Harsh Bhasin | Jamia Hamdard (original) (raw)

Papers by Harsh Bhasin

Alzheimer's & Dementia, 2022

The diagnosis of mild cognitive impairment (MCI), which is deemed a formative phase of dementia, ... more The diagnosis of mild cognitive impairment (MCI), which is deemed a formative phase of dementia, may greatly assist clinicians in delaying its headway toward dementia. This article proposes a deep learning approach based on a triploid genetic algorithm, a proposed variant of genetic algorithms, for classifying MCI converts and non‐converts using structural magnetic resonance imaging data. It also explores the effect of the choice of activation functions and that of the selection of hyper‐parameters on the performance of the model. The proposed work is a step toward automated convolutional neural networks. The performance of the proposed method is measured in terms of accuracy and empirical studies exhibit the preeminence of our proposed method over the existing ones. The proposed model results in a maximum accuracy of 0.97961. Thus, it may contribute to the effective diagnosis of MCI and may prove important in clinical settings.

2021 2nd Asia Conference on Computers and Communications (ACCC), 2021

Mild Cognitive Impairment (MCI) is considered as a formative stage of dementia and therefore its ... more Mild Cognitive Impairment (MCI) is considered as a formative stage of dementia and therefore its diagnosis can significantly assist in providing apposite treatment to the patients to impediment its headway towards dementia. In this paper, a Deep Learning approach is proposed for the classification of MCI-Converts and MCI-Non Converts, using the Structural Magnetic Resonance Imaging data. It investigates the effect of the variation in the number of filters, and the size of the filter on the performance of the model. Furthermore, the features are extracted using the penultimate layer of the proposed architecture. The Fisher Discriminant Ratio is used for the selection of features and the Support Vector Machine for the classification. The results are also compared to those obtained using the Softmax Layer. The proposed pipeline is able to extort germane features, thus improving the classification accuracy. The empirical studies exhibit the supremacy of the proposed method over the existing ones, in terms of accuracy. Consequently, the proposed technique may prove useful in the effectual diagnosis of MCI.

Title: A Combination of 3-D Discrete Wavelet Transform and 3-D Local Binary Pattern for Classific... more Title: A Combination of 3-D Discrete Wavelet Transform and 3-D Local Binary Pattern for Classification of Mild Cognitive Impairment Authors: Harsh Bhasin (i_harsh_bhasin@yahoo.com) Ramesh Agrawal (rkajnu@gmail.com) Version: 1 Date: 29 Jul 2019 Author’s response to reviews: Respected Editor, Thanks for the valuable comments. The comments were very helpful in improving the manuscript. We hope that the following responses would address the issues raised by the reviewers. Kindly let us know if any further change is required.

Advances in Intelligent Systems and Computing, 2015

The testing of a system starts with the crafting of test cases. Not all the test cases are, howev... more The testing of a system starts with the crafting of test cases. Not all the test cases are, however, equally important. The test cases can be prioritized using policies discussed in the work. The work proposes a neural network model to prioritize the test cases. The work has been validated using backpropagation neural network. 200 test cases were crafted and the experiment was carried out using 2, 5, 10, 15, and 20 layers neural network. The results have been reported and lead to the conclusion that neural network-based priority analyzer can predict the priority of a test.

ACM SIGSOFT Software Engineering Notes, 2013

Test Data Generation is the soul of automated testing. The dream of having efficient and robust a... more Test Data Generation is the soul of automated testing. The dream of having efficient and robust automated testing software can be fulfilled only if the task of designing a robust automated test data generator can be accomplished. In the work we explore the gaps in the existing techniques and intend to fill these gaps by proposing new algorithms. The following work presents algorithms that handle almost all the constructs of procedural programming languages. The proposed technique uses cellular automata as its base. The use of Cellular Automata brings a blend of artificial life to the work. The work is a continuation of our earlier attempt to amalgamate Cellular Automata based algorithms to generate test data. The technique has been applied to C programs and is currently being tested on a financial enterprise resource planning system. Since, the solution of most of the problems can be found by observing nature, we must explore artificial nature to accomplish the above task.

