Vous êtes sur la page 1sur 19

AN IMPLEMENTATION OF INTERRUPTION DETECTION

SYSTEM USING HYBRID FEATURE SELECTION


TECHNIQUE WITH KNN AND SVM

Submitted To:- Submitted By:-


Er.
Introduction
 Data Mining :- data mining is a task of retrieve information form large dataset.
we use different type of classification and clustering approach for this task, it
convert data into conceptual level.
 Two type of feature reduction approaches are used for reduction :-
i)feature selection ii) feature extraction

2
Feature selection is an essential role in improving the eminence of learning algorithms in
data mining and machine. In this approch select a subset of the feature form all feature of
data.

Feature Extraction/Reduction(Transform) is highly crucial mathematical task to reach


conceptual level by raw data.

 Concluding that reducing the dimension of data set has following


advantages:

I. It reduces the storage, time and space required.


II. Elimination of multi-co linearity improves the feat of the machine
learning algorithm.
III. It becomes easier to think about the data when reduced to low
dimensions such as 2D or 3D.
IV. computational burden of the learning algorithms can be minimized
3
Introduction of Interruption Detection
During the last few years there is a dramatic increase in growth of computer
networks. There are various private as well as government organizations that
store valuable data over the network.
This tremendous growth has posed challenging issues in network and
information security, and detection of security threats, commonly referred to as
intrusion, has become a very important and critical issue in network, data and
information security.
The security attacks can cause severe disruption to data and networks.
Therefore, Interruption Detection System (IDS) becomes an important part of
every computer or network system. Interruption detection (ID) is a mechanism
that provides security for both computers and networks.
Feature selection and feature reduction is important area of research in intrusion
direction system. The size and attribute of intrusion file are very large.
Due to large size of attribute the detection and classification mechanism of
intrusion detection technique are compromised in terms of detection rate and
alarm generation.
For the improvement of intrusion detection process various authors and
researchers work together for feature reduction and feature selection for
4 intrusion detection system.
Literature Survey
Aditya Chellam et.al.(2018) " Interruption Detection in Computer Networks
using Lazy Learning Algorithm" proposed Interruption Detection Systems (IDS) are
utilized in PC systems to protect the respectability and classification of touchy information. This
paper, proposes the utilization of lethargic learning philosophies to enhance by and large
execution of IDS.
Omar Y. Al-Jarrah et.al. (2018) "Semi-supervised multi-layered clustering model
for interruption detection" considered A Machine Learning (ML)- based Interruption
Detection and Prevention System (IDPS) requires a lot of marked up and coming preparing
information to successfully identify interruptions and sum up well to novel assaults.
N.N. Srinidhi et.al. (2018) " Network optimizations in the Internet of Things: A
review" proposed Internet was at first used to exchange information parcels among clients and
information sources with an explicit IP address. The paper draws a consideration towards the
foundation of IoT and its refinement with different innovations, dialog on system improvement
in IoT and calculations characterization. At long last, best in class systems for IoT specifically to
arrange improvement are talked about dependent on the ongoing works and the audit is closed
with open issues and difficulties for system advancement in IoT.

5
 Mohamed Idhammad et.al. (2018) "Distributed Interruption Detection System for
Cloud Environments based on Data Mining techniques" proposed a dispersed Machine
Learning based interruption identification framework for Cloud situations. The proposed
framework is intended to be embedded in the Cloud one next to the other with the edge arrange
segments of the Cloud supplier. This permits to capture approaching system movement to the edge
arrange switches of the physical layer.
Liang Liang et.al.(2018) " Abnormal Detection of Electric Security Data based on
Scenario Modeling " have considered identification structure for electric security information
is proposed dependent on situation displaying that provisions the depictions of common security
issues. In this structure, the electric power framework security-related information is first gathered
from various assets continuously. At that point we make a situation show dependent on the
chronicled information for the normal security issues, for example, cautioning, framework blunder
or deadly mistake utilizing the affiliation rule mining calculations.
Muhamad Erza Aminanto et.al. (2017) "Wi-Fi Interruption Detection Using
Weighted-Feature Selection for Neural Networks Classifier" have examined the element
weighting techniques in existing machine students and take a gander at how they could be utilized
for the precise choice of the essential highlights. So as to approve our thought, we consider Wi-Fi
systems since unavoidable Internet-of-Things (IoT) gadgets make tremendous traffics and powerless
in the meantime. Recognizing known and obscure assaults in Wi-Fi systems stays extraordinary
testing undertakings.

