Problem 7

searchcoilΛογισμικό & κατασκευή λογ/κού

15 Αυγ 2012 (πριν από 4 χρόνια και 10 μήνες)

266 εμφανίσεις

IT691

Computer Information Systems Projects

School of CSIS, Pace University




User Guide




Project:
Refinement of Mouse Movement Biometric
System
-

Third Iteration Fall 2007


Client:
Larry Immohr


Instructor: Dr. Tappert





by

Nkem Ajufor

Antony Amalraj

Rafael Diaz

Mohammed Islam

Michael Lampe


Due Date of Deliverable: 12/13/2007




Contents:


1.

Overview of the Mouse Biometric System

............................

3

1.1

System Requirements

................................
................................
........

4

2.

Integrated Development Environment User Guide

.............

5

2.1.

Installation

................................
................................
.........................

5

2.2

Start the NetBeans IDE

................................
................................
.....

8

2.3

Opening the project

................................
................................
...........

9

2.4

Launchi
ng the Data Capture module

................................
..............

13

3.

Data Capture module user guide

................................
.........

14

3.1

Software required to run the program:

................................
............

14

3.2

Install the program:

................................
................................
.........

16

3.3

Run the program:

................................
................................
............

16

3.4

Config.txt

................................
................................
........................

16

a.

Data.csv file

................................
................................
........................

17

b.

Structure of the program

................................
................................
.....

19

c.

Screenshots of the t
op level programs:
................................
...............

20

4.

Feature Extraction Module User Guide

..............................

23

4.1 Software required to run the program:

................................
.............

23

4.2 Install the program:

................................
................................
..........

23

4.3

Run the program:

................................
................................
............

23

4.4

FE.conf

................................
................................
............................

24

4.5

Measurement extraction:

................................
................................
.

25

4.6

Measurements:

................................
................................
................

27

4.7

Structure of the
Feature Extraction program:

................................
.

30

5.

Classifier Module User Guide

................................
..............

31

5.1

Software required to run the program:

................................
............

31

5.2

Install the program:

................................
................................
.........

31

5.3

Run the program:

................................
................................
............

32

5.4

Classifier.conf

................................
................................
.................

33

5.5

Classifier functions:

................................
................................
........

34

5.6

KNN Identification Method

................................
............................

36

5.7

Leav
e One Out Method

................................
................................
...

36

5.8

SuccessStatistics.csv

................................
................................
.......

39

6.

Future Web Interface

................................
............................

40


1.

Overview of the Mo
use Biometric System


The Mouse Movements Biometric S
ystem is a pattern recognition system having three
modules: a) Data Capture module b) Feature Extraction module and c) Classifier module.

Each of these three modules needs to be launched separately

in o
rder to capture, extract
and classify the mouse movements
.


Data Capture module:

The Data Capture module
ru
ns as a standalone application on
ones

computer.
It consists of two parts. A Top Level User Area and a Monitoring
Application. For the Top Level user

Area there are three options:
User Task Area, Button
Training Program and Tic Tac Toe. These are selected when creating the config file.
When

a
user start
s

using the mouse for accomplishing his/her tasks, a monitoring
program running in the background gat
hers the data. The data consists of x, y co
-
o
rdinates of the mouse movements as well as a

time stamp on the mo
vements. This data
is written

to
a
file.


Feature Extraction module:

The
Feature extraction module extracts a set of features
from the raw data.
This module has the following major functions:



Read the raw data file



Slice the data in to mouse curves and mouse clicks



Compute individual curve and click measurements



Create a Mouse profile with all curves and clicks and their measurements
(
Represents on
e user
)



Create Mouse profile measurements like average and standard deviation from the
curve and click measurements

The raw data from the data capture module needs to be placed in the feature extraction
module manually.


Classifier module:

classifies the f
eature vector and does the identification. Classifier
program has the following major functions:



Takes the feature vectors as the input



Normalize the feature vectors, so that all the data values are in a normalized scale.



Finds the K Nearest Neighbors for
a test case by computing the Euclidean
distances



Does Identification for an unknown test case



Does a leave one out method for a cross validation between many cases



Prints out the matching cases



Analyze the cross validation results and prints out the succes
s statistics


The feature vector from the feature extractor module needs to be manually placed in the
classifier.

The classifier module also contains most of feature extraction module so classifier
module can be used directly on raw data as an option.


For

the development of the program we used NetBeans as a development platform.

Each of the module can either be launched from the development platform or from its
executable jar file.


