Human Annotator For Imbalanced Dossier

ABSTRACT
An imbalanced dataset is relevant primarily in the context of supervised machine

learning involving two or more classes. Imbalance means that the number of data
points available for a different class is different, Imbalanced data sets is a special
case for classification problem where the class distribution is not uniform among the
classes. Imbalanced classes are a common problem in machine learning classification
where there is a disproportionate ratio of observations in each class. Class imbalance
can be found in many different areas including medical diagnosis, spam filtering, and
fraud detection. Typically, they are composed of two classes, the majority (negative)
class, and the minority (positive) class. Information about experimental studies sets.
These types of data sets are typically found on websites that collect and aggregate
data sets. These aggregators tend to have data sets from multiple sources, without
much creation. In this case, that‟s a good thing too much creation gives us overly neat
data sets that are hard to label. Active learning is undoubtedly effective, but several
recent studies have indicated that active learning is failed when it is applied to data. In
our project, Human Annotator will collect the data‟s from the public post and he will
separate labeled and unlabelled data sets. User needs to register their details and
they can view their learning materials. The labelled and unlabelled data‟s have been
analyzed by human annotator and matched with labelled appropriate data sets are
achieved and it will be available for learners.
v
TABLE OF CONTENTS
CHAPTER TITLE PAGE

NO. NO.
ABSTRACT v
LIST OF FIGURES ix
LIST OF TABLES x
LIST OF ABBREVIATIONS xi
1 INTRODUCTION 1
1.1 OUTLINE OF THE PROJECT 1
1.2 PURPOSE OF THE PROJECT 1
1.3 ACTIVE LEARNING PARADIGM 2
1.4 MACHINE LEARNING 4
1.4.1 Machine Learning Methods 5
1.4.2 Applications of Machine Learning 5
1.4.3 Advantages 5
1.5 SYSTEM REQUIREMENTS 6
1.5.1 Hardware Requirements 6
1.5.2 Software Requirements 6
2 LITERACTURE SURVEY 7
2.1 SURVEY 1 7
2.2 SURVEY 2 7
2.3 SURVEY 3 8
2.4 SURVEY 4 9
2.5 SURVEY 5 10
2.6 SURVEY 6 10
3 METHODOLOGY 12
3.1 SYSTEM ANALYSIS 12
3.1.1 Analysis Model 12
vi
3.2 EXISTING SYSTEM 14
3.2.1 Disadvantage 14
3.3 PROPOSED SYSTEM 15
3.3.1 Advantage 15
3.4 MODULES OF THE PROJECT 16
3.4.1 Authentication and Authorization 16
3.4.2 Material Upload 16
3.4.3 Active Learning 17
3.4.4 Active Learning with Extreme Learning 17
Machine
3.4.5 Learning Material 17
3.5 FEASIBILITY REPORT 17
3.5.1 Technical Feasibility 18
3.5.2 Operational Feasibility 19
3.5.3 Economic Feasibility 19
3.6 SOFTWARE REQUIREMENT SPECIFICATION 19
3.6.1 Developers Responsibilities Overview 20
3.6.2 Functional Requirements 20
3.6.3 Non-Functional Requirements 20
3.6.4 Performance Requirements 21
3.7 THE .NET FRAMEWORK ARCHITECTURE 21
3.7.1 Common Language Runtime Engine 22
3.7.2 Language Independence 23
3.7.3 Framework Class Library 23
3.7.4 Simplified Deployment 23
3.7.5 Security 23
3.7.6 Portability 23
3.7.7 Common Language Specification 24
3.8 SQL SERVER 2014 24
3.9 SYSTEM DESIGN 25
3.10 SYSTEM TESTING AND IMPLEMENTATION 26
vii
3.10.1 Strategic Approach to Software Testing 26
3.11 SYSTEM SECURITY 27
3.11.1 Security in Software 27
3.11.1.1 Client Side Validation 28
3.11.1.2 Server Side Validation 28
4 RESULTS AND DISCUSSIONS 29
4.1 NORMILIZATION 29
4.2 E-R DIAGRAMS 32
4.3 DATA FLOW DIAGRAM 37
4.4 ALGORITHM 38
5 CONCLUSION AND FUTURE WORK 41
REFERENCES 42
APPENDIX 43
A. SAMPLE CODE 43
B. SCREENSHOTS 53
C. PUBLICATION WITH PLAGARISM 59
REPORT
viii
LIST OF FIGURES
FIGURE TITLE PAGE

