further EEG applications including user interfaces, fatigue level identification, and neurological disorders analysis through the enhancement of signal processing and optimization methods. It becomes evident after a detailed evaluation that conventional machine learning algorithms if implemented with correct feature extraction and fine-tuning surpass the deep learning approaches including the frameworks composed of Convolutional Neural Network (CNN) and Principal Component Analysis (PCA) and empirical mode decomposition by approximately 96% across all datasets. This proves the advantage of optimized traditional Machine Learning models over the Deep Learning models in realistic EEG-based eye blink detection.
... EEG serves as a non-invasive technique for measuring the electrical activity within the brain (Binnie & Prior, 1994). This methodology involves the placement of electrodes onto the scalp to capture the electrical signals emitted by neurons within the brain (Khalifa et al., 2012). With its notable temporal resolution, EEG has the capability to promptly identify fluctuations in brain activity, thereby facilitating real-time monitoring. ...
Article
Full-text available
The exploration of emotional responses to spatial design within the context of urban regeneration is a central concern for both researchers and designers. Notably, there exists a current gap in considering emotions in urban regeneration projects. Moreover, existing methods for identifying people's emotional reactions within spatial environments primarily rely on self-reporting, which has significant limitations and heavily depends on participants' subjective expressions. As a result, it becomes essential to collect and analyze participants' physiological data using objective measures. This study introduces a systematic approach to acquiring unbiased data concerning people's emotional responses to spatial design. By employing Virtual Reality and Galvanic Skin Response techniques, this proposed method presents a structured framework that enables precise emotional insights from users before finalizing the design.
... facial recognition and fingerprint recognition, have been widely deployed in real-life applications, the potential dangers of elaborate forgery or clandestine duplication are still unavoidable. Compared with the widely used biometrics such as fingerprint, face, speech, etc electroencephalogram (EEG) has its unique advantages of high time resolution and difficult to camouflage since it is directly generated from the neural activities of central nervous system [1]. Therefore, the brain activity recorded via EEG provides a new path for biometric recognition and some other braincomputer interface applications [2,3], meeting the basic identification requirements [4,5] as a new cognitive biometric trait. ...
Article
Full-text available
Objective. Electroencephalogram (EEG) signals are promising biometrics owning to their invisibility, adapting to the application scenarios with high-security requirements. However, It is challenging to explore EEG identity features without the interference of device and state differences of the subject across sessions. Existing methods treat training sessions as a single domain, affected by the different data distribution among sessions. Although most multi-source unsupervised domain adaptation (MUDA) methods bridge the domain gap between multiple source and target domains individually, relationships among the domain-invariant features of each distribution alignment are neglected. Approach. In this paper, we propose a MUDA method, Tensorized Spatial-Frequency Attention Network (TSFAN), to assist the performance of the target domain for EEG-based biometric recognition. Specifically, significant relationships of domain-invariant features are modeled via a tensorized attention mechanism. It jointly incorporates appropriate common spatial-frequency representations of pairwise source and target but also cross-source domains, without the effect of distribution discrepancy among source domains. Additionally, considering the curse of dimensionality, our TSFAN is approximately represented in Tucker format. Benefiting the low-rank Tucker Network, the TSFAN can scale linearly in the number of domains, providing us the great flexibility to extend TSFAN to the case associated with an arbitrary number of sessions. Main results. Extensive experiments on the representative benchmarks demonstrate the effectiveness of TSFAN in EEG-based biometric recognition, outperforming state-of-the-art approaches, as verified by cross-session validation. Significance. The proposed TSFAN aims to investigate the presence of consistent EEG identity features across sessions. It is achieved by utilizing a novel tensorized attention mechanism that collaborates intra-source transferable information with inter-source interactions, while remaining unaffected by domain shifts in multiple source domains. Furthermore, the electrode selection shows that EEG-based identity features across sessions are distributed across brain regions, and 20 electrodes based on 10–20 standard system are able to extract stable identity information.