1.1

System Requirements


The Mouse Movements Biometric System has been develop
ed on the Microsoft
Windows Operating System in the Java language. Even though Java is a portable
language, the system has not been tested to run on a Linux/Unix system. At minimum,
the system requires to run on a Pentium II class system with 512MB or grea
ter RAM and
Windows XP or greater. It is highly recommended that a user running Microsoft Vista

have atleast 1GB of RAM.
In its current implementation, the Mouse Movements
Biometric System is run as a client application on the users desktop. Future
impleme
ntations will have the ability to run from a web application server and connect to
a database back end to store collected data.


2.

Integrated D
evelopment

Environment User G
uide


We have used NetBeans as our development platform for the java program.

You ca
n get more information the development platform from their web site:

http://www.netbeans.org/


2.1.


Installation


Download the latest version of the NetBeans software from the web site:

http://www.netbeans.info/downloads/index.php



Figure 1:
Once you have downloaded the installer file, double
-
click the installer's icon to
launch the installer.



Figure 2: At Welcome screen click next.


Figure 3:
In the i
nstall wizard, respond to the License Agreement, then click Next.



Figure 4:
Specify an empty directory within which to install NetBeans IDE.

Note: This NetBeans IDE will not disrupt the settings of your other NetBeans
installations because the IDE aut
omatically creates a new user directory when launched
(
${HOME}/.netbeans/5.5
).



Figure 5;
Choose the JDK you want the IDE to use from the list of suitable choices in the
list, then click Next.



Figure 6:
Verify that the installation location is corre
ct and that you have adequate space
on your system for the installation. Click Next to begin the installation.


Figure 7: Click Finish.









2.2


Start the NetBeans IDE




Double
-
click the NetBeans IDE icon on your desktop.



From the Start menu, select NetB
eans IDE 5.5

>

NetBeans IDE.


The NetBeans quick starter guide can accessed from this link:

http://www.netbeans.org/kb/55/quickstart.html








2.3


Opening the project

When the NetBeans started

it will show an empty welc
ome message as shown in
figure 8
.



Figure 8
: Launch the project.


The
n

go to the File
-
>Open menu to open the mmsystem project as shown in figure
9

and
10




Figure 9

Open project


You get an open dialog box and browse to the d
irectory where you mmsystem project
is located



Figure 10
: Select the mmsystem project


Once you open the project the

screen will look as in figure 11




Figure 11
: The project is open.





Once the project is open you can edit the java classes. After

the edit, to build the projects
go to Build and select Build m
ain project as shown in figure 12



Figure 12
: Build project


From the Build menu you can build the java documentation also. To run the project we
need to input the configuration file as
an
ar
gument. In the
MMSYSTEM

the argument
file name is Config.txt. Select the project and right click, in the popup window select
properties. The IDE opens a

dialog box as shown in figure 13
. In this dailog box select
run. On the right side panel, against
Argum
ents
give the configuration file name
Config.txt




Figure 13
: Give the arguments


Now you are ready to run the project. Go to the run menu and select Run main project.




Figure 14
: Run the project


2.4


Launching the Data Capture module

The data capture m
odule will launch with a dialog box to enter your name

as well as
other characteristics
:



Figure 15
: Enter your name as input
to
the program


Then the program will launch one of the user interface as per the co
nfiguration as shown
in figure 16
:



Figure

16
: Button training program


The details of configuration file, output data file are described in section 3 of this
user guide. Please refer section 3 of user guide for more information on the Data
Capture module.










3.

Data Capture module user guide


All modules can be launched from the NetBeans IDE or from the executable jar file.

The section will describe how the program is launched from the jar file.


3.1


Software required to run the program:

In order to run the program,
the
latest java runtime environ
ment is required.
The l
atest
java runtime is version
6
.0 update
3
. The link to download the java runtime is:

http://www.java.com/en/download/manual.jsp



Figure 17: Java download page. Select Off
line version for operating System you have.



Figure 18: accept License Agreement.



Figure 19: Successful installation. Click finish.


3.2


Install the program:

Unzip the mmsystem.zip file to your C drive. In the resulted “mmsystem” folder, there
are
two

fi
les
in order to run the program
a) mmsystem.jar
and b) Config.txt

Config.txt is a required file to run the program.


Figure 20: After program is unzipped to local directory.