NO. NO.
1.1 Overview of Machine Learning 4
3.1 Spiral Model 13
3.2 System Architecture 16
3.3 .NET Architecture 22
3.4 Server Architecture 24
4.1 ER Diagram for Human Annotator 33
4.2 Flowchart of Human Annotator 34
4.3 Sequence Diagram for Human Annotator 35
4.4 Activity Diagram for Human Annotator 36
4.5 Level-0 Data Flow Diagram 37
4.6 Level-1 Data Flow Diagram 38
ix
LIST OF TABLES
TABLE TITLE PAGE

NO. NO.
4.1 Design for User Register 29
4.2 File Upload 31
x
LIST OF ABBREVIATIONS
ABBREVIATIONS EXPANSION
ANSI American National Standards Institute
AOW-ELM Active Online Weighted-Extreme Learning Machine
ASP Active Server Page
ASR Automatic Speech Recognition
BCL Base Class Library
BLOBs Binary Large Objects
CBIR Content Based Image Retrieval
CLR Common Language Runtime
ELM Extreme Learning Machine
FCL Framework Class Library
MLP Multiple Level Perceptron
SQL Structured Query language
SRS Software Requirement Specification
SVM Support Vector Machine
WCF Windows Communication Foundation
xi
CHAPTER 1
INTRODUCTION
1.1. OUTLINE OF THE PROJECT
Imbalanced data typically refers to classification tasks where the classes

are not represented equally. Most of the real-world classification problems display
some level of class imbalance, which happens when there are not sufficient instances
of the data that correspond to either of the class labels. Therefore, it is imperative to
choose the evaluation metric of your model correctly. If it is not done, then you might
end up adjusting/optimizing a useless parameter. In a real business-first scenario, this
may lead to a complete waste.
There are problems where a class imbalance is not just common it is bound
to happen. For example, the datasets that deal with fraudulent and non-fraudulent
transactions, it is very likely that the number of fraudulent transactions as compares
to the number of non-fraudulent transactions will be very much less. And this is where
the problem arises. When the dataset has underrepresented data, the class
distribution starts skew.
Due to the inherent complex characteristics of the dataset, learning from
such data requires new understandings, new approaches, new principles, and new
tools to transform data. And moreover, this cannot anyway guarantee an efficient
solution to your business problem.
1.2. PURPOSE OF THE PROJECT
Dealing with imbalanced datasets includes various strategies such as

improving classification algorithms or balancing classes in the training data
(essentially a data pre-processing step) before providing the data as input to the
machine learning. The latter technique is preferred as it has a broader application and
adaptation. Moreover, the time taken to enhance is often higher than to generate the
required samples. But for research purposes, in our project Human Annotator will
1
collect the imbalanced data, he will segregate labeled and unlabelled data to provide
a complete learning material.
1.3. ACTIVE LEARNING PARADIGM
Active learning is a popular machine learning paradigm and it is frequently

deployed in the scenarios when large scale instances are easily collected, but
labeling them is expensive and/or time-consuming. By adopting active learning, a
classification model can iteratively interact with human experts to only select those
most significant instances for labeling and to further promote its performance as
quickly as possible. Therefore, the merits of active learning lie in decreasing both the
burden of human experts and the complexity of training instances but acquiring a
classification model that delivers superior or comparable performance to the model
with labeling all instances.
Past research has accumulated a large number of active learning models,

and generally, we have several different taxonomies to organize these models. Based
on different ways of entering the unlabeled data, active learning can be divided into
pool-based and stream-based models. The former previously collects and prepares
all unlabeled instances, while the latter can only visit a batch of newly arrived
unlabeled data at each specific time point.
According to different numbers of the labeled instances in each round, we

