Detection of Cyber Attack in Network using

Machine Learning Techniques

A documentation report submitted in partial fulfillment of the

requirements for the Award of the degree of
Bachelor In Technology In
Computer Science and Engineering

SUBMITTED BY
G. HIMA HARSHA - 121710301017
G. PAVAN KUMAR -121710301018
N. SURESH - 121710301033
C. RAKESH KUMAR-121710301043
UNDER THE ESTEEMED GUIDANCE OF
Ms. Bhanu Sree
ASSOCIATE PROFESSOR
GITAM

DEPARTMENT OF COMPUTER SCIENCE AND ENGINEERING

GITAM
(DEEMED TO BE UNIVERSITY)
VISAKHAPATNAM

DEPARTMENT OF COMPUTER SCIENCE AND ENGINEERING

GITAM
(DEEMED TO BE UNIVERSITY)
1
 CERTIFICATE
This is to certify that the project report titled “Detection of Cyber
Attack in Network using Machine Learning Techniques” is a
Bonafede record of work carried out by G. HIMA HARSHA –
(121710301017) G. PAVAN KUMAR (121710301018)
N. SURESH (121710301033) C. RAKESH KUMAR (121710301043)
submitted in partial fulfillment of requirements for the award of
Bachelor of Technology in Computer Science and Engineering
degree.

Project Guide: Ms. Bhanu Sree

Project Reviewer 1: Dr. SRINIVAS
L. CHAKRAVARTHY
Project Reviewer 2: Mrs. K Naga Sowjanya

2
3
 ABSTRACT

Contrasted with the past, improvements in PC and correspondence

innovations have given broad and propelled changes. The use of new
innovations give incredible advantages to people, organizations, and
governments, be that as it may, messes some up against them. For
instance, the protection of significant data, security of put away
information stages, accessibility of information and so forth. Contingent
upon these issues, digital fear based oppression is one of the most
significant issues in this day and age. Digital fear, which made a great deal
of issues people and establishments, has arrived at a level that could
undermine open and nation security by different gatherings, for example,
criminal association, proficient people and digital activists. Along these
lines, Intrusion Detection Systems (IDS) has been created to maintain a
strategic distance from digital assaults. Right now, learning the bolster
support vector machine (SVM) calculations were utilized to recognize port
sweep endeavors dependent on the new CICIDS2017 dataset with 97.80%,
69.79% precision rates were accomplished individually. Rather than SVM
we can introduce some other algorithms like random forest, CNN, ANN
where these algorithms can acquire accuracies like SVM – 93.29, CNN –
63.52, Random Forest – 99.93, ANN – 99.11.
.

TABLE OF CONTENTS

1.INTRODUCTION

1.1 Motivation
1.2 Existing System
1.3 Objective
1.4 Outcome
1.5 Applications

4
1.6 Structure of project
1.5.2 System Design
1.5.3 Implementation
1.5.4 Testing
1.5.5 Deployment of System and Maintenance

1.6 Functional Requirements

1.7 Non-Functional Requirements

1.7.1 Advantages of Non-Functional Requirements

1.7.2 Disadvantages Non-Functional Requirements

2.LITERATURE SURVEY
3.PROBLEM IDENTIFICATION & OBJECTIVES
3.1 Existing Approach

3.2 Proposed System

3.3 Modules
3.4 Algorithms
4.SYSTEM DESIGN
5.IMPLEMENTATION
5.1 Flowchart
5.2 Code
6.RESULTS AND DISCUSSIONS
7.CONCLUSION AND FUTURE SCOPE
8.REFERENCES

5
Detection of Cyber Attack in Network using Machine Learning
Techniques

6
 ABSTRACT

Contrasted with the past, improvements in PC and correspondence

innovations have given broad and propelled changes. The use of new
innovations give incredible advantages to people, organizations, and
governments, be that as it may, messes some up against them. For
instance, the protection of significant data, security of put away
information stages, accessibility of information and so forth. Contingent
upon these issues, digital fear based oppression is one of the most
significant issues in this day and age. Digital fear, which made a great deal
of issues people and establishments, has arrived at a level that could
undermine open and nation security by different gatherings, for example,
criminal association, proficient people and digital activists. Along these
lines, Intrusion Detection Systems (IDS) has been created to maintain a
strategic distance from digital assaults. Right now, learning the bolster
support vector machine (SVM) calculations were utilized to recognize port
sweep endeavors dependent on the new CICIDS2017 dataset with 97.80%,
69.79% precision rates were accomplished individually. Rather than SVM
we can introduce some other algorithms like random forest, CNN, ANN
where these algorithms can acquire accuracies like SVM – 93.29, CNN –
63.52, Random Forest – 99.93, ANN – 99.11.
.