3.3


Run the program:

Double click the “mmsystem.jar” file

Or

Run the MS
-
DOS prompt
and point to the “mmsystem” directory and type “java
-
jar
mmsystem.jar” and enter as shown below:

C:
\
mmsystem>java
-
jar mmsystem.jar

The program will be launched.
At the starting of the program, a dialog box will appear
and ask for your name

and other char
acteristics.
The data file written will be appended
with your name.

For example: if I give my name “
John
1
” in the dialog box then there will be a data file


John
1
_data.csv”


Figure 21:
Data Capture opening screen.

3.4


Config.txt

The “Config.txt” is
used
to c
onfigure the program for certain parameters. This is a text
file having the following format. The values are separated by coma (,). There are
two
lines

for the configuration file.

Line1:

Timer, Program

Choice,
Number of buttons, N
umber of buttons to be en
abled

Line 2:
Pattern for the buttons

(the pattern corresponds to the number of buttons to be
enable
d
)

Timer: is an integer value
which
user can input to control the frequency of the data
gathering. The unit

of frequency is taken in milli
seconds. Some exam
ple values are: 25
,

50 or 100

ProgramChoice: This allows the user to select the top level program to run. Currently
three top level programs are supported

1
-

Blank Screen

2
-

Button Training Program

3
-

TicTacToe Game.


An example
Config.txt
file:

50,2,25,
10

1,2,3,4,5,6,7,8,9,10


This will give
the
timer as 50 ms and launch the button training program with 25 buttons.
Out
of these

25 buttons
,

only 10 buttons will be enable
d
. The 10 buttons will
enable

the
click in
the
following order given in the second lin
e.

Note: There are not rigorous checks for validating the patte
rn. I
f the pattern is not valid it
will give an error message and
the program will
stop
.

A good pattern should contain numbers starting from 1 to the number of enabled buttons.
The numbers can
be in any order.
The numbers cannot repeat. All the numbers (between
1
-
number of buttons) should be given in the pattern
.

a.


Data.csv file

The data gathered during the mouse movement

Data Capture

is printed in
the

Name_
Data.csv” file
s
.

Make sure you enter

your name differently each time you run the program otherwise it
will overwrite the current Name_Data.csv file.
See figure 22 for an example.


Figure 21: Multiple instances of Name_Data.csv file by appending a numeral to name.




The data in this file is

printed in the following format:


Figure 22: Name_Data.csv file


First
line

in the
spreadsheet lists the name and characteristics
:


John1/M/44

Right
-
handed

Serial Mouse

Fixed 25 button sequence/used right hand


Third and fourth line lists the width and h
eight of the user’s screen size and timer


User
screen
Size

width

1680

height

1050

timer

10





Rest of the lines in the program

contain the mouse movements
:


<mouseAction><TimeMilliSec><x
-
cord><y
-
cord>

Example for mouse move data:

mouseMoved

11616361734
53

541

201

mouseMoved

1161636173562

174

190

mouseMoved

1161636173718

165

193

mouseMoved

1161636173828

59

12


The time difference between first entry and the second entry is (562
-
453)=109 ms

Example for mouse clicks data:

mousePressed :left click 1
time
(s)

1161636174062

59

11

mouseReleased

1161636174296

59

11


A click is a combination two actions mouse press and mouse release.

Time require to complete the click = press time


release time= 4296
-
4062=230 ms.









b.


Structure of the program



Figure 23: Structure of the
MMSYSTEM

program.



The program is divided in to many modules to make it a modular system. The
program is divided in
to

three groups
:

o

Top level user area


three programs (UserTaskArea.java,
ButtonTrainingProgr
am.java, TicTacToeAI.java) The
tic
-
tac
-
toe

code was
downloaded from the i
nternet but modified it to suit

our application.

o

Monitoring program
-

one program (MouseMonitoringComponent.java).
This program monitors and gathers data of the top level program

o

Tim
eStamp.java
-

to make time stamp on the data

o

Application wrap up
-

(MouseMovementSystemApp. Java) which makes it
an application.


UserTaskArea.java

Mouse
MonitoringComponet.java

ButtonTrainingProgram.java

TicTacToeAI.java

TimeStamp.Java

MouseMovement
SystemApp.java

c.


Screenshots of the top level programs:



Figure 24
: Enter your name



Figure 25: Select your gender



Figure 26: select your

age.



Figure 27: Select hand used.



Figure 28: S
elect type of mouse.



Figure 29: Select Test Screen Type.



Figure 2
9
: Blank Screen



Figure 3
0
: Button Training Program


Figure 31:

TicTacToe Game

4.

