Qualification: 
Ph.D, MCA
s_thangavel@cb.amrita.edu

Dr. Thangavelu S. currently serves as Assistant Professor at the Department of Computer Science and Engineering, School of Engineering, Amrita Vishwa Vidyapeetham, Coimbatore Campus. His areas of research include Evolutionary Algorithms and Database Technologies.

Publications

Publication Type: Conference Paper

Year of Publication Title

2020

V. SandhyaSree and Dr. Thangavelu S., “Performance Analysis of Differential Evolution Algorithm Variants in Solving Image Segmentation”, in Computational Vision and Bio-Inspired Computing, Cham, 2020.[Abstract]


Image segmentation is an activity of dividing an image into multiple segments. Thresholding is a typical step for analyzing image, recognizing the pattern, and computer vision. Threshold value can be calculated using histogram as well as using Gaussian mixture model. but those threshold values are not the exact solution to do the image segmentation. To overcome this problem and to find the exact threshold value, differential evolution algorithm is applied. Differential evolution is considered to be meta-heuristic search and useful in solving optimization problems. DE algorithms can be applied to process Image Segmentation by viewing it as an optimization problem. In this paper, Different Differential evolution (DE) algorithms are used to perform the image segmentation and their performance is compared in solving image segmentation. Both 2 class and 3-class segmentation is applied and the algorithm performance is analyzed. Experimental results shows that DE/best/1/bin algorithm out performs than the other variants of DE algorithms

More »»

2019

Dr. Thangavelu S., S., A., Naetra, K. C., A.C., K. S., and Lasya, V., “Feature Selection in Cancer Genetics using Hybrid Soft Computing”, in 2019 Third International conference on I-SMAC (IoT in Social, Mobile, Analytics and Cloud) (I-SMAC), 2019.[Abstract]


Microarray databases are the most frequently used datasets for cancer analytics. Microarray databases are characterized by the presence of a very large number of genes, which exceeds the very little number of samples. So, the feature set accumulates the curse of dimensionality. Therefore, selecting a small subset of genes among thousands of genes in microarray data can potentially increase the accuracy for the classification of cancer. Many approaches, from the field of classical machine learning and soft computing, have been used to address the issue of feature selection and feature extraction for better classifications and clustering accuracy. The research outlined in this paper strives to look at a two-stage approach using minimum Redundancy Maximum Relevancy (mRMR), a feature ranking framework as the first stage followed by a hybrid genetic algorithm in the second stage that works on the features ranked by the mRMR. The proposed method is aimed to select the optimal feature subsets for better classification results in binary and multi class datasets to compensate for the curse of dimensionality in microarray datasets. The classifiers used to test the two-stage proposition are SVM, Naive-Bayes, Linear Discriminant Analysis, decision trees and random forest classifiers. The experimental results show that the gene subset selected by the mRMR-GA pipeline gives good results.

More »»

2019

S. S. Shinde, Dr. Thangavelu S., and Dr. Jeyakumar G., “Evolutionary Computing Approaches for Solving Multi-Objective and Many-Objective Optimization Problems: A Review”, in 2019 5th International Conference On Computing, Communication, Control And Automation (ICCUBEA), 2019.[Abstract]


In today world, there exists different algorithmic approaches in Computer Science to solve the complex real-world optimization problems around us. Based on the number of objectives they are classified as single-objective, multi-objective or many-objective optimization problems. Though many approaches are available for solving single-objective optimization problems, they are not directly scalable to solve Multi-objective and Many-objective optimization problems. However, the Evolutionary Computing (EC) based approaches are becoming most popular and common to solve all the types of optimization problems, due to their simplicity and wide applicability. This paper presents the details of the Evolutionary Algorithms (EAs) to solve the optimization problem with the category of EAs for Multi-objective and EAs for Many-Objective optimization problems. Insight about the Hybridization of EAs is also added. The details about the algorithmic components required for designing EAs is also presented along with the Benchmarking Problems and Performance Metrics available to validate the newly designed EAs.

More »»

2019

Dr. Thangavelu S., Rao, V., Shyamala, C. K., and Velayutham, C. S., “Introductory Programming Using Non-Textual Modalities - An Empirical Study on Skill Assessment Using Rainfall Problem”, in 2019 IEEE Global Engineering Education Conference (EDUCON), 2019, vol. April-2019, pp. 867-871.[Abstract]


This paper presents an analysis of programming skills of first semester undergraduate computer science & engineering students using Rainfall programming problem. 345 students, in a 2 hour laboratory session, solved the Rainfall programming problem using either (block-based) Scratch or (flowchart-based) Flowgorithm tool. The success rate, in terms of the number of correctly implemented sub-goals, of the students has been analyzed. The possible influence of factors such as the non-texual modality tool, gender difference, school board from which students had their higher secondary education and prior computing experience in their higher secondary stage has been presented. The possible factors for difficulty towards conceiving and implementing the sub-goals has been indicated based on the observations while evaluating the student programs.