7
 1.INTRODUCTION
Contrasted with the past, improvements in PC and correspondence
innovations have given broad and propelled changes. The use of new
innovations give incredible advantages to people, organizations, and
governments, be that as it may, messes some up against them. For
instance, the protection of significant data, security of put away
information stages, accessibility of information and so forth. Contingent
upon these issues, digital fear based oppression is one of the most
significant issues in this day and age. Digital fear, which made a great deal
of issues people and establishments, has arrived at a level that could
undermine open and nation security by different gatherings, for example,
criminal association, proficient people and digital activists. Along these
lines, Intrusion Detection Systems (IDS) has been created to maintain a
strategic distance from digital assaults. Right now, learning the bolster
support vector machine (SVM) calculations were utilized to recognize port
sweep endeavors dependent on the new CICIDS2017 dataset with 97.80%,
69.79% precision rates were accomplished individually. Rather than SVM
we can introduce some other algorithms like random forest, CNN, ANN
where these algorithms can acquire accuracies like SVM – 93.29, CNN –
63.52, Random Forest – 99.93, ANN – 99.11.

1.1 MOTIVATION
The use of new innovations give incredible advantages to people,
organizations, and governments, be that as it may, messes some up against
them. For instance, the protection of significant data, security of put away
information stages, accessibility of information and so forth. Contingent
upon these issues, digital fear based oppression is one of the most
significant issues in this day and age. Digital fear, which made a great deal
of issues people and establishments, has arrived at a level that could
undermine open and nation security by different gatherings, for example,
criminal association, proficient people and digital activists. Along these
lines, Intrusion Detection Systems (IDS) has been created to maintain a
strategic distance from digital assaults.

8
 1.2 Objectives
Objective of this project is to detect cyber attacks by using machine
learning algorithms like
• ANN
• CNN
• Random forest
1.3 Objectives

The objective of the project is to detect cyber attacks

1.4 Applications

It can be used in cyber attacks detection

1.5 STRUCTURE OF PROJECT (SYSTEM ANALYSIS)

9
Fig: 1 Project SDLC

• Project Requisites Accumulating and Analysis

• Application System Design

• Practical Implementation

• Manual Testing of My Application

• Application Deployment of System

• Maintenance of the Project

1.5.1REQUISITES ACCUMULATING AND ANALYSIS

It’s the first and foremost stage of the any project as our is an
academic leave for requisites amassing, we followed of IEEE
Journals and Amassed so many IEEE Relegated papers and
final culled a Paper designated “Individual web revisitation by
setting and substance importance input and for analysis stage

10
we took referees from the paper and did literature survey of
some papers and amassed all the Requisites of the project in
this stage

1.5.2SYSTEM DESIGN

In System Design has divided into three types like GUI

Designing, UML Designing with avails in development of
project in facile way with different actor and its utilizer case
by utilizer case diagram, flow of the project utilizing
sequence, Class diagram gives information about different
class in the project with methods that have to be utilized in
the project if comes to our project our UML Will utilizable in
this way The third and post import for the project in system
design is Database design where we
endeavor to design database predicated on the number of
modules in our project 1.5.3.IMPLEMENTATION

The Implementation is Phase where we endeavor to give the

practical output of the work done in designing stage and most
of Coding in Business logic lay coms into action in this stage
its main and crucial part of the project

1.5.4.TESTING

UNIT TESTING

It is done by the developer itself in every stage of the project

and fine-tuning the bug and module predicated additionally
done by the developer only here we are going to
solve all the runtime errors
11
MANUAL TESTING

As our Project is academic Leave, we can do any

automatic testing so we follow manual testing by
endeavor and error methods.

1.5.5.DEPLOYMENT OF SYSTEM AND MAINTENANCE

When the undertaking is complete yare, we will come to
sending of customer framework in truly world as its scholastic
leave we did organization in our school lab just with all need
Software's with Windows OS .