ACM SIGSOFT Software Engineering Notes, 2013

Manual Test Data Generation is an expensive, error prone and tedious task. Therefore, there is an... more Manual Test Data Generation is an expensive, error prone and tedious task. Therefore, there is an immediate need to make the automation of this process as efficient and effective as possible. The work presented intends to automate the process of Test Data Generation with a goal of attaining maximum coverage. A Cellular Automata system is discrete in space and time. Cellular Automata have been applied to things like designing water distribution systems and studying the patterns of migration. This fascinating technique has been amalgamated with standard test data generation techniques to give rise to a technique which generates better test cases than the existing techniques. The approach has been verified on programs selected in accordance with their Lines of Code and utility. The results obtained have been verified. The proposed work is a part of a larger system being developed, which takes into account both black box and white box testing.

BMC Medical Informatics and Decision Making, 2020

Background The detection of Alzheimer’s Disease (AD) in its formative stages, especially in Mild ... more Background The detection of Alzheimer’s Disease (AD) in its formative stages, especially in Mild Cognitive Impairments (MCI), has the potential of helping the clinicians in understanding the condition. The literature review shows that the classification of MCI-converts and MCI-non-converts has not been explored profusely and the maximum classification accuracy reported is rather low. Thus, this paper proposes a Machine Learning approach for classifying patients of MCI into two groups one who converted to AD and the others who are not diagnosed with any signs of AD. The proposed algorithm is also used to distinguish MCI patients from controls (CN). This work uses the Structural Magnetic Resonance Imaging data. Methods This work proposes a 3-D variant of Local Binary Pattern (LBP), called LBP-20 for extracting features. The method has been compared with 3D-Discrete Wavelet Transform (3D-DWT). Subsequently, a combination of 3D-DWT and LBP-20 has been used for extracting features. The r...

2014 International Conference on Soft Computing and Machine Intelligence, 2014

ABSTRACT Diploid Genetic Algorithms promise robustness as against Simple Genetic Algorithms which... more ABSTRACT Diploid Genetic Algorithms promise robustness as against Simple Genetic Algorithms which only work towards optimization. Moreover, these algorithms outperform others in dynamic environments. The work examines the theoretical aspect of the concept by examining the existing literature. The present work takes the example of Dynamic TSP to compare Greedy Approach, Genetic Algorithms and Diploid Genetic Algorithms. The work also implements a Greedy Genetic Approach for the problem. In the experiments carried out, the three variants of dominance were implemented and 115 runs proved the point that none of them outperforms other.

Test Data Generation is an intricate process which requires intensive manual labor and thus a lot... more Test Data Generation is an intricate process which requires intensive manual labor and thus a lot of project time. There is an immediate need of finding out an effective technique for automating the process as manual Test Data Generation escalates the project cost. The paper proposes the use of Artificial Life in generating and minimizing the Test Cases. The work has been applied on some programs and the initial results are encouraging. The technique makes sure that all the modules are tested in accordance with their functional specifications by the Artificial Life Test Suite Generator (ALTSG). The initial results even points to an indication of the technique being better than its counterparts.

ACM SIGSOFT Software Engineering Notes, 2014

Automated test-data generation is a convoluted task. The quality of test cases generated determin... more Automated test-data generation is a convoluted task. The quality of test cases generated determines the quality of the program under test. This paper proposes two major changes in the architecture of the automated test-data generator proposed in our earlier work. The new model of artificial-life-based test-data generation uses an s-box-based component. The earlier paper used an artificial-life based component. The component that generated black box test cases has been replaced by an s-box-based component in this paper. The test cases generated have also been encrypted using a block cipher encryption system. The encryption of test cases makes the system less prone to intrusion. This work has been done to make the system secure and prevent attacks on the proposed system by accessing the test data. The proposed model has been implemented, tested and validated using an enterprise resource planning system.

ACM SIGSOFT Software Engineering Notes, 2014

ABSTRACT Black Box Testing is immensely important because the source code of a module is not alwa... more ABSTRACT Black Box Testing is immensely important because the source code of a module is not always available. Enterprise Resource Planning systems are also tested using Black Box Testing wherein all the test cases are not equally important. The prioritization of these test cases would be helpful in case of premature termination of testing, due to lack of resources. This paper proposes a Neural Network based method to prioritize test cases. The paper also presents guidelines for prioritizing test cases. The technique has been tested using a financial management system and the results are encouraging. This paper paves way for applying Neural Network in Black Box Testing and presents a framework, which would help both researchers and practitioners.