More »»

2017

K. C. Haritha, Dr. Jeyakumar G., and Dr. Thangavelu S., “Image fusion using evolutionary algorithms: A survey”, in 2017 4th International Conference on Advanced Computing and Communication Systems, ICACCS 2017, Sri Eeshwar College of Engineering, 2017.[Abstract]


This paper is mainly intended to compare image fusion method using different evolutionary algorithms and a comparison between these methods. The survey focuses on region-based fusion techniques, which is a major area of research. The paper compares image fusion processes using various evolutionary algorithms and illustrates the advantages and disadvantages of these algorithms. This survey illustrates that a method of image fusion can also be included in the DE optimization stage with the block size optimization. Finally, it is concluded that spatial frequency can be used as the sharpness criterion and Evolutionary algorithms perform better in block size optimization. © 2017 IEEE.

More »»

Publication Type: Journal Article

Year of Publication Title

2020

Dr. Shriram K Vasudevan, M, V. Krishna Ki, Sini Raj P., and Dr. Thangavelu S., “AI approach with increased accuracy to extract the tabular content from PDF and Image files”, International Journal of Advanced Trends in Computer Science and Engineering, vol. 9, pp. 1013-1019, 2020.[Abstract]


In today's era of computerized banking, management, billings and what not, we use tabular data in every sector. The most commonly used format of storing tabular data by us is through excel format. It is very easy to retrieve information from excel sheets. But, tabular data extraction from PDFs or images has remained as an inherent problem since many years. In order to reduce such issues and automate the process we have designed a system using artificial intelligence that can take a PDF or an image as an input and outputs a CSV or excel file directly with the extracted tabular data.

More »»

2020

R. Giridhararajan, Vasudevan, S. K., and Dr. Thangavelu S., “IoT Based Approach for the Increased and Improved sales for the Brick and Mortar Stores”, International Journal of Advanced Trends in Computer Science and Engineering, vol. 9, pp. 3048-3052, 2020.[Abstract]


The consistent use of smartphones in daily life has brought the wave of online shopping and digital marketing into action. Over the last few years, brick-and-mortar store owners have increasingly seen consumers migrate away from retail stores in favor of convenient and quickly accessible digital outlets. Many shoppers feel that, with the ease of smartphone and "one-click" shopping, browsing for products in a physical store is almost obsolete. However, what if the very same technology that caused this crisis in retail industry could be used to drive customers into their stores? Bluetooth Low Energy (BLE) beacons provide a great way to guide customers through a larger store and find the intended product. In addition to customer convenience, beacons are a powerful tool for storeowners to deliver targeted product recommendations. Thus, the overall solution consists of two components - Proximity sensing and Product recommendation systems. When a customer comes into the proximity of a particular section of a store, the recommender system generates total and personal recommendation for the customer. The matching entries between recommended products and products from the particular section are the final optimized recommendation delivered to the customer via mobile application. Product recommendations are powered by the purchase history data of Future group stores. That way a new lease of life can be brought to struggling brick-and-mortar stores.

More »»

2019

S. S. Shinde, K., D., Dr. Thangavelu S., and Dr. Jeyakumar G., “Multi-objective evolutionary algorithm based approach for solving rfid reader placement problem using weight-vector approach with opposition-based learning method”, International Journal of Recent Technology and Engineering, vol. 7, pp. 177-184, 2019.[Abstract]


For smart building applications, identifying and tracking the objects and people in and around a building is an inevitable problem. There exist many approaches for solving this problem. Nowadays, the RFID network based approaches have become most popular for its speed and accuracy. However, placing the RFID readers at optimal places in a building to cover all the areas in order to identify and track the objects and people is a cumbersome task. This paper proposes a model in which the RFID reader placement problem is formulated as a multi-objective optimization problem and also proposes an algorithmic framework to solve the same. The proposed algorithmic frame work consists of a multi-objective Differential Evolution algorithm which adds weights to each of the objective and also follows the opposition-based learning approach for initializing the populations. The results obtained in solving the RFID reader placement problem with proposed algorithmic framework is studied and reported in details for individual objectives, combined objectives with different schemes and for two different population initialization techniques.

More »»

2019

D. Vasumathi and Dr. Thangavelu S., “Scalarizing functions in solving multi-objective problem-an evolutionary approach”, Indonesian Journal of Electrical Engineering and Computer Science, vol. 13, no. 3, pp. 974-981, 2019.[Abstract]


Scalarizing functions had long been observed for optimization of multi-objective problems. Scalarizing functions on multi-objective problem along with Differential Evolution (DE) algorithm variants had been used to analyze the effect of scalarizing functions. The main purpose is to find the better scalarizing function which can be applied for optimization. The effective solution of the multi-objective problem depends on the various factors like the DE algorithm and the scalarizing functions used. Multi objective evolutionary algorithm (MOEA) framework in java had been used for performing the analysis. The Obtained results showed that Tchebysheff scalarization function performs better than the other scalarizing functions on various indicator functions used.