Feature Extraction Module User G
uide


All modules

can be launched from the NetBeans IDE or from the executable jar file.

The section will describe how the program is launched from the jar file.


4.1 Software required to run the program:

In order to run the program,
the
latest java runtime environment
is required. Latest java
runtime is version
6
.0 update
3
. The link to download the java runtime is:

http://www.java.com/en/download/manual.jsp

4.2 Install the program:

Unzip the featureextracti
on_IDE2.zip file
to your C drive. In the resulting

“featureextraction_IDE2” folder, there are a) FeatureExtraction.jar b) “FE.conf” file c)
Data folder d) FeatureOutput folder


4.3

Run the program:

Bring up a

MS
-
DOS prompt and point to the “featureextract
ion_IDE2” directory and type
“java
-
jar featureextraction.jar
FE.conf
” and press enter key.

FE.conf

is the configuration file for the feature extraction module.

A screenshot of the program running is given below:



Figure 32
: Screenshot of running the p
rogram


4.4

FE.conf


FE.conf is
the configuration file for the Feature E
xtraction module.

The configuration
requirements for the feature extraction module are stored in this file. This file is r
ead in
when the program starts and
is essential to run the pro
gram. The file has one line and the
parameters are separated by com
m
a (“,). The line should be written in the following
format:



Input Folder Name, O
utput Folder Name, Input Load Choice, Input File Name


Details of each parameter in this f
ile are given below:

Input Folder Name:

This is the folder in which the raw data files are stored.

The profileDB.txt (described in the next section) files also should be stored in this folder.
The suggested name of the folder is “Data”

O
utput Folder Name:

This the folder in which output files of feature extraction module is
stored. The suggested name of the folder is “FeatureOutput”

Note:

The ab
ove folder names can be changed. M
ake sure that there are folder
s

with the new
names in the director
y

of the prog
ram. If there
are

no folder
s

to read or write the data,
then the program will give an error.

Input Load Choice:

The Feature E
xtraction module allows loading one or more files. This
parameter takes two values
:

1 or 2. The value 1 indicates there is only one

file to load.
The value 2 indicates there are many files to load.

Input File Name:
This parameter gives the file name
to load to the program. If the Input
Load C
hoice value is equal to 1 then the file name will be the actual file name to load.

If the Inpu
t Load C
hoice is equal
to 2,
then the file name is “profileDB.txt”. The file
“profileDB.txt” contains all the file names to load.



Figure 33: FE.conf





Figure 34: ProfileDB.txt



4.5

Measurement extraction:

The Feature E
xtraction module allows taking
one or more raw data files and extracting
the
features of each file. The program reads the names of the raw data from the file called
“profileDB.txt”.
There are two ways to create a user mouse profile from the raw data file,
this is depending upon how line
s in “profileDB.txt” is given
.



a.

One file creates one user profile, if user has multiple files in the
“profileDB.txt” then multiple user profiles can be created with the same user
name.

b.

Multiple files can create one user profile with one user name.


The “p
rofileDB.txt” is written in the following format

Line1: usernameA,fileA1,fileA2fileA3

Line2: usernameB, fileB1,fileB2

----------
etc.


An example “profileDB.txt” is given below:


Figure 35: One way to configure ProfileDB.txt


In the above case each of the
usernames will have five user profiles.


Another example:


Figure 36: Another way to configure ProfileDB.txt.


In this case each of the usernames will have only one profile.


A mix of the above two cases is also possible.


After successful execution of th
e program, it outputs a measurement file for each line in
the “profileDB.txt” in the following name convention: “
yourdatafile_Measure.csv

“. For
example if “ XY_data.csv” is your data file name, then program will output
“XY_data_Measure.csv”. The file can
be opened in an excel spreadsheet. Details of the
single file measurements are given in section 6.




After the successful execution it also outputs a “FeatureVector.csv” file. This files gives
the feature vector created for each user. The feature vector c
onsists of average and
standard deviations for each of the individual curve and click measures. A sample feature
vector file is shown below. The column headings are added for clarity.






Figure 37
: A sample Feature Vector file, the column headings are a
dded for clarity.






4.6

Measurements:

This section provides the details for individual feature extraction from a file.

The feature extraction module has the following major functions:

i.

Read the raw data file

ii.

Slice the data in to mouse curves and mouse cl
icks

iii.

Compute individual curve and click measurements

iv.

Create a Mouse profile with all curves and clicks and their
measurements (
Represents one user
)

v.