BMC Medical Informatics and Decision Making, 2020

Background The detection of Alzheimer’s Disease (AD) in its formative stages, especially in Mild ... more Background
The detection of Alzheimer’s Disease (AD) in its formative stages, especially in Mild Cognitive Impairments (MCI), has the potential of helping the clinicians in understanding the condition. The literature review shows that the classification of MCI-converts and MCI-non-converts has not been explored profusely and the maximum classification accuracy reported is rather low. Thus, this paper proposes a Machine Learning approach for classifying patients of MCI into two groups one who converted to AD and the others who are not diagnosed with any signs of AD. The proposed algorithm is also used to distinguish MCI patients from controls (CN). This work uses the Structural Magnetic Resonance Imaging data.

Methods
This work proposes a 3-D variant of Local Binary Pattern (LBP), called LBP-20 for extracting features. The method has been compared with 3D-Discrete Wavelet Transform (3D-DWT). Subsequently, a combination of 3D-DWT and LBP-20 has been used for extracting features. The relevant features are selected using the Fisher Discriminant Ratio (FDR) and finally the classification has been carried out using the Support Vector Machine.

Results
The combination of 3D-DWT with LBP-20 results in a maximum accuracy of 88.77. Similarly, the proposed combination of methods is also applied to distinguish MCI from CN. The proposed method results in the classification accuracy of 90.31 in this data.

Conclusion
The proposed combination is able to extract relevant distribution of microstructures from each component, obtained with the use of DWT and thereby improving the classification accuracy. Moreover, the number of features used for classification is significantly less as compared to those obtained by 3D-DWT. The performance of the proposed method is measured in terms of accuracy, specificity and sensitivity and is found superior in comparison to the existing methods. Thus, the proposed method may contribute to effective diagnosis of MCI and may prove advantageous in clinical settings.

Nonlinear equations represent highly complex systems and their solutions by conventional methods ... more Nonlinear equations represent highly complex systems and their solutions by conventional methods have high computational complexity. Methods like Bisection, Regula Falsi, Newton–Raphson, Secant, Muller, etc., are used to solve such problems. This work find gaps in the existing methods and justifies the applicability of Genetic Algorithm to the problem. A Genetic Algorithm-based method has been proposed, which is more efficient and produces better results as compared to the existing methods.

Cryptanalysis refers to finding the plaintext from the given cipher text. The problem reduces to ... more Cryptanalysis refers to finding the plaintext from the given cipher text. The problem reduces to finding the correct key from a set of possible keys, which is basically a search problem. Many researchers have put in a lot of effort to accomplish this task. Most of the efforts used conventional techniques. However, soft computing techniques like Genetic Algorithms are generally good in optimized search, though the applicability of such techniques to cryptanalysis is still a contentious point. This work carries out an extensive literature review of the cryptanalysis techniques, finds the gaps there in, in order to put the proposed technique in the perspective. The work also finds the applicability of Cellular Automata in cryptanalysis. A new technique has been proposed and verified for texts of around 1000 words. Each text is encrypted 10 times and then decrypted using the proposed technique. The work has also been compared with that employing Genetic Algorithm. The experiments carried out prove the veracity of the technique and paves way of Cellular automata in cryptanalysis. The paper also discusses the future scope of the work.

ACM SIGSOFT Software Engineering Notes, 2014

ABSTRACT Manual test data generation is carried out by using the ability of neurons to recognize ... more ABSTRACT Manual test data generation is carried out by using the ability of neurons to recognize patterns. The nervous system and the brain coordinate to generate test cases, which are capable of finding potential faults. Automated test data generators lack the ability to produce efficient test cases because they do not imitate natural processes. This paper proposes using Artificial Life based systems for generating test cases. Cellular Automata and Langton's loop have been used to accomplish the above task. Cellular Automata are parallel distributed systems capable of reproducing using self generated patterns. These fascinating techniques have been amalgamated with standard test data generation techniques to give rise to a methodology, which generates test cases for white box testing. Langton's Loops have been used to generate test cases for Black Box Testing. The approach has been verified on a set of 20 programs. The programs have been selected on the basis of their Lines of Code and utility. The results obtained have been verified using Average Probability of Fault Detection. This paper also proposes a new framework capable of crafting test cases taking into account the oracle cost.

