DDD: A New Ensemble
Approach for Dealing with Concept Drift
ABSTRACT:
Online learning algorithms often have to
operate in the presence of concept drifts. A recent study revealed that
different diversity levels in an ensemble of learning machines are required in
order to maintain high generalization on both old and new
concepts. Inspired by this study and
based on a further study of diversity with different strategies to deal with
drifts, we propose a new online ensemble learning approach called Diversity for
Dealing with Drifts (DDD). DDD maintains ensembles with different diversity levels
and is able to attain better accuracy than other approaches. Furthermore, it is
very robust, outperforming other drift handling approaches in terms of accuracy
when there are false positive drift detections. In all the experimental
comparisons we have carried out, DDD always performed at least as well as other
drift handling approaches under various conditions, with very few exceptions.
EXISTING SYSTEM:
·
Online learning
algorithms often have to operate in the presence of concept drifts.
·
The real-world
data sets, it is not possible to know exactly when a drift starts to occur,
which type of drift is present, or even if there really is a drift.
·
The existing
ensemble approaches yet, as they do not encourage different levels of diversity
in different situations.
PROPOSED SYSTEM:
·
DDD maintains
ensembles with different diversity levels and is able to attain better accuracy
than other approaches.
·
It is very
robust, outperforming other drift handling approaches in terms of accuracy when
there are false positive drift detections.
·
DDD always
performed at least as well as other drift handling approaches under various
conditions, with very few exceptions.
MODULES:
A New Ensemble Approach for Dealing
with Concept Drift consists of the following modules.
Ø Data
Conversion
Ø Protocol
View
Ø Prequential
Accuracy
Ø DDD
Description
MODULES
DESCRIPTION:
Data Conversion:
§ Data Conversion is the process of converting
the dataset of KDD99Cup to our database using specialized splitting process.
§ The Dataset is entirely converted as fields in
our database in order to freely access the information regarding the database
KDD99Cup.
§ Where as the Database contains information on
required parameters. Such as:
• Duration
• Protocol Type
• Service
• Flag
• Source Bytes
• Destination Bytes,
• Land,
• Logged in
• Etc,.
Protocol
View:
The protocol view module can be used
to view the protocol and its total records from the KDD99Cup.This dataset already
extracted and stored in the database. Here 3 protocols are available. Such as,
·
TCP
·
UDP
·
IMCP
Prequential
Accuracy:
The prequential
accuray module consists of the following sub modules.Such as,
·
Online Bagging
·
R2L
·
U2R
·
Probe
Online
Bagging is a well-known ensemble learning method which is used to
improving generalization performance.R2L is an unautherized access from a
remote machine. U2R is an unautherized access from a local super user.Probe is
used to learning something about the state of the network.
DDD
Description:
DDD
operates in two modes: prior to drift detection and after drift detection. We
chose to use a drift detection method, instead of treating drifts implicitly,
because it allows immediate treatment of drifts once they are detected. So, if
the parameters of the drift detection method are tuned to detect drifts the
earliest possible and the approach is designed to be robust to false alarms, we
can obtain fast adaptation to new concepts.
HARDWARE
REQUIREMENTS
•
SYSTEM : Pentium IV 2.4 GHz
•
HARD
DISK : 40 GB
•
FLOPPY
DRIVE : 1.44 MB
•
MONITOR : 15 VGA colour
•
MOUSE : Logitech.
•
RAM : 256 MB
•
KEYBOARD :
110 keys enhanced.
SOFTWARE
REQUIREMENTS
•
Operating system :- Windows XP
Professional
•
Front End :- Microsoft Visual Studio .Net 2008
•
Coding Language : - C# .NET.
•
Database :-
SQL Server 2005
REFERENCE:
Leandro L. Minku, Member, IEEE, and Xin
Yao, Fellow, IEEE, “DDD: A New Ensemble Approach for Dealing with Concept
Drift”, IEEE TRANSACTIONS ON KNOWLEDGE
AND DATA ENGINEERING, VOL. 24, NO. 4, APRIL 2012.