Difference between revisions of "AFSecurity Seminar"

From mn/ifi/AFSecurity
Jump to: navigation, search
Line 7: Line 7:
 
'''AGENDA:'''<br />
 
'''AGENDA:'''<br />
 
14:00h Welcom at UiO<br /><br />14:15h Invited Talk:
 
14:00h Welcom at UiO<br /><br />14:15h Invited Talk:
| [[File:logo-ENSICAEN.jpg|300px|link=https://ccdcoe.org/]]
+
| [[File:logo-ENSICAEN.png|300px|link=https://ccdcoe.org/]]
 
|}
 
|}
 
* '''TALK:''' &nbsp;''Privacy Threat of Keystroke Profiling on the Web''<br />'''SPEAKER:''' ''Denis Migdal'' (ENSICAEN) &nbsp; <br />'''ABSTRACT:''' In a browser environment, Keystroke Dynamics (the way of typing on a keyboard) is an interesting biometric modality as it requires neither additional sensors (just your keyboard), nor additional actions from the user.  Keystroke Dynamics can easily be collected through a Web page to authenticate,  identify,  or profile visitors...  even without their knowledge and consent. Contrary  to  learning-based  Keystroke  Dynamics  (s.a.  based  on  Deep  learning  or SVM), distance-based Keystroke Dynamics can be used with very few data.  However, it generally provides deceiving authentication and identification performances.  In a first part, we will see how to improve such performances, and how Keystroke Dynamics can pose serious threats to users privacy, even with only few information. Fortunately, thanks to Keystroke Dynamics Anonymization Systems, it is possible to protect our Keystroke Dynamics, or at last to disturb identification and profiling systems. Several  Keystroke  Dynamics  Anonymization  Systems  will  be  presented  in  the  second part,  as  well  as  some  recommendations  to  build  and  implement  Keystroke  Dynamics Anonymization Systems. In a third part, I will present a multi-modal privacy-compliant authentication based, among other, on Keystroke Dynamics, as well as some schemes and uses cases.  Proof of authorship in an online collaborative document writing, or proof of identity on a Social Network constitute application of our proposed authentication. And if we still have time, the fourth part will be dedicated to synthetic generation of Keystroke Dynamics.  Usurpation of Keystroke Dynamics, Keystroke Dynamics dataset creation or augmentation, and better understanding of Keystroke Dynamics are goals of Keystroke Dynamics synthetic generation
 
* '''TALK:''' &nbsp;''Privacy Threat of Keystroke Profiling on the Web''<br />'''SPEAKER:''' ''Denis Migdal'' (ENSICAEN) &nbsp; <br />'''ABSTRACT:''' In a browser environment, Keystroke Dynamics (the way of typing on a keyboard) is an interesting biometric modality as it requires neither additional sensors (just your keyboard), nor additional actions from the user.  Keystroke Dynamics can easily be collected through a Web page to authenticate,  identify,  or profile visitors...  even without their knowledge and consent. Contrary  to  learning-based  Keystroke  Dynamics  (s.a.  based  on  Deep  learning  or SVM), distance-based Keystroke Dynamics can be used with very few data.  However, it generally provides deceiving authentication and identification performances.  In a first part, we will see how to improve such performances, and how Keystroke Dynamics can pose serious threats to users privacy, even with only few information. Fortunately, thanks to Keystroke Dynamics Anonymization Systems, it is possible to protect our Keystroke Dynamics, or at last to disturb identification and profiling systems. Several  Keystroke  Dynamics  Anonymization  Systems  will  be  presented  in  the  second part,  as  well  as  some  recommendations  to  build  and  implement  Keystroke  Dynamics Anonymization Systems. In a third part, I will present a multi-modal privacy-compliant authentication based, among other, on Keystroke Dynamics, as well as some schemes and uses cases.  Proof of authorship in an online collaborative document writing, or proof of identity on a Social Network constitute application of our proposed authentication. And if we still have time, the fourth part will be dedicated to synthetic generation of Keystroke Dynamics.  Usurpation of Keystroke Dynamics, Keystroke Dynamics dataset creation or augmentation, and better understanding of Keystroke Dynamics are goals of Keystroke Dynamics synthetic generation

Revision as of 07:31, 16 August 2019

Red Teaming in Cyber Exercises

DATE:  30 August 2019

PLACE:  Kristan Nygaards Hall (Room 5370), IFI, UiO - OJD House .

AGENDA:
14:00h Welcom at UiO

14:15h Invited Talk:

Logo-ENSICAEN.png
  • TALK:  Privacy Threat of Keystroke Profiling on the Web
    SPEAKER: Denis Migdal (ENSICAEN)  
    ABSTRACT: In a browser environment, Keystroke Dynamics (the way of typing on a keyboard) is an interesting biometric modality as it requires neither additional sensors (just your keyboard), nor additional actions from the user. Keystroke Dynamics can easily be collected through a Web page to authenticate, identify, or profile visitors... even without their knowledge and consent. Contrary to learning-based Keystroke Dynamics (s.a. based on Deep learning or SVM), distance-based Keystroke Dynamics can be used with very few data. However, it generally provides deceiving authentication and identification performances. In a first part, we will see how to improve such performances, and how Keystroke Dynamics can pose serious threats to users privacy, even with only few information. Fortunately, thanks to Keystroke Dynamics Anonymization Systems, it is possible to protect our Keystroke Dynamics, or at last to disturb identification and profiling systems. Several Keystroke Dynamics Anonymization Systems will be presented in the second part, as well as some recommendations to build and implement Keystroke Dynamics Anonymization Systems. In a third part, I will present a multi-modal privacy-compliant authentication based, among other, on Keystroke Dynamics, as well as some schemes and uses cases. Proof of authorship in an online collaborative document writing, or proof of identity on a Social Network constitute application of our proposed authentication. And if we still have time, the fourth part will be dedicated to synthetic generation of Keystroke Dynamics. Usurpation of Keystroke Dynamics, Keystroke Dynamics dataset creation or augmentation, and better understanding of Keystroke Dynamics are goals of Keystroke Dynamics synthetic generation

16:00h Discussion


SPEAKER BIO
Denis Migdal is PhD student at ENSICAEN in Caen, France. His PhD research project focuses on privacy protection against user profiling which can exploit biometric keystroke dynamics of normal user activity on the Web.

AFSecurity-small.png AFSecurity is organised by the UiO Research Group on Information & Cyber Security Sec-uio-light-1000.png