You Are Who You Know: User Authentication by Face Recognition

brasscoffeeAI and Robotics

Nov 17, 2013 (3 years and 6 months ago)

69 views

You Are Who You Know: User
Authentication by Face Recognition

M
artin

Klíma
1
, A
dam
J Sporka
1,2
, J
akub

Franc
3


1
CTU Prague, Czech Republic

2
University of Trento, Italy

3
Sun Microsystems, Inc., Czech Republic

Kl
íma, Sporka, Franc: User Authentication by Face Recognition

ICDVRAT 2008,
07

11 September, Maia, Portugal

2

Outline


Motivation: Why authentication by pictures


Use case


Demo


User study


Results

Kl
íma, Sporka, Franc: User Authentication by Face Recognition

ICDVRAT 2008,
07

11 September, Maia, Portugal

3

Introduction


Pervasion of the computing equipment




Computerization of household appliances


UI not being adapted for the needs of the
elderly population


Increasing demand on skills such as IT,
perception, etc.


Addressed by i2home project

http://www.i2home.org


Kl
íma, Sporka, Franc: User Authentication by Face Recognition

ICDVRAT 2008,
07

11 September, Maia, Portugal

4

Target Group and Use Case


Our focus:


Elderly people (65+ of age)


Use case:


Accessing own room in a senior house


Problem:


Authentication of the user

Kl
íma, Sporka, Franc: User Authentication by Face Recognition

ICDVRAT 2008,
07

11 September, Maia, Portugal

5

Problem We Address


Still, a lack of privacy in senior houses


“Houseless vs. homeless”


Territories (Gifford 1997)


Primary


Secondary


Public


Distinguishing territories


Principal requirement of the inhabitants


Solution


Physical separation of territories

Kl
íma, Sporka, Franc: User Authentication by Face Recognition

ICDVRAT 2008,
07

11 September, Maia, Portugal

6

Authentication


Tokens


Keys, etc.


Authentication by passwords


PIN code (4 digits)


Memory
-
challenging problem:


Requirements of
active

memory


Insecure (easy
-
to
-
guess passwords)


Biometrics


Kl
íma, Sporka, Franc: User Authentication by Face Recognition

ICDVRAT 2008,
07

11 September, Maia, Portugal

7

Face Recognition


Remarkable human skills in recognizing
each other, regardless of …


Angle


Distance


Light


“Wired” into our brain


Very important for common social
transactions

Kl
íma, Sporka, Franc: User Authentication by Face Recognition

ICDVRAT 2008,
07

11 September, Maia, Portugal

8

Face Recognition

Developmental psychology perspective:


In neonates (Pascalis et al, 1995)


Acquired very early


In senior population


Preserved in old age


Relatively resistant to senile dementia


Knowledge of faces learnt in preceding stages

Kl
íma, Sporka, Franc: User Authentication by Face Recognition

ICDVRAT 2008,
07

11 September, Maia, Portugal

9

Previous Work


Weidenbeck et al (2005)


Clicking on pre
-
defined parts of the picture


Tullis et al (2005), Weinshall et al (2004)


Pictures of faces assigned to the user account


Commercial solution:
PassFaces



Common traits:


Involve learning


Outcome can be affected by the senile dementia

Kl
íma, Sporka, Franc: User Authentication by Face Recognition

ICDVRAT 2008,
07

11 September, Maia, Portugal

10

Our Method


Selection of
unknown

face in a 3x3 matrix


Three rounds



Known pictures:


Taken from a family album of the user


Unknown pictures:


Randomly selected


Characteristics matching to the known pictures

Kl
íma, Sporka, Franc: User Authentication by Face Recognition

ICDVRAT 2008,
07

11 September, Maia, Portugal

11

Our Method


<demo>

Kl
íma, Sporka, Franc: User Authentication by Face Recognition

ICDVRAT 2008,
07

11 September, Maia, Portugal

12

User Study


Pilot study


Comparison of 2 methods


Pictures


PIN (realistically generated; not easy
-
to
-
guess)


Within
-
subject design

Kl
íma, Sporka, Franc: User Authentication by Face Recognition

ICDVRAT 2008,
07

11 September, Maia, Portugal

13

Participants


9 participants


74 years old (SD 4.2)


Residents of a senior house near Prague

Kl
íma, Sporka, Franc: User Authentication by Face Recognition

ICDVRAT 2008,
07

11 September, Maia, Portugal

14

Procedure


Session 1:


Upload of pictures


Selection of PIN


Run of both methods


Session 2:


14 days after


Run of both methods


Kl
íma, Sporka, Franc: User Authentication by Face Recognition

ICDVRAT 2008,
07

11 September, Maia, Portugal

15

Results


Authentication by pictures


People more successful, confident, and pleased


No failures during both sessions


Method reported as “more satisfying”,
“friendlier”, and “easier
-
to
-
use” than PIN

Kl
íma, Sporka, Franc: User Authentication by Face Recognition

ICDVRAT 2008,
07

11 September, Maia, Portugal

16

Results


PIN authentication


People more confused and annoyed


Higher cognitive load observed and reported


Some people were “practicing” the PIN between the
sessions


3 failures (out of 9) during Session 2


“struggling around” the correct one


Early Alzheimer participants failed in this task

Kl
íma, Sporka, Franc: User Authentication by Face Recognition

ICDVRAT 2008,
07

11 September, Maia, Portugal

17

Conclusion


Authentication by pictures more successful
than traditional PIN authentication.


Method may be suitable for the Alzheimer
disease patients


Will be studied in a separate quantitative study


Kl
íma, Sporka, Franc: User Authentication by Face Recognition

ICDVRAT 2008,
07

11 September, Maia, Portugal

18

Acknowledgements


Conducted within i2home project,

FP6
-
033502,
http://www.i2home.org
.


Adam Sporka’s research at the University of
Trento is supported by the EC Marie Currie
Excellence Grant for the ADAMACH project
(contract. No. 022593).


The authors would like to express their
thanks to Sri Hastuti Kurniawan for her
comments.