... Since the beginning of EEG experimentation for user authentication purposes, numerous experiments have been performed, employing different EEG acquisition protocols, applying a variety of EEG montages raging from single channel up to 128-channel devices, leveraging both traditional machine learning and deep learning techniques with noticeable accuracy [36]. In this context, researchers are often required to decide on the signal-acquisition specifics regarding the brain areas that will be monitored as well as simultaneously determine which of these acquiring channels can generate useful information for a given EEG-user authentication protocol. ...
Article
Full-text available
The employment of Electroencephalography (EEG) in the User Authentication (UA) scientific research has recently unlocked state-of-the-art experimentation, aiming at identifying and authenticating individuals given their brainiac activity within specific contexts of use. Indeed, utilizing EEG signals that are derived from brainiac activities can be used for tackling existing UA security threats such as shoulder surfing, thus providing a novel solution to contemporary security problems in traditional knowledge-based user authentication. In this survey, we aim to complete previous literature surveys by providing a systematic classification and presentation of existing research that is based on the following pillars: a) the user experimental setup, with an emphasis on the applied EEG-acquisition protocols (e.g., rest, external stimuli driven, mental and hybrid); b) the artificial intelligence techniques employed and finally c) the security and privacy preservation aspects. The reviewed papers cover a broad area of experiment protocols and various algorithms used for EEG signal classification. Moreover, most cited works include results from more than one experiments with different approaches and configurations. This leads to a discussion on best practices for EEG-based User Authentication and conclusions suggesting future research directions that consists, among others, of considering homomorphically encrypted biometric templates for information leakage prevention through federated learning approaches in decentralized architectures. We anticipate that the present literature review will provide a roadmap for future research by considering efficiently and effective EEG-based User Authentication methods while at the same time preserving privacy.
... EEG bands or rhythms are classified based on frequency ranges, that are named after Greek letters delta band δ (1 to 4 Hz), theta band θ (4 to 8 Hz), alpha band α (8 to 12 Hz), beta band β (13 to 25 Hz) and gamma band γ (> 25 Hz). (Niedermeyer and da Silva, 2005;Khalifa et al., 2012). The results offered by (Juhl et al., 2010) found that there is slow wave activity and reduced alpha activity before a cognitive dysfunction coupled with hypoglycemia. ...
Conference Paper
During COVID 19 pandemic the global needs for online shopping, ride-sharing transportation, and food delivery services have been dramatically increased. The drivers who suffer from diabetes especially low blood sugar level (Hypoglycemia) are more likely at risk than others. Earlier literature has revealed that hypoglycemic issues in patients with diabetes are correlated with significant changes in scalp electroencephalography (EEG); signals amplitude (time domain) or power spectral density (frequency domain). In addition, Haret rate variability HRV which reflects the balance between the sympathetic and parasympathetic nervous system has been proven as one of the indicators of Hypoglycemia. The aim of this paper to propose a conceptual design of a Vehicle detection system using both EEG and HRV measures at the same time in real time feed using non-invasive sensors to reduce the potential of driver’s cognitive dysfunction.