Oxford University Press, 2015

The book has been divided into four sections: Algorithm Basics, Data Structures, Design Techniques and Advanced Topics. The first section explains the importance of algorithms, growth of functions, recursion and analysis of algorithms. The second section covers the data structures basics, trees, graphs, sorting in linear and quadratic time. Section three discusses the various design techniques namely, divide and conquer, greedy approach, dynamic approach, backtracking, branch and bound and randomized algorithms used for solving problems in separate chapters. The fourth section includes the advanced topics such as transform and conquer, decrease and conquer, number thoeretics, string matching, computational geometry, complexity classes, approximation algorithms, and parallel algorithms. Finally, the applications of algorithms in Machine Learning and Computational Biology areas are dealt with in the subsequent chapters. This section will be useful for those interested in advanced courses in algorithms.

The book also has 10 appendixes which include topics like probability, matrix operations, Red-black tress, linear programming, DFT, scheduling, a reprise of sorting, searching and amortized analysis and problems based on writing algorithms.

The concepts and algorithms in the book are explained with the help of examples which are solved using one or more methods for better understanding. The book includes variety of chapter-end pedagogical features such as point-wise summary, glossary, multiple choice questions with answers, review questions, application-based exercises to help readers test their understanding of the learnt concepts.

he testing of a system starts with the crafting of test cases. Not all the test cases are, howeve... more he testing of a system starts with the crafting of test cases. Not all the test cases are, however, equally important. The test cases can be prioritized using policies discussed in the work. The work proposes a neural network model to prioritize the test cases. The work has been validated using backpropagation neural network. 200 test cases were crafted and the experiment was carried out using 2, 5, 10, 15, and 20 layers neural network. The results have been reported and lead to the conclusion that neural network-based priority analyzer can predict the priority of a test.

Soft Computing, Springer

Diploid Genetic Algorithms promise robustness as against Simple Genetic Algorithms which only wor... more Diploid Genetic Algorithms promise robustness as against Simple Genetic Algorithms which only work towards optimization. Moreover, these algorithms outperform others in dynamic environments. The work examines the theoretical aspect of the concept by examining the existing literature. The present work takes the example of Dynamic TSP to compare Greedy Approach, Genetic Algorithms and Diploid Genetic Algorithms. The work also implements a Greedy Genetic Approach for the problem. In the experiments carried out, the three variants of dominance were implemented and 115 runs proved the point that none of them outperforms other.

AI & Society, Springer

The heuristic search processes like Simple Genetic Algorithms helps in achieving optimization but... more The heuristic search processes like Simple Genetic Algorithms helps in achieving optimization but do not guarantee robustness so there is an immediate need of a machine learning technique which also promises robustness. Diploid Genetic Algorithms ensure consistent results and can therefore replace Simple Genetic Algorithms in applications like Test Data Generation and Regression Testing where robustness is more important. However, there is a need to review the work that has been done so far in the field. It is also important to analyze the applicability of the premise of the dominance techniques applied so far in order to implement the technique. The work presents a systematic review of Diploid Genetic Algorithms, examines the premise of the dominance relation used in different works and discusses the future scope. The work also discusses the biological basis of evaluating dominance. The work is important as the future of machine learning relies on techniques that are robust as well as efficient.

2014 Intl. Conference on SoftComputing and Machine Intelligence, Sep 27, 2014

Diploidity is the essence of the nature. However, it has largely been ignored by the computer sci... more Diploidity is the essence of the nature. However, it has largely been ignored by the computer science fraternity. Simple Genetic Algorithms and their variants have extensively been used in solving NP hard problems in-spite of the fact that Diploid Genetic Algorithms assure robustness as against Simple Genetic Algorithms which solitary guarantee optimization. Moreover, the past endeavors proved that these algorithms are more successful in dynamic environments as compared to their haploid counterpart. The work proves the above point by applying Diploid genetic Algorithms to Dynamic Travelling Salesman Problem and comparing the results to Greedy Approach and Simple Genetic Algorithms. The works also presents a hybrid approach namely Greedy Genetic Approach. The results of the experiments established that diploidity ensures robustness. In the experiments carried out, the three variants of dominance were implemented and 115 trials bought forth the point that though Haploid and Greedy Approaches do not outperform the other, Diploid are the best bet for dynamic environments.

