Data analysis by using machine

International Journal on Computational Sciences & Applications (IJCSA) Vol.4, No.6, December 2014
DOI:10.5121/ijcsa.2014.4602 19
DATA ANALYSIS BY USING MACHINE
LEARNING ALGORITHM ON CONTROLLER
FOR ESTIMATING EMOTIONS
Tanu Sharma and Dr.Bhanu Kapoor
Department of Computer Engineering, Chitkara University, Baddi,Solan
ABSTRACT
Emotions are an unstoppable and uncontrollable aspect of mental state of human. Some bad situations give
stress and leads to different sufferings. One can’t avoid situation but can have awareness when body feel
stress or any other emotion. It becomes easy for doctors whose patient is not in condition to speak. In that
case person’s physiological parameters are measured to decide emotional status. While experiencing
different emotion, there are also physiological changes taking place in the human body, like variations in
the heart rate (ECG/HRV), skin conductance (GSR), breathing rate(BR), blood volume pulse(BVP),brain
waves (EEG), temperature and muscle tension. These were some of the metrics to sense emotive co-
efficient. This research paper objective is to design and develop a portable, cost effective and low power
embedded system that can predict different emotions by using Naïve Bayes classifiers which are based on
probability models that incorporate class conditional independence assumptions. Inputs to this system are
various physiological signals and are extracted by using different sensors. Portable microcontroller used
in this embedded system is MSP430F2013 to automatically monitor the level of stress in computer. This
paper reports on the hardware and software instrumentation development and signal processing approach
used to detect the stress level of a subject.To check the device's performance, few experiments were done in
which 20 adults (ten women and ten men) who completed different tests requiring a certain degree of effort,
such as showing facing intense interviews in office.
KEYWORDS
Emotions, Model, equation, Result, pattern, GSR, BVP, Temperature, Microcontroller
1.INTRODUCTION
Human experiencing emotions in a higher magnitude differ from those who can regulate these
emotional experiences; such type of factor is named as Emotional Intelligence (EI).Emotional
intelligence has four aspects that are also known as branches. These branches perceive emotion to
facilitate thoughts, understanding emotions and managing emotions[2].
This paper explains contribution to work in perceiving emotions. As perceiving emotion has
ability to identify emotion in oneself and others. With this it also has ability to tell difference
between honest and dishonest emotions. Recognizing emotions is not just dependent upon facial
expressions rather many more kinds of cues are there now like: voice, gestures, actions and
biofeedback modalities. Various biofeedback modalities which exist are electromyography
(EMG) in which muscle contraction and relaxation is measured, temperature change is accessed
via fingertip thermometers. Resistance of skin influenced by sweat is evaluated (GSR),

20
Cardiovascular activity is measured via heart rate [3]. Emotion depends upon ANS (autonomic
nervous system) of the person. As emotion varies and it bring changes in sympathetic nerve of the
ANS in excited condition, the sweat is secreted form the sweat gland and it decreases the GSR
and heart rate is also under continuous control of autonomic nervous system which also increase
heart rate [9].
This paper targets on the GSR with parallel work based upon BVP and temperature analysis.
Measuring instrument was designed and developed by authors. It is based on low power
consuming microcontroller(MSP430F2013).This measuring instrument can measure different
biofeedback modalities and can also predict emotion of a person. In particular, popular learning
algorithm was employed: Naïve Bayes Classifier. The Bayesian Classification represents a
supervised learning method as well as a statistical method. It assumes a probabilistic model and it
allows us to capture uncertainty about the model in a principled way by determining probabilities
of the outcomes.
Through this healthcare application certain experiments was conducted on different people
working in same office, by stimulating them with different scenarios like in meetings, before
presentations, in breaks, people appearing for interview etc. Monitoring and analyzing emotions
is important as it contains information that can help in improving human wellbeing. Emotions
which were observer by system are joyful, anxiety, calm and will also reflect illness in case of
high temperature. Twenty subjects were chosen to take part in this experiment with age from 23-
55.This was carried out for 10 male and 10 female subjects with consistent experimental setup.
When there is variation in ANS, it gives changes in biofeedback modalities as result emotions are
observed. Table 1[4] gives an example of same.
Table I.Example of biofeedback modalities In autonomic activity
Emotions GSR BVP Temperature
Anxiety Decreases Decreases Decreases
Joyful Increases Normal Normal
Calm Normal Normal Normal
Illness Normal Increases Increases
2.State of art
There exist different studies which try to detect with subject emotions in different manner. Study
for blind people [5] Analyze autonomic nervous system activity is a subject. Estimation of
emotion on one modality (GSR) [6] has shown possibilities to achieve outcome. Varying Heart
Rate is another parameter used to measure different stress levels [7].Analysis of Emotion
Recognition from GSR Based on PSO [8] is an application, which came up with different idea. A
stress sensor has implemented different algorithm techniques on GSR [4].
3.Architecture of Emotion Estimation System
The architecture of emotion estimation monitoring system is shown (fig1), which has mechanism
to measure different bio- modalities or bio-signals (GSR/BVP/Temperature). In this system
validation of subject is done at priority. It sense biosignals from human and then sends it to