Article
Sleep stage scoring, traditionally done manually by specialists through the inspection of neurophysiological data from sleep studies, is a labor-intensive, monotonous, and time-consuming activity. This has led to an increased interest in the development of Automated Sleep Stage Classification (ASSC) technologies. Such systems are vital for assisting medical professionals in diagnosing and managing sleep-related disorders and neurological conditions, including Alzheimer's disease. This paper presents a cutting-edge classification technique that combines deep learning strategies, delivering outstanding outcomes. It also reviews progress and hurdles in current methods of sleep stage determination using Electroencephalogram (EEG) signals, covering steps like preprocessing, feature detection, and categorization. The paper's goal is to unveil a new classifier design that promises real-time, high-accuracy solutions recognized by the scientific community. This includes the classification of EEG signals into different patient sleep stages: Wake, N1, N2, N3, and REM. Employing a robust classifier system within the Electroencephalography Analysis System (EAS) based on a Brain-Computer Interface (BCI), this system utilizes hybrid classifiers beginning with feature extraction methods such as WDT and PCA, followed by a combination of BiLSTM and LightGBM classifiers. The process starts with training the BiLSTM model on raw EEG data to learn temporal patterns and feature extraction. Features from the BiLSTM outputs are then used as inputs for LightGBM, creating a potent classification system. Unlike previous approaches that often required multiple EEG channels and longer epochs, this research introduces an effective method for 10-second epochs from a single-channel EEG, incorporating novel statistical features and utilizing the PhysioNet Sleep Database, EDFx sleep DATA. The proposed method has shown an average classification sensitivity of 92.1%, specificity of 98.8%, and an overall accuracy of 97.42% using a decision tree classifier, outperforming previous studies in classification accuracy.
Article
Since the first recording in 1924, modern developments in technology have enabled human electroencephalogram (EEG) acquisition as a non-invasive process, enabling a multitude of opportunities to learn about human brain dynamics. With the capability to tap into localized brain dynamics, it is possible to correlate event-related potentials such as meditation, concentration, and motor control with localized brain activities, opening up a broad spectrum for exploration and implementation in prosthetic, control, and brain computer interfaces. In this work, an attempt has been made to explore human emotions and other intelligent states that can be interpreted to automate and control electrical appliances for differently abled people. A smart home automation model has been designed and implemented using a Think Gear Application-specific integrated circuit (ASIC) Module (TGAM) EEG sensor module integrated with a Bluetooth module, which serves as the core for control applications. A combination of external triggers and brain states are interpreted and forwarded to gain control of the connected appliances via a local server over the internet. Equipped with internet connectivity and Internet of Things (IoT), the system also facilitates long-distance communication and control, which can be easily translated to industrial control and automation applications. Based on a single-channel analog EEG signal acquisition module, this study details the development of a Brain Computer Interface (BCI) control and monitoring system for smart home automation with blink and attention features. The designed BCI system has a large bandwidth of 400 Hz, an easy setup, Morphological Component Analysis (MCA) based blink detection, monitoring, control of devices, and high accuracy at a low cost. Each subject completed three trials separated by one minute. The smart devices underwent testing in two states, namely on and off. The response time and accuracy of the system were recorded for each trial. The average system response time for the devices was determined to be 17.13 sec for switching ON and 20.66 sec for switching OFF, with an accuracy of 98.73% and 95.75% for ON and OFF states respectively.
Article
Full-text available
This paper reports a new approach on identifying the individuality of persons by using parametric classification of multiple mental thoughts. In the approach, electroencephalogram (EEG) signals were recorded when the subjects were thinking of one or more (up to five) mental thoughts. Autoregressive features were computed from these EEG signals and classified by Linear Discriminant classifier. The results here indicate that near perfect identification of 400 test EEG patterns from four subjects was possible, thereby opening up a new avenue in biometrics. Keywords—Autoregressive, Biometrics, Electroencephalogram, Linear discrimination, Mental thoughts. I. INTRODUCTION
Article
Full-text available
Electroencephalographic measurements are commonly used in medical and research areas. This review article presents an introduction into EEG measurement. Its purpose is to help with orientation in EEG field and with building basic knowledge for performing EEG recordings. The article is divided into two parts. In the first part, background of the subject, a brief historical overview, and some EEG related research areas are given. The second part explains EEG recording.