Alzheimer's & Dementia, 2022

2021 2nd Asia Conference on Computers and Communications (ACCC), 2021

Advances in Intelligent Systems and Computing, 2015

ACM SIGSOFT Software Engineering Notes, 2013

BMC Medical Informatics and Decision Making, 2020

2014 International Conference on Soft Computing and Machine Intelligence, 2014

ACM SIGSOFT Software Engineering Notes, 2014

BMC Medical Informatics and Decision Making, 2020

Background The detection of Alzheimer’s Disease (AD) in its formative stages, especially in Mild ... more Background
The detection of Alzheimer’s Disease (AD) in its formative stages, especially in Mild Cognitive Impairments (MCI), has the potential of helping the clinicians in understanding the condition. The literature review shows that the classification of MCI-converts and MCI-non-converts has not been explored profusely and the maximum classification accuracy reported is rather low. Thus, this paper proposes a Machine Learning approach for classifying patients of MCI into two groups one who converted to AD and the others who are not diagnosed with any signs of AD. The proposed algorithm is also used to distinguish MCI patients from controls (CN). This work uses the Structural Magnetic Resonance Imaging data.

ACM SIGSOFT Software Engineering Notes, 2014

Oxford University Press, 2015

Soft Computing, Springer

AI & Society, Springer

2014 Intl. Conference on SoftComputing and Machine Intelligence, Sep 27, 2014

Python for Beginners is a textbook designed for the undergraduate and postgraduate students of Co... more Python for Beginners is a textbook designed for the undergraduate and postgraduate students of Computer Science Engineering (CSE), Information Technology (IT), and computer applications. It helps the students to understand the fundamentals and applications of Python. The book will also serve as a useful reference for researchers and practising programmers who intend to pursue a career in machine learning. The book is also indented for students preparing for interviews and for programmers who intend to switch to Python. It covers both basic and intermediate levels.

Algorithms: Design and Analysis of is a textbook designed for the undergraduate and postgraduate ... more Algorithms: Design and Analysis of is a textbook designed for the undergraduate and postgraduate students of computer science engineering, information technology, and computer applications. It helps the students to understand the fundamentals and applications of algorithms. The book will serve as a useful reference for researchers and practising programmers in the field of algorithm designing. It is also indented for students preparing for interviews and competitive examinations.

The book has been divided into four sections: Algorithm Basics, Data Structures, Design Techniques and Advanced Topics. The first section explains the importance of algorithms, growth of functions, recursion and analysis of algorithms. The second section covers the data structures basics, trees, graphs, sorting in linear and quadratic time. Section three discusses the various design techniques namely, divide and conquer, greedy approach, dynamic approach, backtracking, branch and bound and randomized algorithms used for solving problems in detail in separate chapters. The fourth section includes the advanced topics such as transform and conquer, decrease and conquer, number thoeretics, string matching, computational geometry, complexity classes, approximation algorithms, and parallel algorithms. Finally, the applications of algorithms in Machine Learning and Computational Biology areas are dealt with in the subsequent chapters. This section will be useful for those interested in advanced courses in algorithms. Appendixes of the book include topics such as probability, matrix operations, Red-black tress, linear programming, DFT, scheduling, a reprise of sorting, searching and amortized analysis, and problems based on writing algorithms.

The concepts and algorithms in the book are explained with the help of examples which are solved using more than one method for better understanding. Each chapter of the book includes a variety of end-chapter exercises in the form of MCQs with answers, review questions, and programming exercises to help readers test their knowledge.

rogramming in C# is a textbook designed for the undergraduate and postgraduate students of comput... more rogramming in C# is a textbook designed for the undergraduate and postgraduate students of computer science engineering, information technology, and computer applications. It helps the students to understand the fundamentals and applications of C# programming using .NET Framework. The book will also serve as a useful reference for researchers and practising programmers who intend to pursue a career in C# programming.

Broadly divided into three parts, the first part of the book serves as an introduction to the .NET Framework and C# and procedural programming, followed by the second part discussing object-oriented programming concepts, and the third part focusing on component object model (COM) and advanced topics. Part one, beginning with an introduction to the .NET Framework and C#, goes on to discuss data types and operators, conditional statements, loops, collections, strings, arrays, structure and enumerations providing a thorough coverage of procedural programming concepts. The second part, object-oriented programming, starts with the fundamentals of classes and objects, inheritance, interfaces and then discusses topics such as operator overloading, error and exception handling, generics, and threads. The third part deals with topics related to COM, namely, Windows Form and controls, menus and common dialogs, and also advanced topics such as data connectivity, ASP.NET, networking, deployment, and Windows Presentation Foundation (WPF) and Windows Communication Foundation (WCF).

Each chapter includes a variety of end-chapter exercises in the form of MCQs with answers, review questions, and programming exercises to help readers test their knowledge.