21
microcontroller (MSP430F2013). This is a new model which is solution to previous conventional
models, with more portability low cost and more power efficient.
Fig.2.Biofeedback monitoring system which display different parameters connected to microcontroller [6]
Firstly system starts sensing galvanic skin response (GSR).GSR is based on sweat gland activity.
The autonomic nervous system (ANS) controls eccrine sweating, with this skin conductance
reflects arousal of the sympathetic ANS which accompanies different psychological processes.
GSR is measured by passing a small current through a pair of electrodes placed on a surface of
skin.
Fig. 2 Electrodes used as sensor for GSR
That measured value can be in form of conductance, resistance, current value or voltage value.
By placing two fingers on the surface of electrodes Galvanic Skin Response (GSR) was
measured. To have value, resistance was used. To voltage divider was formed to calculate
resistance.

22
Fig. 3 GSR Input setup
Where, Rs/R1 is the resistance of the skin.
The GSR resistance oscillates between 10 k and 10 M [4].As ADC in MSP430 is 12 bits so it
has been observed that ADC saturates at 2.30V and resolution is 0.570mv.A minimum tension
level a human can have is 136mv, this helped to avoid operational amplifier and it was not
included. By avoiding operational amplifier, efficiency of system got improved because it became
less power consuming sensor system.
Blood volume pulse varies when heart rate variation is observed. It is due to the synergistic action
of the two branches of the autonomic nervous system (ANS) - vagus nerve and sympathetic
nerve. Vagus nerve: Vagal nerves innervate the S-A node, the atrioventricular node and the atrial
muscle. The stimulation of the vagus nerve slows down the heart rate. Sympathetic nerve:
Sympathetic nerve fibers innervate the entire heart, including the sinus node and A-V node.
Increased activity of the sympathetic nerves results in the increase in heart rate of human
body.BVP is recorded by technique known as a photoplethysmography.
Photoplethysmography (PPG) works by placing an individual’s finger tip or ear-lobe between two
parts of a transducer consisting of a light source and a light sensitive detector. A beam of infrared
light is projected towards the detector. The blood in the finger or ear-lobe scatters the light in the
infrared range, and the amount of light reaching the detector is inversely related to the volume of
blood in the skin periphery. This is shown in fig 3.
Fig 4 Arrangement of plethysmograph components
(1)