Article
Full-text available
We present a novel idea for user authentication that we call pass-thoughts. Recent advances in Brain-Computer Interface (BCI) technology indicate that there is potential for a new type of human-computer interaction: a user transmitting thoughts directly to a computer. The goal of a pass-thought system would be to extract as much entropy as possible from a user's brain signals upon "transmitting" a thought. Provided that these brain signals can be recorded and processed in an accurate and repeatable way, a pass-thought system might provide a quasi two-factor, changeable, authentication method resistant to shoulder-surfing. The potential size of the space of a pass-thought system would seem to be unbounded in theory, although in practice it will be finite due to system constraints. In this paper, we discuss the motivation and potential of pass-thought authentication, the status quo of BCI technology, and outline the design of what we believe to be a currently feasible pass-thought system. We also briefly mention the need for general exploration and open debate regarding ethical considerations for such technologies.
Article
Cognitive biometrics is a new authentication scheme that utilises the cognitive, emotional, and conative state of an individual as the basis of user authentication and/or identification. These states of mind (and their derivatives) are extracted by recording various biosignals such as the EEG, ECG, and electrodermal response (EDR) of the individual in response to the presentation of the authentication stimulus. Stimuli are selected which elicit characteristic changes within the acquired biosignal(s) that represent unique responses from the individual. These characteristic changes are processed using a variety of machine learning algorithms, resulting in a unique signature that identifies or authenticates the individual. This approach can be applied in both static mode (single point of authentication), or in continuous mode, either alone, or in a multi-modal approach. The data suggest that the classification accuracy can reach 100% in many scenarios, providing support for the efficacy of this new approach to both static and continuous biometrics.
Article
This work deals with an optimization of the EEG–based biometric classification. Recent results and other published works indicate that it is possible to identify subjects on the base of their and µ rhythms spectral envelopes. Our latest work was targeted to the explanation of some past results and to the optimization of unfeasibly high frequency resolution used previously. The results showed that the problems with identification of some subjects have roots in the EEG signal itself. Further, it is shown that the frequency resolution might be reduced by 50% (from 1/180Hz to 1/90Hz) reducing the required segment length to identify subject from 180 sec to 90 sec. Despite of the encouraging results it is clear that linear FFT parameterization is not suitable for this application and other parameterizations are under investigation now.
Conference Paper
Biometrics based on electroencephalogram (EEG) signals is an emerging research topic. Several recent results have shown its feasibility and potential for per- sonal identification. However, they all use a single task (e.g., signals recorded during imagination of repetitive left hand movements or during resting with eyes open) for classifier design and subsequent identification. In contrast with this, in this paper multiple related tasks are used simultaneously for classifier learning. This mechanism has the advantage of integrating informa- tion from extra tasks and thus hopefully can guide clas- sifier learning in a hypothesis space more effectively. Experimental results on EEG-based personal identifi- cation show the effectiveness of the proposed multitask learning approach.
Article
Abstract not available. ©1996 American Institute of Physics
Article
We estimated the genetic and nongenetic (environmental) contributions to individual differences in the background EEG power spectrum in two age cohorts with mean ages of 26.2 and 49.4 years. Nineteen-lead EEG was recorded with eyes closed from 142 monozygotic and 167 dizygotic twin pairs and their siblings, totaling 760 subjects. We obtained power spectra in 24 bins of 1 Hz ranging from 1.0 to 25.0 Hz. Generally, heritability was highest around the alpha peak frequency and lower in the theta and delta bands. In the beta band heritability gradually decreased with increasing frequency, especially in the temporal regions. Genetic correlations between power in the classical broad bands indicated that half to three-quarters of the genetic variance can be attributed to a common source. We conclude that across the scalp and most of the frequency spectrum, individual differences in adult EEG are largely determined by genetic factors.
Cognitive Biometrics: A Novel Approach to Person Authentication, ch 7, in Continuous Authentication using Biometrics: Data, Methods, and Models ISBN: 1613501293, publication date: 30 September
  • K Revett
Revett, K., Cognitive Biometrics: A Novel Approach to Person Authentication, ch 7, in Continuous Authentication using Biometrics: Data, Methods, and Models, (eds) I. Traore and A. Ahmed, Information Science Reference, ISBN: 1613501293, publication date: 30 September, 2011.