The Maintenance of our Project is one-time measure as it were

1.6 FUNCTIONAL REQUIREMENTS

1.Data Collection

2.Data Pre-handling

3.Training and Testing

4.Modeling

5.Predicting

1.7 NON-FUNCTIONAL REQUIREMENTS NON-

12
FUNCTIONAL REQUIREMENT (NFR) indicates the quality
property of a product framework. They judge the product
framework dependent on Responsiveness, Usability, Security,
Portability and other non-useful principles that are basic to the
accomplishment of the product framework. Illustration of
nonfunctional prerequisite, "how quick does the site load?"
Failing to meet non-utilitarian necessities can bring about
frameworks that neglect to fulfill client needs. Non-practical
Requirements permits you to force imperatives or limitations on
the plan of the framework across the different light-footed
accumulations. Model, the site should stack in 3 seconds when
the quantity of concurrent clients is > 10000. Portrayal of non-
utilitarian necessities is similarly just about as basic as a useful
prerequisite.

Instances OF NON-FUNCTIONAL REQUIREMENTS

Here, are a few instances of non-useful necessity:

1.7.1.1 Users should transfer dataset

1.7.1.2 The product ought to be compact. So moving from one

OS to other OS doesn't make any issue.

1.7.1.3 Privacy of data, the fare of confined advances, scholarly

13
property rights, and so forth ought to be reviewed.

1.7.2 ADVANTAGES OF NON-FUNCTIONAL

REQUIREMENT

Advantages/aces of Non-utilitarian testing are:

• The nonfunctional necessities guarantee the product

framework adheres to lawful and consistence rules.

• They guarantee the unwavering quality, accessibility, and

execution of the product framework

• They guarantee great client experience and simplicity of

working the product.

• They help in figuring security strategy of the product

framework.

1.7.3 DISADVANTAGES OF NON-FUNCTIONAL

REQUIREMENT

Cons/disadvantages of non-practical necessity are:

• Non useful necessity may influence the different significant

14
level programming subsystem

• They require exceptional thought during the product

engineering/significant level plan stage which builds costs.

• Their execution doesn't normally guide to the particular

programming sub-framework,

• It is difficult to adjust non-practical once you pass the design

stage.

2.LITERATURE SURVEY

15
2.1 R. Christopher, “Port scanning techniques and the defense against
them,” SANS Institute, 2001.

Port Scanning is quite possibly the most well known strategies assailants
use to find benefits that they can adventure to break into frameworks. All
frameworks that are associated with a LAN or the Internet by means of a
modem run benefits that tune in to notable and not so notable ports. By
port filtering, the assailant can track down the accompanying data about
the focused on frameworks: what administrations are running, what clients
own those administrations, regardless of whether unknown logins are
upheld, and whether certain organization administrations require
verification. Port examining is cultivated by making an impression on each
port, each in turn. The sort of reaction got demonstrates whether the port is
utilized and can be examined for additional shortcomings. Port scanners
are imperative to organize security experts since they can uncover
conceivable security weaknesses on the focused on framework. Similarly
as port outputs can be ran against your frameworks, port sweeps can be
recognized and the measure of data about open administrations can be
restricted using the legitimate apparatuses. Each freely accessible
framework has ports that are open and accessible for use. The item is to
restrict the openness of open ports to approved clients and to deny
admittance to the shut ports.

2.2 S. Staniford, J. A. Hoagland, and J. M. McAlerney, "Useful

computerized recognition of subtle portscans," Journal of Computer
Security, vol. 10, no. 1-2, pp. 105–136, 2002.

Portscanning is a typical movement of extensive significance. It is

regularly utilized by PC assailants to portray has or networks which they
are thinking about threatening movement against. Accordingly it is helpful
for framework chairmen and other organization protectors to recognize
portscans as potential fundamentals to a more genuine assault. It is
additionally broadly utilized by network safeguards to comprehend and
discover weaknesses in their own organizations. Along these lines it is of
impressive interest to aggressors to decide if the safeguards of an

16
organization are portscanning it routinely. Nonetheless, protectors won't
typically wish to conceal their portscanning, while assailants will. For
definiteness, in the rest of this paper, we will discuss the assailants
checking the organization, and the safeguards attempting to recognize the
sweep. There are a few legitimate/moral discussions about portscanning
what break out routinely on Internet mailing records and newsgroups.

2.3 M. C. Raja and M. M. A. Rabbani, "Joined examination of help

vector machine and rule part investigation for ids," in IEEE
International Conference on Communication and Electronics Systems,
2016, pp. 1–5.