23
Heart rate is derived by measuring either the average or instantaneous time interval between two
successive pulses. An average rate i.e. beats/min. is calculated by counting the number of pulses
in a given time. The true picture of the heart’s responds to exercise, stress and environment. This
is done by measuring the time (T) in seconds, between the two consecutive pulses and converting
this time into beats/min. [9]
Fig .5 Beats/min. = 60/T, calculates beat per minute in human body
Heart rate reserve is the difference of the maximum heart rate and the resting heart rate. Heart
Rate Reserve = Maximum HR – Resting HR.
(HR is Heart Rate of a human body) (2)
For temperature when we feel stress muscles get contracted (tense), with this temperature of body
get reduced because of less blood reaches the fingers of the patient. This coldness can be
experienced in our hands when we are stressed. An example for the same can be, when appearing
for job interview that nervousness can be experienced by anyone. It is common for human
temperature to have readings as low as 70 to 80 degrees Fahrenheit
The eZ430-F2013 is a complete MSP430 development tool providing all the hardware and
software to evaluate the MSP430F2013 and complete an entire project in a convenient USB stick
form factor. All 14-pins on the MSP430F2013 are accessible on the MSP-EZ430D target board
for easy debugging and interfacing with peripherals. Using the integrated 16-bit SigmaDelta
analog-to digital converter, the MSP430F2013 provides all the required elements for interfacing
.Following algorithm help to prepare AD-converter:
SD16CTL = SD16REFON + SD16SSEL_1 + SD16XDIV_2 + SD16DIV_1; VREF=1.2V,
Clock=SMCLK, Clock-Divider2=16, Clock-Divider1=2 500kHz MF)
P1SEL |= BIT3; (internal reference voltage at P1.3 (Pin5) )
SD16INCTL0 = SD16INCH_4; (using input A4 (P1.1=A4+, P1.2=A4-))
SD16AE = SD16AE1 ;( negative inputs are internally connected to Vss (GND); P1.1 an A4+)
SD16CCTL0 = SD16IE + SD16UNI ;( Interrupt + unipolar Input, and continues mode)
SD16CCTL0 |= SD16XOSR; oversample rate
SD16CCTL0 |= SD16OSR_512; (over sample rate is 512)
SD16CCTL0 |= SD16SC; (start conversion)
_BIS_SR (GIE); (enter global interrupt)
The result sample rate is clock frequency divided by modulation frequency of the SD16, divided
by the oversampling rate
T
Time

24
3.1 Signals Processing
In GSR resistance calculation of Body is done using Voltage Divider Circuit with 10K variable
resistance. That variable voltage according to the body resistance is fed into the msp430
microcontroller for analogue processing which is send to the 8051 microcontroller via 3 wire self
designed protocol which is further send to the pc from serial port using UART Communication.
In BVP Sensor it measure Blood Volume Pulse Rate we are using a sensor that gives us the high
pulse in synchronization with the heart rate. A light is fed into the human finger using a LED
which is reflected back from the amount of blood accordingly the phototransistor receives the
amount of light and gives the output voltage that is fed into the msp430 microcontroller for
analogue processing which is send to the 8051 microcontroller via 3 wire self designed protocol
which is further send to the pc from serial port using UART Communication. In Temp Sensor
(LM-35) MSP microcontroller has temperature sensor inbuilt with which temp is measured
directly using the internal SD16 of MSP. The digital output value is send to the 8051
microcontroller via 3 wire self designed protocol which is further send to the pc from serial port
using UART Communication.
The architecture of emotion estimation with Biofeedback (biosignals) monitoring system is given
below (fig.1), which is having the mechanism of stimulation and measuring (with self designed
instrument) and then finally the estimation of the emotional state (anger/happy etc) of person,
which is the variation of the objective evaluation (experimental value) and ideal or expected
values (subjective).
Fig 6 Emotion Estimation block diagram [6]
3.2Hardware/Software Requirements
To design and implement the actual device, main components used are listed below:
• Electrodes (silver)
• light emitting diode (LED)
• phototransistor
• LM35
• MSP430F2013
• Display (LCD )( 16 Character x 2 Line) /Personal Computer
• IAR/Code studio compiler
Emotions
Induction
Measurement
Of Different ANS
values
Emotion
Estimation
Subjective
Assessment

25
4.Mapping of emotion Model with Emotions
The main objective is this work was to design and develop real time monitoring system which is
capable to estimate different emotions, specially for the people who all can’t express their
emotions like suffering from paralyzed body etc. This system will be user friendly as it will be
following diagram shown in fig 5.
Emotion Physiological Feeling of producing response
Emotion
stimulus
Fig 6 Sequence of emotion
The expected values of different biofeedback modalities are mapping with different ranges of
emotions are as mentioned in the table given below [10].
TABLE II. values and emotions
Stressed calm Joy Fear/
Aroused
GSR
Amplitud
e
0.28 0.15-
0.25
0.15 0.29
BVP 63 72 73 85
Temp 98 98.2 98.4 96.4
Stress is seen as a state of emergency that is preceded by arousal due to an external stimulus.
External stimulus can be considered anything which is creating distractions. After the factor
causing stress (the stressor) disappears; the body relaxes get calm and returns to a normal state.
Figure 6 shows the general case with more relationships between the four different states
depicting the inner process of emotions. These states are interrelated by each other and are also
associated by external stimuli. Emotion varies when input parameter GSR/TEM/BVP varies,
depending upon input values result (emotions) are detected.
Fig.7 Relation between different states
Normal
Arousing
Joy
Stress

