116
Internet of Things (IoT)
6.9 Conclusions
The chapter employs machine learning methods to develop a model that predicts gender
and age group based on keystroke dynamics features which significantly improve accuracy. Gender and age group alone are not sufficient features to identify the individual user,
but they can be used as additional features. We have used three public authentic datasets
on keystroke dynamics through keyboard, and one dataset through touch screen to verify
whether or not this technique is applicable in both environments. Our proposed approach,
FRNN-VQRS, a new approach to FRNN, achieved a gender and age group prediction accuracy of more than 94% in desktop environment, and 84.74% accuracy in android environment. We have used paired t-test where FRNN-VQRS is most significant than previously
used libSVM by Giot et al. [7]. This is a very positive outcome in keystroke dynamics system
for a single predefined text which can be used as soft biometric additional features in identification/authentication technique which improves the gain accuracy by 3.5% to 14.38%.
As per the obtained results listed in Tables 6.3 and 6.4, gender as well as age group of
the users can be extracted from the typing pattern. It is also observed that both gender
and age group information can be extracted from the dataset collected through the touch
screen device. This is the first time we have used FRNN as per our knowledge on keystroke dynamics datasets instead of the very popular machine learning method, libSVM.
In this chapter, we have also fused these two soft biometric scores with the timing features to enhance the performance of keystroke dynamics user authentication systems. It is
also observed that gender and age group information as extra features increase the user
authentication performance instead of using only gender information. So, this technique
can be used to predict the gender and age group of the Internet users as it is evident from
our experiment, as keystroke dynamics is a common measurable distance-based activity
to monitor the use of the Internet through keyboard/touch screen. It could be used to deal
with the problem of fake accounts and would facilitate creation of a more loyal and authentic social networking sites. This keystroke Dynamics user recognition with inclusion of
personal traits as additional features is the modest and efficient approach.
References
1. Bleha S., Slivinsky C., and Hussien B. Computer-access security systems using keystroke
dynamics. IEEE Transactions on Pattern Analysis and Machine Intelligence, 12(12):1217–1222, 1990.
2. Gaines R., Lisowski W., Press S., and Shapiro N. Authentication by keystroke timing: Some preliminary results. Technical Report Rand Rep. R-2560-NSF, RAND Corporation, Santa Monica, CA
90406, 1980.
3. Joyce R., and Gupta G. Identity authentication based on keystroke latencies. Communications of
the ACM, 33(2):168–176, 1990.
4. Killourhy K., Maxion R., A scientific understanding of keystroke dynamics. Carnegie Mellon
University, Pittsburgh, PA, 2012.
5. El-Abed M., Dafer M., El Khayat R. RHU Keystroke: A mobile-based benchmark for keystroke
dynamics systems. 48th IEEE International Carnahan Conference on Security Technology (ICCST),
Rome, Italy, 2014.
6. Modi S., and Elliott S.J. Keystroke dynamics verification using a spontaneously generated
password. In Proceedings of the 40th Annual IEEE International Carnahan Conference on Security
Technology (ICCST ’06), USA, October 2006, pp. 116–121.
Internet of Things (IoT)
6.9 Conclusions
The chapter employs machine learning methods to develop a model that predicts gender
and age group based on keystroke dynamics features which significantly improve accuracy. Gender and age group alone are not sufficient features to identify the individual user,
but they can be used as additional features. We have used three public authentic datasets
on keystroke dynamics through keyboard, and one dataset through touch screen to verify
whether or not this technique is applicable in both environments. Our proposed approach,
FRNN-VQRS, a new approach to FRNN, achieved a gender and age group prediction accuracy of more than 94% in desktop environment, and 84.74% accuracy in android environment. We have used paired t-test where FRNN-VQRS is most significant than previously
used libSVM by Giot et al. [7]. This is a very positive outcome in keystroke dynamics system
for a single predefined text which can be used as soft biometric additional features in identification/authentication technique which improves the gain accuracy by 3.5% to 14.38%.
As per the obtained results listed in Tables 6.3 and 6.4, gender as well as age group of
the users can be extracted from the typing pattern. It is also observed that both gender
and age group information can be extracted from the dataset collected through the touch
screen device. This is the first time we have used FRNN as per our knowledge on keystroke dynamics datasets instead of the very popular machine learning method, libSVM.
In this chapter, we have also fused these two soft biometric scores with the timing features to enhance the performance of keystroke dynamics user authentication systems. It is
also observed that gender and age group information as extra features increase the user
authentication performance instead of using only gender information. So, this technique
can be used to predict the gender and age group of the Internet users as it is evident from
our experiment, as keystroke dynamics is a common measurable distance-based activity
to monitor the use of the Internet through keyboard/touch screen. It could be used to deal
with the problem of fake accounts and would facilitate creation of a more loyal and authentic social networking sites. This keystroke Dynamics user recognition with inclusion of
personal traits as additional features is the modest and efficient approach.
References
1. Bleha S., Slivinsky C., and Hussien B. Computer-access security systems using keystroke
dynamics. IEEE Transactions on Pattern Analysis and Machine Intelligence, 12(12):1217–1222, 1990.
2. Gaines R., Lisowski W., Press S., and Shapiro N. Authentication by keystroke timing: Some preliminary results. Technical Report Rand Rep. R-2560-NSF, RAND Corporation, Santa Monica, CA
90406, 1980.
3. Joyce R., and Gupta G. Identity authentication based on keystroke latencies. Communications of
the ACM, 33(2):168–176, 1990.
4. Killourhy K., Maxion R., A scientific understanding of keystroke dynamics. Carnegie Mellon
University, Pittsburgh, PA, 2012.
5. El-Abed M., Dafer M., El Khayat R. RHU Keystroke: A mobile-based benchmark for keystroke
dynamics systems. 48th IEEE International Carnahan Conference on Security Technology (ICCST),
Rome, Italy, 2014.
6. Modi S., and Elliott S.J. Keystroke dynamics verification using a spontaneously generated
password. In Proceedings of the 40th Annual IEEE International Carnahan Conference on Security
Technology (ICCST ’06), USA, October 2006, pp. 116–121.