Contrasted with the past security of organized frameworks has become a

basic all inclusive issue that impacts people, ventures and governments.
The pace of assaults against arranged frameworks has expanded
exaggeratedly, and the procedures utilized by the aggressors are
proceeding to develop. For instance, the protection of significant data,
security of put away information stages, accessibility of information and so
on Contingent upon these issues, digital psychological oppression is
perhaps the main issues in this day and age. Digital dread, which made a
great deal of issues people and foundations, has arrived at a level that
could compromise public and country security by different gatherings like
criminal associations, proficient people and digital activists. Interruption
recognition is one of the arrangements against these assaults. A free and
compelling methodology for planning Intrusion Detection Systems (IDS)
is Machine Learning. In this examination, profound learning and backing
vector machine (SVM) calculations were utilized to recognize port output
endeavors dependent on the new CICIDS2017 dataset Introduction
Network Intrusion Detection System (IDS) is a product based application
or an equipment gadget that is utilized to distinguish noxious conduct in
the organization [1,2]. In light of the discovery strategy, interruption
recognition is characterized into abnormality based and signature-based.

17
2.4 S. Aljawarneh, M. Aldwairi, and M. B. Yassein, "Inconsistency
based interruption discovery framework through include
determination examination and building crossover productive model,"
Journal of Computational Science, vol. 25, pp. 152–160, 2018.

n network security, interruption location assumes a significant part.

Highlight subsets got by various element choice techniques will prompt
distinctive exactness of interruption discovery. Utilizing singular
component determination technique can be precarious in various
interruption location situations. In this paper, the possibility of troupe is
applied to include choice to change highlight subsets. Highlight choice is
changed over into a two-class issue, and odd number of highlight
determination techniques is utilized for casting a ballot strategy to choose
whether an element is required or disposed of. In real activity, mean
diminishing pollution, irregular backwoods classifier, security choice,
recursive element end and chi-square test are utilized. Highlight subsets
acquired from them will be changed by our proposed technique to get
gathering highlight subsets. To test the presentation, support vector
machine, choice tree, knn and multi-facet discernment are utilized to
notice and contrast the grouping exactness and gathering highlight subsets.
Three interruption identification informational indexes, including
kddcup99, cidds-001 and unsw_nb15 are utilized in our tests. The best
outcome is considered cidds-001 with a 99.40% order precision.

18
 3. PROBLEM ANALYSIS
3.1 EXISTING APPROACH:
Chaste Bayes and Principal Component Analysis (PCA) were been utilized
with the KDD99 dataset by Almansob and Lomte [9].Similarly, PCA,
SVM, and KDD99 were utilized Chithik and Rabbani for IDS [10]. In
Aljawarneh et al's. Paper, their appraisal and assessments were passed on
dependent on the NSL-KDD dataset for their IDS model [11] Composing
investigates show that KDD99 dataset is ceaselessly utilized for IDS [6]–
[10].There are 41 features in KDD99 and it was made in 1999.
Subsequently, KDD99 is old and doesn't give any information about
bleeding edge new attack types, model, multi day abuses, etc. As such we
used a forefront and new CICIDS2017 dataset [12] in our examination.

3.11 Drawbacks

1) Strict Regulations

2) Difficult to work with for non-specialized clients

3) Restrictive to assets

4) Constantly requirements Patching

5) Constantly being assaulted

3.2 Proposed System

significant strides of the calculation are given in underneath. 1)

Normalization of each dataset. 2) Convert that dataset into the testing and
preparing. 3) Form IDS models with the assistance of utilizing RF, ANN,
CNN and SVM calculations. 4) Evaluate each model's exhibitions

.
19
3.2.1 Advantages

• Protection from malevolent assaults on your organization.

• Deletion or potentially ensuring pernicious components inside a

previous organization.

• Prevents clients from unapproved admittance to the organization.

• Deny's projects from specific assets that could be tainted.

• Securing classified data

20
3.3 Software And Hardware Requirements

SOFTWARE REQUIREMENTS
The functional requirements or the overall description documents
include the product perspective and features, operating system and
operating environment, graphics requirements, design constraints and user
documentation.
The appropriation of requirements and implementation constraints
gives the general overview of the project in regards to what the areas of
strength and deficit are and how to tackle them.