International Journal on Computational Sciences & Applications (I
In this paper we consider a simplified setting assuming that a person is either in the normal state
or in a stressed state. The change between the two states can be sudden or incr
arousal is more rapid and relaxation takes considerably longer. [11] In following fig 3 we can see
various emotions are categorized which are based upon the degree of arousal from low to high
and valence i.e. positive to negative of emo
Fig. 8 Two dimensional emotion models with four quadrants
Following are observation from various Activities:
• when anxiety will increases then at same time with skin conductivity GSR will decreases
and HR/BVP will increase
• When parson will be joyful then with skin conductivity GSR will increases and HR/BVP
will (increase/decrease)
• When person is calm no change is observed
All features were extracted from experiment held by author, and then they were provided as input
to learning systems, which were trained to differentiate the stress from the different state of a
person.
Bayesian reasoning was applied to predict the emotional state. Bayesian is responsible for
decision making and inferential statistics by using probabilities techniques [11].
Bayes Theorem:
Method above computes the conditional probabilities of the different classes given the values of
attributes of an unknown sample and then the classifier will predict that the sample belongs to the
class having the highest posterior probability in that case. Insta
ndimensional feature vector, (x1,x2,…,xn), sample is classified to a class c from a set of possible
classes C according to the maximum a posteriori (MAP) decision rule:
Classify(x1,x2,…..xn)=argmax p(C=c)
D: Set of tuples
Each Tuple is an ‘n’ dimensional attribute vector
or in a stressed state. The change between the two states can be sudden or incremental; typically,
and valence i.e. positive to negative of emotions.[10]
Fig. 8 Two dimensional emotion models with four quadrants
Following are observation from various Activities:
when anxiety will increases then at same time with skin conductivity GSR will decreases
and HR/BVP will increase
ll be joyful then with skin conductivity GSR will increases and HR/BVP
will (increase/decrease)
When person is calm no change is observed
s, which were trained to differentiate the stress from the different state of a
decision making and inferential statistics by using probabilities techniques [11].
(3)
above computes the conditional probabilities of the different classes given the values of
class having the highest posterior probability in that case. Instance is represented by an
classes C according to the maximum a posteriori (MAP) decision rule:
Classify(x1,x2,…..xn)=argmax p(C=c) (4)
Each Tuple is an ‘n’ dimensional attribute vector
6, December 2014
26
emental; typically,
when anxiety will increases then at same time with skin conductivity GSR will decreases
ll be joyful then with skin conductivity GSR will increases and HR/BVP
s, which were trained to differentiate the stress from the different state of a
above computes the conditional probabilities of the different classes given the values of
nce is represented by an