6
S.No. Name of Author & Discussion Technique Used Result
Year
1. Aditya Chellam This paper, Tale heuristic The NSL-KDD dataset for
et.al.(2018) proposes the weight based reproducing a certifiable like
utilization of ordering procedure. situation and contrasting their
lethargic learning relative exhibitions and hw-
philosophies to IBk.
enhance by and
large execution of
IDS.
2. Omar Y. Al-Jarrah This paper proposes Semi-regulated Test results demonstrate that
et.al. (2018) a Semi-regulated Multi-Layered SMLC is better than tri-
Multi-Layered Clustering (SMLC) preparing, furnishing
Clustering equivalent recognition
((SMLC)) show for exactness with 20% less
the identification marked cases of preparing
and aversion of information.
system interruption.

7
3. N.N. Srinidhi et.al.
The paper draws a It was a review of This paper not just surveys, looks
(2018) consideration towards different papers at and combines the ongoing
the foundation of IoT related works, yet additionally
and its refinement with respects the creator's discoveries,
different innovations, arrangements and examines its
dialog on system convenience towards system
improvement in IoT streamlining in IoT.
and calculations
characterization.
4. Mohamed Idhammad A dispersed Machine standard Random The framework accomplished a
et.al. (2018) Learning based Forest classifier normal precision of 97%, a normal
interruption false positive rate of 0.21% and a
identification normal running time of 6.23s.
framework for Cloud
situations. The
proposed framework
was intended to be
embedded in the
Cloud one next to the
other with the edge
arrange segments of
the Cloud supplier.

8
5. Liang Liang Identification Electric Security The test results recommend
et.al.(2018) structure for Data based on that the proposed technique
electric security Scenario Modeling can successfully recognize and
information is keep the considerable lot of
proposed dependent the regular issues in the
on situation electric system.
displaying that
provisions the
depictions of
common security
issues.
6. Muhamad Erza examined the weighted-based accomplishing 99.72% F1
Aminanto element weighting machine learning score
et.al.[2017] techniques in model
existing machine
students and take a
gander at how they
could be utilized for
the precise choice
of the essential
highlights.
9
7. Aditya Shrivastava1 proposed a mixture PCNN Identification rate is high in
et.al [2013] show for highlight pressure of other neural
choice and system model
interruption
identification.
Highlight
determination is
imperative issue in
interruption
recognition.
8. Jayshri R. Patel et.al a strategy utilizing C4.5, CART, Controlling the interruption
[2013] Decision Trees Random Forest and
arrangement of REP Tree.
Interruption
recognition, as
indicated by their
highlights into
either meddling or
non nosy class is a
generally
considered issue.
10
Problem formulation
In the research work Interruption detection different problems are
faced. A hybrid model for feature selection and Interruption
detection is important issue in Interruption detection. The selection
of feature in attack attribute and normal traffic attribute is
challenging task. There are different problems that are given
below:
There are selection of known and unknown attack is also faced a
problem of classification. There is multiclass problem during the
classification of data.
Interruption detection is a problem of transportation infrastructure
protection owing to the fact that computer networks are at the core
of the operational control of much of the nation’s transportation.
The feature ranking and selection problem for Interruption
detection is similar in nature to various engineering

11
Another major problem that is also faced in the intrusion detection that
are given below:
 Security
 Authentication
 Attackers

12
OBJECTIVE
 To implement Interruption Detection System using hybrid
feature selection technique with KNN and SVM.
 To evaluate the performance on KNN and SVM in terms of
root mean square error and computational time.
 Calculate and analyse the result with different parameters
like Detection Rate (DR), False Alarm Rate (FAR), time to
build the model (TBM), and time to test the model (TT).
True Positive (TP) is the number of interruptions correctly
classified as an attack. True Negative (TN) is the number of
normal instances correctly classified as a benign packet.