• Python idel 3.7 version (or)

• Anaconda 3.7 ( or)
• Jupiter (or)
• Google colab

HARDWARE REQUIREMENTS

21
Minimum hardware requirements are very dependent on the particular
software being developed by a given Enthought Python / Canopy / VS
Code user. Applications that need to store large arrays/objects in
memory will require more RAM, whereas applications that need to
perform numerous calculations or tasks more quickly will require a
faster processor.
• Operating system : windows, linux
• Processor : minimum intel i3
• Ram : minimum 4 gb
• Hard disk : minimum 250gb

4. SYSTEM DESIGN
22
UML DIAGRAMS
The System Design Document depicts the framework necessities, working
climate, framework and subsystem engineering, records and information
base plan, input designs, yield formats, human-machine interfaces, nitty
gritty plan, preparing rationale, and outer interfaces.

Worldwide Use Case Diagrams:

Recognizable proof of entertainers:

Entertainer: Actor addresses the job a client plays as for the framework.
An entertainer communicates with, yet has no influence over the utilization
cases.

USE CASE DIAGRAM

23
 Start

Localhost

Register & Login to Application

Real Time Malware Detection

Data Stores in SQL

User Add Data

User

Attack Classification based on

model

Detection of Attack

Visualisation

End

Fig 1: Use Case Diagram

CLASS DIAGRAM

24
 User
agriculture

Start()
Localhost()
Register & Login to Application() System
Real Time Malware Detection()
Data Stores in SQL()
User Add Data()
Attack Classification based on model()
Detection of Attack()
Visualisation()
end()

Fig 2:Class Diagram

25
SEQUENCE DIAGRAM

Us er Sy s tem

Start

Loc alhos t

Regis ter & Login to Application

Real Time Malware Detec tion

Data Stores in SQL

Us er Add Data

Attac k Clas s ifi cation based on model

Detec t ion of A ttac k

Vi sualisation

Fig 3: Sequence Diagram

26
 5.IMPLEMENTATION

5.1 FLOW CHART:

27
6.CODE

28
29
30
31
32
33
34
35
36
37
38
 7.RESULTS AND DISCUSSIONS

Data preprocessing

39
Data EDA

40
41
ML Deploy

42
From the score accuracy we concluding the DT & RF give better accuracy and
building pickle file for predicting the user input

43
Application

44
Localhost - in cmd python app.py

45
Enter the input

46
Predict attack -

47
 8. CONCLUSION
Right now, estimations of help vector machine, ANN, CNN, Random Forest
and profound learning calculations dependent on modern CICIDS2017
dataset were introduced relatively. Results show that the profound learning
calculation performed fundamentally preferable outcomes over SVM, ANN,
RF and CNN. We are going to utilize port sweep endeavors as well as other
assault types with AI and profound learning calculations, apache Hadoop
and sparkle innovations together dependent on this dataset later on. All these
calculation helps us to detect the cyber attack in network. It happens in the
way that when we consider long back years there may be so many attacks
happened so when these attacks are recognized then the features at which
values these attacks are happening will be stored in some datasets. So by
using these datasets we are going to predict whether cyber attack is done or
not. These predictions can be done by four algorithms like SVM, ANN, RF,
CNN this paper helps to identify which algorithm predicts the best accuracy
rates which helps to predict best results to identify the cyber attacks
happened or not.