International Journal on Computational Sciences & Applications (I
X : (x1,x2,x3,…. xn)
Let there be ‘m’ Classes : C1,C2,C3…Cm
Classifier predicts X belongs to Class Ci if
P (Ci/X) > P(Cj/X) for 1<= j <= m , j <> i
Maximum Posteriori Hypothesis
P(Ci/X) = P(X/Ci) P(Ci) / P(X)
Maximize P(X/Ci) P(Ci) as P(X) is constant With many attributes, computing of this is
expensive to evaluate P(X/Ci).Naïve Assumption of “class conditional independence”
P (X |Ci)=
P(X/Ci) = P(x1/CI) * P(x2/CI) *…* P(xn/ CI)
The conditional probability in the given above formula is obtained from the estimates of the
probability mass function using training data of that
assumption may not be a realistic model of the probabilities involved; it may still permit
relatively accurate classification performance [6].
5.PROCEDURE FOR EXPERIMENTATION
Prerequisites to start the experiment are
1. The author arrived with a setup to various areas of office where different activities were
going on like meeting, interviews, lunch break, official work, tea time and official
presentations.
2. Readings were taken three times before the task,
3. Experiment was done with age from 23
4. As circuit is portable designed and was also helping with feature with low power supply,
having 1.8V of power supply.
5. Before performing above experiment, permission was granted by offi
6. Different emotions were detected, as it also helped in professional growth ( by building
strong emotions).
7. While experiment, subject was not allowed to have water as it can change emotion.
5.1.The experiment
All the events or activities were found to trigger specific emotions. Both genders were
considered 20 odd male and female. Sample was taken from various people were involved in
different activities and reading were taken suddenly to avoid manipulated emoti
to put their left and right fingers, one each on to the sensors and also check each parameter turn
wise. Then according to the connections showed in diagram (fig.2,3) the GSR,BVP and
temperature values are sensed by sensors , on the basis
Initially, the person’, under neutral conditions is measured, that serves as the reference for us to
Let there be ‘m’ Classes : C1,C2,C3…Cm
Classifier predicts X belongs to Class Ci if
P (Ci/X) > P(Cj/X) for 1<= j <= m , j <> i
Hypothesis
P(Ci/X) = P(X/Ci) P(Ci) / P(X)
(5)
P(X/Ci) = P(x1/CI) * P(x2/CI) *…* P(xn/ CI) (6)
probability mass function using training data of that scenario. Moreover the independence
relatively accurate classification performance [6].
5.PROCEDURE FOR EXPERIMENTATION
Prerequisites to start the experiment are given below:
The author arrived with a setup to various areas of office where different activities were
Readings were taken three times before the task, within task and after task.
Experiment was done with age from 23-55.
As circuit is portable designed and was also helping with feature with low power supply,
having 1.8V of power supply.
Before performing above experiment, permission was granted by officials.
Different emotions were detected, as it also helped in professional growth ( by building
While experiment, subject was not allowed to have water as it can change emotion.
different activities and reading were taken suddenly to avoid manipulated emotions. Person need
temperature values are sensed by sensors , on the basis of these values emotions are predicted.
6, December 2014
27
scenario. Moreover the independence
The author arrived with a setup to various areas of office where different activities were
within task and after task.
As circuit is portable designed and was also helping with feature with low power supply,
cials.
Different emotions were detected, as it also helped in professional growth ( by building
While experiment, subject was not allowed to have water as it can change emotion.
ons. Person need
of these values emotions are predicted.

28
estimate the variance of the values, in different emotional states. Once the state is reached,
person tends to be in that state for a finite amount of time. The total time of stimulus for each
emotion is in between 5 to10 minutes and with a gap of 2-3 minutes is must between different
emotions, so to full fill these criteria emotion was estimated thrice. This design is not suitable for
people having Hyperhidrosis. This is disease of excessive sweating, which becomes drawback of
system.
5.2Observations and Results
Values of GSR between < 5 K ohms indicates high level of brain arousal, and >25Kohms
indicates low arousal (calm level).The normal heart rate of an adult varies between 60 to
90bpm.At rest, an adult has an average heart rate of 72bpm [7].
Table 3. Values of different people
PERSON GSR
BVP TEMPERATURE Emotions
Subject 1
Low Low Low Anxiety
Subject 2
Low High Low Anxiety
Subject 3
High Low Low RELAX
Subject 4
High High Low JOYFULL
Subject 5
Normal High Low RELAX
Subject 6
Normal Low Low RELAX
Subject 7
High Normal Low JOYPULL
Subject 8
Low Normal Low Anxiety
Subject 9
High High Normal JOYFULL
Subject 10
High Low Normal RELAX
Subject 11
Normal Normal Normal RELAX
Subject 12
Low Low Low Anxiety
Subject 13
Subject 14
Subject 15
Low Low Normal Anxiety
Subject 16
Low High Normal Anxiety
Subject 17
Subject 18
Low Low Low Anxiety