Create Mouse profile measurements like average and standard
deviation from the curve and click measurements


Currently the following measurements are included:

For each curve:

a) Number of mouse points in the curve (curve sample size)

b) Length of the curve (pixels)

c) Total time to complete the curve (millisecond)

d) Average speed of the curve (pixels/milliseco
nd)

For each click:


a) Click duration

For each profile:


a) Average curve time of all the curve times


b) Standard deviation of all the curve times


c) Average speed of all the curve average speeds


d) Standard deviation of all the curve average speeds


e) Average click duration of all click durations


f) Standard deviation of all click duration

A sample measurement file for one data file with
25

buttons

is shown below:





Figure 38: A sample Name_Data_Measure file, the column headings are added for
clarity.


4.7

Structure of the Feature E
xtraction program:

The Feature E
xtraction module will follow the following hierarchy. Each “data.csv” file
represents a mouse profile in the software. Each profile contains vectors of two objects,
Mouse curves and M
ouse clicks. A mouse curve and mouse click contains the data as
well as the measures. The profile object contains profile measures as well. Profile
measures contain measures between all the curves and clicks in the profile. For example:
standard deviation
of average velocities of a profile.






Figure 39: Structure of the Feature Extraction program.


MouseProfile

Is a vector of curves
and clicks?


MouseCurves

This is a vector of
curves. A profile
contains many curves


MouseClicks

This is a vector of
cli
cks. A profile
contains many clicks


MousePoints

Each curve is a
vector of
mouse points.


CurveMeasure

Each curve has
one Measure
object


MouseData

Each point is
represented
with
MouseData:



Action



Time



x



y


Each measure
objec
t can have
many measures
in it.



Speed



Angle




MousePoints

Each click is
having
two

mouse points.


ClickMeasure

Each click has
one Measure
object


MouseData

Each point is
represented
with
MouseData:



Action



Time



x



y


Each measure
object can have
many measures
in it.



duration


ProfileMeasure

measure of many
curves and clicks



5.


Classifier
Module User G
uide


5.1

Software required to run the program:

In order to run the program, latest java runtime environ
ment is required. Latest java
runtime is version
6
.0 update
3
. The link to download the java runtime is:

http://www.java.com/en/download/manual.jsp

5.2

Install the program:



Unzip the classifier_
IDE1.zip file in to your C: drive. The resultant folder is
Classifier_IDE1. In the folder the following files are included: a) Classifier.jar b)
Classifier.conf c) Data folder and d) ClassifierOutput


Figure 40: Contents of Classifier_IDE1 directory


The
Data folder contains files that
input

to the classifier program

The Data folder contains a) profileDB.txt b) test_data.csv c) FeatureVectors.csv d) Many
data files


The ClassifierOutput
directory
contains files that are
output

from the classifier program.

The ClassifierOutput contains the following files: a) FeatureVectors.csv b)
NormalizedFeatureVectors.csv c) ClassifierOut.csv and d) SuccessStatistics.csv




5.3

Run the program:

Run the MS
-
DOS prompt and point to the Classifier_IDE1 and type “java

jar

Classifier.jar Classifier.conf” and press enter key.

i.e.
C:
\
IT691Project
\
Classfier_IDE1> java

jar Classifier.jar Classifier.conf

“Classifier.conf” is the configuration file for the classifier program.

A screenshot of the program running is given below:


Figure
41: Screenshot of the C
lassifier program running

with KNN Identification


5.4

Classifier.conf


Classifier.conf is the configuration file for the classifier module. This file is read in when
the program starts. This file is essential to run the pr
ogram. The file has one line and
parameters are separated by com
m
a (“,”). The line should be written in the following
format:



Input Method, Classifier Method, test data file



Figure 42:
Classifier.conf configuration file.


Details of each parameter in
this file are given below:

Input Method:

Classifier program has the facility to accept input in two different
methods. It can take either one of the two values 1 or 2.

1: Directly from RAW data files. In this case the program reads the profileDB.txt fil
e
and extracting the features before classifying the training set.

or

2: From the feature vectors extracted by the Feature Extraction program. In this case the
program reads “FeatureVectors.csv”.



Classifier Method:
Classifier program can do classificat
ion in two different ways, either
KNN identification or Leave one out method. This parameter can take either one of the
two values 1 or 2.

1: KNN Identification

2: Leave One out Method


Test data file:
This is the unknown data file to be used in the KNN Id
entification.