FUTURE SCOPE
In enhancement we will add some ML Algorithms to increase accuracy

48
 8.REFERENCES
[1]iK.iGraves,iCeh:iOfficialicertifiediethicalihackerireviewiguide:iExami312-
50.iJohniWileyi&iSons,i2007.i
[2]iR.iChristopher,i“Portiscanningitechniquesianditheidefenseiagainstithem,”iSA
NSiInstitute,i2001.i
[3]iM.iBaykara,iR.iDas¸,iandiI.iKaradoi˘gan,i“Bilgiigi¨uvenlii˘giisistemlerindeik
ullanilaniarac¸lariniincelenmesi,”iini1stiInternationaliSymposiumioniDigitaliFore
nsicsiandiSecurityi(ISDFS13),i2013,ipp.i231–239.i
[4]iS.iStaniford,iJ.iA.iHoagland,iandiJ.iM.iMcAlerney,i“Practicaliautomatedidete
ctioniofistealthyiportscans,”iJournaliofiComputeriSecurity,ivol.i10,ino.i1-
2,ipp.i105–136,i2002.i
[5]iS.iRobertson,iE.iV.iSiegel,iM.iMiller,iandiS.iJ.iStolfo,i“Surveillanceidetectio
niinihighibandwidthienvironments,”iiniDARPAiInformationiSurvivabilityiConfer
enceiandiExposition,i2003.iProceedings,ivol.i1.iIEEE,i2003,ipp.i130–138.i
[6]iK.iIbrahimiiandiM.iOuaddane,i“Managementiofiintrusionidetectionisystemsib
asedkdd99:iAnalysisiwithildaiandipca,”iiniWirelessiNetworksiandiMobileiComm
unicationsi(WINCOM),i2017iInternationaliConferenceion.iIEEE,i2017,ipp.i1–6.i
[7]iN.iMoustafaiandiJ.iSlay,i“Theisignificantifeaturesiofitheiunswnb15ianditheik
dd99idataisetsiforinetworkiintrusionidetectionisystems,”iiniBuildingiAnalysisiDat
asetsiandiGatheringiExperienceiReturnsiforiSecurityi(BADGERS),i2015i4thiInte
rnationaliWorkshopion.iIEEE,i2015,ipp.i25–31.i
[8]iL.iSun,iT.iAnthony,iH.iZ.iXia,iJ.iChen,iX.iHuang,iandiY.iZhang,i“Detectioni
andiclassificationiofimaliciousipatternsiininetworkitrafficiusingibenford’silaw,”ii
niAsiaPacificiSignaliandiInformationiProcessingiAssociationiAnnualiSummitian
diConferencei(APSIPAiASC),i2017.iIEEE,i2017,ipp.i864–872.i
[9]iS.iM.iAlmansobiandiS.iS.iLomte,i“Addressingichallengesiforiintrusionidetect
ionisystemiusinginaiveibayesiandipcaialgorithm,”iiniConvergenceiiniTechnology
i(I2CT),i2017i2ndiInternationaliConferenceifor.iIEEE,i2017,ipp.i565–568.i
[10]iM.iC.iRajaiandiM.iM.iA.iRabbani,i“Combinedianalysisiofisupportivectorim
achineiandiprincipleicomponentianalysisiforiids,”iiniIEEEiInternationaliConferen
ceioniCommunicationiandiElectronicsiSystems,i2016,ipp.i1–5.i

49
[11]iS.iAljawarneh,iM.iAldwairi,iandiM.iB.iYassein,i“Anomaly-
basediintrusionidetectionisystemithroughifeatureiselectionianalysisiandibuildingi
hybridiefficientimodel,”iJournaliofiComputationaliScience,ivol.i25,ipp.i152–
160,i2018.i
[12]iI.iSharafaldin,iA.iH.iLashkari,iandiA.iA.iGhorbani,i“Towardigeneratingiain
ewiintrusionidetectionidatasetiandiintrusionitrafficicharacterization.”iiniICISSP,i
2018,ipp.i108–116.i
[13]iD.iAksu,iS.iUstebay,iM.iA.iAydin,iandiT.iAtmaca,i“Intrusionidetectioniwith
icomparativeianalysisiofisupervisedilearningitechniquesiandifisheriscoreifeatureis
electionialgorithm,”iiniInternationaliSymposiumioniComputeriandiInformationiS
ciences.iSpringer,i2018,ipp.i141–149.i
[14]iN.iMarir,iH.iWang,iG.iFeng,iB.iLi,iandiM.iJia,i“Distributediabnormalibeha
vioridetectioniapproachibasedionideepibeliefinetworkiandiensembleisvmiusingisp
ark,”iIEEEiAccess,i2018.i
[15]iP.iA.iA.iResendeiandiA.iC.iDrummond,i“Adaptiveianomalybasediintrusioni
detectionisystemiusingigeneticialgorithmiandiprofiling,”iSecurityiandiPrivacy,ivo
l.i1,ino.i4,ip.ie36,i2018.i
[16]iC.iCortesiandiV.iVapnik,i“Supportvectorinetworks,”iMachineilearning,ivol.i
20,ino.i3,ipp.i273–297,i1995.i
[17]iR.iShouval,iO.iBondi,iH.iMishan,iA.iShimoni,iR.iUnger,iandiA.iNagler,i“A
pplicationiofimachineilearningialgorithmsiforiclinicalipredictiveimodeling:iaidata
miningiapproachiinisct,”iBoneimarrowitransplantation,ivol.i49,ino.i3,ip.i332,i201
4.i

50