29
Subject 19
Subject 20
6.CONCLUSION
Emotion regulation is a capability of emotional Intelligence. The work finds its
application in the domains for personnel usage especially for people who are paralyzed.
Situations and emotions where there is great arousal like anger, fear, horror and
melancholy could be easily identified, while lower arousal emotions like calm and sad
were meagerly distinguishable. Research also talks of emotions like joy as high arousal
emotions and hence easily distinguishable. The heart rate varies between individuals and
gives different reading according to their emotions. The normal human heart rate at rest is
60 to 80 bpm. At rest, an adult has an average heart rate of 72bpm and GSR varies from 0
to 35 K ohms’. Further empirical studies need to be conducted to give a clearer picture on
this aspect. However the present work is an attempt to such an end and hopes to find out
methods and ways to achieve the goal of affective communication. Future working is also
being done parallel by adding more input parameters to the same.
6.ACKNOWLEDGEMENTS
This research paper got possible through the help and support from: My guide, parents,
family, friends, and god. Especially, I would like to acknowledge my gratitude toward the
following significant advisor of my work:
Foremost, I would like to thank Dr. Bhanu Kapoor for his most support and
encouragement. With this my sincere thank to my parents, family who provide all
essential support. This research paper would not be successful without all support.
Fig 9 Variation in GSR Fig 10 Variation in BVP

30
REFERENCES
[1] Patrick E. McSharryb, Gari D. Clifforda,”( May 25, 2004) A realistic coupled nonlinear artificial, BP,
ECG and respiratory signal generator for assessing noise performance of biomedical signal
processing algorithms” aHarvard-MIT Division of Health Sciences, address-45 Carleton St.,
Cambridge MA 02142, US. B Department of Engineering Science at University of Oxford in Parks
Road, Oxford OX1 3PJ, UK; Volume 5467
[2] Cynthia j. Tanis, (2008 )“The effects of heart rhythm variability biofeedback with emotional
regulation on the athletic performance of women collegiate volleyball players”, a dissertation
presented in partial fulfillment of the requirements for the degree doctor of philosophy,
[3] Muhammad Tausef, (2012). “Human emotion Recognition using smart sensors”, Thesis of Master of
Engineering in Massey University,new zealand feb
[4] María Viqueira Villarejo, Begoña García Zapirain and Amaia Méndez Zorrill,( May 2012) “A Stress
Sensor Based on Galvanic Skin Response (GSR) Controlled by ZigBee” Article sensors10 ISSN
1424-8220
[5] Massot B., Baltenneck N., Gehin C., Dittmarm A., McAdams , (2011) ”E. EmoSense: An Ambulatory
Device for the Assessment of ANS Activity—Application in the Objective User Evaluation of Stress
with the Blind”. Sensors J. ;12:543–551.
[6] Tanu S,Bhanu K (2013) “Emotion estimation of physiological signals by using low power embedded
system” ,. In Proceedings of the conference on advances in communication and control systems.
Atlantis Press, April.The authors - Published by Atlantis Press
[7] Jongyoon C., Gutierrez-Osuna R. (2011),”Removal of Respiratory Influences from Heart Rate
Variability in Stress Monitoring.” IEEE Sensor J. ;11:2649–2656.
[8] Wu G., Liu G., Hao M.( October 2010), “ The Analysis of Emotion Recognition from GSR Based on
PSO. Proceedings of International Symposium on Intelligence Information Processing and Trusted
Computing; Huanggang,” China. 28–29.
[9] Patrick E. McSharryb, Gari D. Clifforda,” (May 25, 2004) A realistic coupled nonlinear artificial, BP,
ECG and respiratory signal generator for assessing noise performance of biomedical signal
processing algorithms” aHarvard-MIT Division of Health Sciences, address-45 Carleton St.,
Cambridge MA 02142, US. B Department of Engineering Science at University of Oxford in Parks
Road, Oxford OX1 3PJ, UK; Volume 5467
[10] Bakker, J., Pechenizkiy, M. & Sidorova, N. (2011) What's your current stress level? Detection of
stress patterns from GSR sensor data, In Proceedings of ICDM Workshops. 2nd HACDAIS
Workshop @ ICDM 2011 (HACDAIS 2011), pp. 573-580
[11] Jing zhai1, armando barreto,( 2006) “ stress detection in computer users through non-invasive
monitoring of physiological signals” electrical and computer engineering department, 2biomedical
engineering department florida international university, Biomed Sci Instrum. ;42:495-500
[12] Latha M ,Senthilmurugan M. and Dr. Malmurugan N. ,( April 2011)” Classification in EEG-Based
Brain Computer Interfaces Using Inverse Model” International Journal of Computer Theory and
Engineering, Vol. 3, No. 2, ISSN: 1793-8201
[13] Hsun-Hsien Chang and Jos´e M. F. Moura, (2010) “Biomedical Signal Processing” ed .Myer Kutz , in
Biomedical Engineering and Design handbook, 2nd edition , Volume 1, pp.559-579. McGraw Hill.
[14] Dalila Trad1,Tarik Al-ani1,,( January 2010) “Signal Processing and Classification Approaches for
Brain-computer Interface” 2Department of Informatics, ESIEE-Paris, Cité, University of TUNIS
France 3Tunisia, , pp. 386, ISBN 978-953-7619-58-9
[15] Justin Dauwels_ and Franc¸ois Vialatte, “Topics in Brain Signal Processing” Nanyang Technological
University, Singapore(IEEE J-STSP), ISSN1941-0484
[16] María Viqueira Villarejo, Begoña García Zapirain and Amaia Méndez Zorrill, (May 2012 )“A Stress
Sensor Based on Galvanic Skin Response (GSR) Controlled by ZigBee” Article sensors10 ISSN
1424-8220
[17] Trotta A. Kachouri1 , A. L. Ekuakille2,H. Ghariani1,Ben Hmida1, A.,( June 2009) “extracting
electric power from human body supplying neural recording system” international journal on smart
sensing in intelligent systems, Volume. 2,
[18] Fuji Rent, Kazuyuki Matsumotot. Shunji Mitsuyoshij, Shingo Kuroiwat, Gai Lin,”( 2003) Researches
on the emotion measurement system” ‘0-7803-7952-7/03 IEEE.