have single-mode and batch-mode learning models. As their names indicate, the
single-mode model only labels one unlabeled instance on each round, while the
batch-mode labels a batch of unlabeled examples once. In addition, we have several
different significance measures to rank unlabeled instances, including uncertainty,
representativeness, inconsistency, variance, and error. Each significance measure
has a criterion for evaluating which instances are the most important for improving the
performance of the classification model. For example, uncertainty considers the most
important unlabeled instance to be the nearest one to the current classification
boundary; representativeness considers the unlabeled instance that can represent a
new group of instances, e.g., a cluster, to be more important, and inconsistency
2
considers the unlabeled instance that has the most predictive divergence among
multiple diverse baseline classifiers to be more significant. In addition, active learning
models can also be divided into different categories according to which kind of
classifier has been adopted.
Some popular classifiers, including naive Bayes, k-nearest neighbors,
decision tree, multiple level perceptron (MLP), logistic regression, support vector
machine (SVM), and extreme learning machine (ELM), have all been developed to
satisfy the requirements of active learning. In the past decade, active learning has
also been deployed in a variety of real world applications, such as video annotation,
image retrieval, text classification, remote sensing, image annotation, speech
recognition, network intrusion detection, and bioinformatics.
The proposed algorithm is named active online weighted ELM (AOW-
ELM), and it should be applied in the pool-based batch-mode active learning scenario
with an uncertainty significance measure and ELM classifier. In AOW-ELM, we first
take advantage of the idea of cost-sensitive learning to select the weighted ELM
(WELM) as the base learner to address the class imbalance problem existing in the
procedure of active learning. Then, we adopt the AL-ELM algorithm presented in our
previous paper to construct an active learning framework. Next, we deduce an
efficient online learning mode of WELM in theory and design an effective weight
update rule. Finally, benefiting from the idea of the margin exhaustion criterion, we
present a more flexible and effective early stopping criterion. Moreover, we try to
simply discuss why active learning can be disturbed by skewed instance distribution,
further investigating the influence of three main distribution factors, including the class
imbalance ratio, class overlapping, and small disjunction. Specifically, we suggest
adopting the clustering techniques to previously select the initially labeled seed set,
and thereby avoid the missed cluster effect and cold start phenomenon as much as
possible. Experiments are conducted on 32 binary-class imbalanced data sets, and
the results demonstrate that the proposed algorithmic framework is generally more
effective and efficient than several state-of-theart active learning algorithms that were
specifically designed for the class imbalance scenario.
3
1.4.1. Machine Learning Methods
Some of the methods of Machine Learning algorithm are categorized as

 SUPERVISED LEARNING
A Supervised learning algorithm learns from labelled training data, helps

you to predict outcomes for unforeseen data. It is highly accurate and trustworthy
method.
 UNSUPERVISED LEARNING
Unsupervised learning algorithm is the type of self - organized with the

help of previously unknown patterns in dataset without pre-existing labels.
 SEMI-SUPERVISED LEARNING
Semi-supervised learning is the combination of both supervised and

unsupervised which means labeled and unlabelled data.
 REINFORCEMENT MACHINE LEARNING
Reinforcement machine learning is an area of machine learning

concerned with how software agents ought to take actions in an environment so as to
maximize some notion of cumulative reward.
1.4.2. Applications of Machine Learning
 Video Surveillance
 Social Media Services
 Email Spam and Malware Filtering
 Financial Services
 Health Care
 Retail
 Transportation
1.4.3. Advantages
 Computational property is cheaper and more powerful.
 Affordable data storage.
 It can analysis complex data quickly and automatically.
 It produces more accurate results.

Human Annotator For Imbalanced Dossier

Uploaded by

Copyright:

Human Annotator For Imbalanced Dossier

Uploaded by

Document Information

Original Title

Copyright

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Human Annotator For Imbalanced Dossier

Uploaded by

Copyright:

ABSTRACT

An imbalanced dataset is relevant primarily in the context of supervised machine

CHAPTER TITLE PAGE

FIGURE TITLE PAGE

TABLE TITLE PAGE

1.1. OUTLINE OF THE PROJECT

Imbalanced data typically refers to classification tasks where the classes

1.2. PURPOSE OF THE PROJECT

Dealing with imbalanced datasets includes various strategies such as

1.3. ACTIVE LEARNING PARADIGM

Active learning is a popular machine learning paradigm and it is frequently

Past research has accumulated a large number of active learning models,

According to different numbers of the labeled instances in each round, we

Some of the methods of Machine Learning algorithm are categorized as

A Supervised learning algorithm learns from labelled training data, helps

Unsupervised learning algorithm is the type of self - organized with the

Semi-supervised learning is the combination of both supervised and

Reinforcement machine learning is an area of machine learning

You might also like