Some example configurations:

If one wants to give raw data file input and KNN identification, the configuration file is:

1, 1, test_data.csv

If one wants to give raw data file input and Leave one out method, the configuration file
is:

1, 2,
test_data.csv




5.5

Classifier functions:

The
Classifier program has the following major functions:


a.

Take

the feature vectors as the input

b.

Normalize the feature vectors, so that all the data values are in a normalized
scale.

c.

Find

the K Nearest Neighbors
for a test case by computing the Euclidean
distances

d.

Does an Identification for an unknown test case

e.

Does a leave one out method for a cross validation between many cases

f.

Prints out the matching cases

g.

Analyze the cross validation results and prints out the

success statistics


The structure of the classif
ier program is given in figure 43
. The program takes input in
two different ways. Either directly from the raw data file or from the feature vectors.
When the input is from the raw data files, the program ex
tracts the features and creates
the feature vectors.



Figure 43
: Structure of the classifier program







Takes
any one
the input
and
outputs Feature
Vector

FeatureVectors.csv

File

Feature Extraction

Raw Data File

Feature Vectors

Normalized Feature Vectors

Classifier
can
output
any one

K Nearest Neighbor

KNN
Identification

Leave One out
Method

Test Case

5.6

KNN Identification Method

In the KNN Identification method
,

an unknown user is tested against
a
known set of
users.

The nearest neighbors are listed by computing the Euclidean distance between the
unknown case and each of entries in the known database of users. A sample output report
for the KNN Ide
ntification is given in figure 44

below:

In this case
, a copy of one of

the files of the user “
Mohammed
” is used as the test case
.



Figure 44
: Sample KNN report



5.7

Leave One
Out
Method

The Leave O
ne
Out
method is used for cross validation with all the data files in the
database. In this method
,

one file is selected as t
he test case and tested against
the
rest of
the files in the training set. For example, let
’s

say we have three users A,B,C and each of
the
m

have three files each A={A1,A2,A3} B={B1,B2,B3}, C= {C1,C2,C3}.

Create a training set with all these files

Trainin
g Set =

{A1, A2, A3, B1, B2, B3, C1, C2, C3}

Then take one and compare it with others.

A1
---
Compare with

{A2,A3,B1,B2,B3, C1,C2,C3}

If there is a match, A2/A3 will be first and second shortest distances

Next, take out another one

A2
---
Compare with

{A1,
A3,B1,B2,B3, C1,C2,C3}

If there is a match then A1/A3 will be the closest.

Continue
this program for all the cases.


This approach will produce cross validating results with each other. From the statistics of
such results
,

we will be able to derive good co
nclusions.

A sample output file of the leave one out method is given in figure 4
5

below. In this case
we used 5 data files from each user. When
the first file of the user “
Rafael
” compared
with all
the other files,
four

files of the same user came in the f
irst
four

places.



Figure 4
5
: A

sample Leave One O
ut method report.


The case shown above has one file matched to all other four files of the same user in the
first
four

positions of the Nearest Neighbors. However this most successful case may not
happen

to all the users. Another test case in the same leave out one cross valid
ation is
given below in figure 46
. Note that both of these output samples are taken from the same
“ClassifierOutput.csv” file.


In this case, testing one file of the user “
Mohammed

with all other files in the database.
The first two positions have files from same user. However third

through seventh

positions are having files of other users. This result is having 100% success in the first
two cases.



Figure 46
: Another sample in Le
ave out One method


5.8

SuccessStatistics.csv

After the leave one out classification, the program does an analysis on the success on
different cases. The report on such analysis is written in SuccessStatistics.csv file. A
sample file is given in figure 47
.

In this study there are 25

data files, 5 files each from
5

users.

When each of these 25

files is tested, the first choice was another file from the same user.

So there was
72

percent success in the closest neighbor case.

When each of these 25

files is tes
ted, the second choice was another file from the same
user. So there was
72

percent success in the second closest neighbor case.

Matching

first and second nearest neighbors together makes
44

percent.

When each of these 25

files is tested, the third choice
was another fil
e from the same user
only for 11 files out of 25

files. So there is only
44

percent success in the third closest
neighbor case.



Figure 47
: A sample SuccessStatistics.csv file


6.

Future Web Interface



The screenshot above is of the planne
d new Java mouse Movement Systems user
registration interface which will collect more user characteristics. This will run locally on
the client’s machine.



This screenshot shows a prototype of a future web interface for the Mouse Movement
Systems Biometr
ic Identification program. This is accessible for viewing at
http://utopia.csis.pace.edu/cs691/2007
-
2008/team1/register.html