31
[19] Rana El Kaliouby Rolfe Swinton . ” Measuring emotions through a mobile device across borders,
ages, genders “Copyright © ESOMAR 2012)
[20] Johannes Wagner, Florian Lingenfelser, Elisabeth André, Jonghwa Kim and Thurid Vogt University
of Augsburg, Augsburg,” Exploring Fusion Methods for Multimodal Emotion Recognition with
Missing Data” IEEE trasactions on affective computing October-December 2011 (vol. 2 no. 4) pp.
206-218[emotion Model]
[21] Bakker, J., Pechenizkiy, M. & Sidorova, N. (2011) What's your current stress level? Detection of
stress patterns from GSR sensor data, In Proceedings of ICDM Workshops. 2nd HACDAIS
Workshop @ ICDM 2011 (HACDAIS 2011), pp. 573-580
[22] W. K. Pratt, Digital Image Processing, 3rd ed., Wiley-Interscience, New York, 2001
[23] J. Lin, E. J. Keogh, L. Wei, and S. Lonardi. ,( 2007.)”Experiencing SAX: a novel symbolic
representation of time series”. Data Min. Knowl. Discov., 15(2):107–144,
[24] Bakker, J., Holenderski, L., Kocielnik, R., Pechenizkiy, M. & Sidorova, N. (2012) Stess@work: From
Measuring Stress to its Understanding, Prediction and Handling with Personalized Coaching,
In Proceedings of ACM SIGHIT International Health Informatics Symposium (IHI 2012), ACM
Press, pp. 673-678
[25] Chang, C.-Y.; Zheng, J.-Y.; Wang, C.-J.,( July 2010) ” Based on Support Vector Regression for
EmotionRecognition Using Pshysiological Signals”. In Proceedings of International Joint Conference
on Neural Networks, Barcelona, Spain, 18–23
Authors
Tanu Sharma,Research scholar in the department of computer science and
engineering, Chitkara Un iversity with research interests primarily in development
and designing of embedded systems .Also focuses on implementation of machine
learning algorithm on microcontroller.
Dr. Kapoor started his technical career in 1987 with Texas Instruments, where he held
various roles in the company’s research and development labs. He moved on to other
positions at several electronic design automation companies. He is a senior member of
the Institute of Electrical and Electronics En gineers (IEEE) and the Association for
Computing Machinery (ACM). He has received six U.S. patents, has participated in
various industry panels, and has served as an IEEE conference chairperson. His
written works include more than 50 papers that have been presented at IEEE/ACM
conferences or published in journals.

Data analysis by using machine

More Related Content

Similar to Data analysis by using machine (20)

Recently uploaded (20)

Data analysis by using machine