13
Methodology

Use goal oriented approach and reduced some irrelevant feature of


instance under feature selection method.

Build a hybrid method that reduced execution complexity of mining


system and train classifier with low number of data set.

14
15
THE PROPOSED STEP FOR WORK IS

Step 1: Start weka tool.


Step 2: Read the KDDCups 99 dataset on the preprocessing.
Step 3: Apply the attribute selection to select the attributes.
Step 4: Note down the selected attribute and remove the unselected
attribute.
Step5: Classify the selected attribute with SVM and KNN classifier.
Step 6: Analyze the different values after the classification.
Step 7: visualize the resulted graph with different values.
Step 8: repeat the step 3 to step 7 for different classifiers.
Step 9: Stop

16
References
 Shrivastava, A., Baghel, M. and Gupta, H., 2013. A novel hybrid feature selection and
interruption detection based on PCNN and support vector machine. International journal of
computer technology and applications, 4(6), p.922.
 Gong, S., Gong, X. and Bi, X., 2011, July. Feature selection method for network interruption
based on GQPSO attribute reduction. In Multimedia Technology (ICMT), 2011 International
Conference on (pp. 6365-6368). IEEE.
 Patel, J.R., Performance Evaluation of Decision Tree Classifiers for Ranked Features of
Interruption Detection. Journal of Information, Knowledge and Research in Information
Technology, Nov, 12, pp.152-155.
 Liang, L., 2018. Abnormal Detection of Electric Security Data based on Scenario
Modeling. Procedia computer science, 139, pp.578-582.
 Aggarwal, M., 2013. Performance analysis of different feature selection methods in
interruption detection. International Journal of Scientific & Technology Research, 2(6),
pp.225-231.
 Idhammad, M., Afdel, K. and Belouch, M., 2018. Distributed Interruption Detection System
for Cloud Environments based on Data Mining techniques. Procedia Computer Science, 127,
pp.35-41.

17
 Aminanto, M.E., Tanuwidjaja, H.C., Yoo, P.D. and Kim, K., 2017, September. Wi-Fi
interruption detection using weighted-feature selection for neural networks classifier.
In Big Data and Information Security (IWBIS), 2017 International Workshop on(pp.
99-104). IEEE.
 Srinidhi, N.N., Kumar, S.D. and Venugopal, K.R., 2018. Network optimizations in
the Internet of Things: A review. Engineering Science and Technology, an
International Journal.
 Nguyen, H.T., Franke, K. and Petrovic, S., 2010, August. Towards a generic feature-
selection measure for interruption detection. In Pattern Recognition (ICPR), 2010
20th International Conference on (pp. 1529-1532). IEEE.
 Takkellapati, V.S. and Prasad, G.V.S.N.R.V., 2012. Network interruption detection
system based on feature selection and triangle area support vector
machine. International Journal of Engineering Trends and Technology, 3(4).
 Al-Jarrah, O.Y., Al-Hammdi, Y., Yoo, P.D., Muhaidat, S. and Al-Qutayri, M., 2017.
Semi-supervised Multi-Layered Clustering Model for Interruption Detection. Digital
Communications and Networks.
 Chellam, A., Ramanathan, L. and Ramani, S., 2018. Interruption Detection in
Computer Networks using Lazy Learning Algorithm. Procedia Computer
Science, 132, pp.928-936.

18
19

Vous aimerez peut-être aussi