More »»

2018

A. Gandhi, Kumari, V., and Dr. Thangavelu S., “A Comprehensive Study and Analysis of LEACH and HEED Routing Protocols for Wireless Sensor Networks – With Suggestion for Improvements”, Indonesian Journal of Electrical Engineering and Computer Science, vol. 9, no. 3, pp. 778 - 783, 2018.[Abstract]


The main interesting aspect of the digital era is the widely spread ease of communication from one end of the world to the other end of the world. There is a revolution in communication, digitalization, globalization, video calling, wireless data transfer and this is possible due to networking. Initially computer networks is the data sharing where data such as documents, file, reports, presentation files, videos, images etc can be shared within a local network or remotely connected networks. Traditional data networking is to empower end-to-end information transfer. The data in such networks are carried across point-to-point links and the intermediate nodes just forward the packets, where the payload of the packets is not modified. Traditional LANs need wires, which may be difficult to set up in some situations.It is very much understandable and clearly visible that wired communication is being completely overtaken by wireless technologies in the recent past. Wireless LANs, by its very nature, empowers with increased mobility and flexibility. Wi-Fi devices get connected to the internet through WLAN and access points. 2.4 GHz and 5 GHz ISM bands are used by Wi-Fi. Also, it is to be understood that, a wireless adhoc network is distributed in its nature. It is also to be noted that, the adhoc nature makes these network to rely on any of the pre-existing infrastructure. The data forwarding shall happen from the nodes very much dynamically based on the connectivity and the routing algorithm used.

More »»

2018

K. C. Haritha and Dr. Thangavelu S., “Multi focus region-based image fusion using differential evolution algorithm variants”, Lecture Notes in Computational Vision and Biomechanics, vol. 28, pp. 579-592, 2018.[Abstract]


This work focuses on an optimum process of image fusion on multiple focus images using an optimization algorithm viz., Differential Evolution (DE) algorithm. The input image is divided into regions and sharper regions are selected from these two images. The selected clear blocks are used for constructing final resultant image. The main purpose of using differential evolution algorithm is to find out optimum block size, which is more useful during division of image rather than fixed block size. And also, this work compares different variants of differential evolution algorithm based image fusion to find out which one will be suitable for getting more focused image. The major focus of the research is finding out which type of differential evolution algorithm is best suitable for almost all type of images. Block based and pixel based method are used together to achieve a better resultant image. Performance of fused image is calculated using image quality measures and found out better fusion method, which can be used in almost all situations. © 2018, Springer International Publishing AG.

More »»

2017

A. Sn, Dr. Shriram K Vasudevan, Prashant R. Nair, Dr. Thangavelu S., and Rmd, S., “A proposal for mitigating fishermen killing in indian sea borders through technology – maritime boundary identification device”, Indonesian Journal of Electrical Engineering and Computer Science, vol. 6, pp. 704-710, 2017.[Abstract]


The Tamil Nadu – Sri Lanka maritime boundary has been responsible for frequent controversies in the global front, due to fishermen from Rameshwaram and Ramanthapuram districts recklessly straying past Indian waters. Instances of Indian fishermen being captured and killed by the Sri Lankan navy have spun vivid images of violence and human rights violation among the masses. Social activists are desperate for an automatic alarm system to warn the fishermen when they are about to cross the border, and avert a possible impending crisis. The following discussion focuses on the design of an alarm signal system that could alert the fishermen on a periodic basis as they approach closer to the maritime boundary. Installing Global Positioning System (GPS) devices would pose several economic challenges. Instead of the conventional approach of using GPS devices to track location, a transmitter – receiver system exquisitely designed to send signals to the boat would be effective. This system helps in keeping a continuous track of the boats. It provides a reliable solution to alert fishermen before they could inadvertently exceed the boundary. © 2017 Institute of Advanced Engineering and Science. All rights reserved.

More »»

2015

A. V., Praveen, V., and Dr. Thangavelu S., “Performance Analysis of Variants of Differential Evolution on Multi-Objective Optimization Problems”, Indian Journal of Science and Technology, vol. 8, no. 17, pp. 1-6, 2015.[Abstract]


Differential Evolution (DE) algorithm is a stochastic search algorithm, applied to solve various optimization problems. Different DE variants such as rand/1/bin, best/1/bin, rand/2/bin, best/2/bin, etc. are existed in the literature and many comparative performance analyses among these DE variants in solving different single-objective optimization problems were already done by many researchers. But many real world applications are of the category of multi-objective optimization problems. Only minimal amount of research work has been found in the literature on the performance analysis of these DE variants on solving multi-objective kind of problems. In this paper, we analyze the performance of the Differential Evolution (DE) variants to solve Multi-objective Optimization Problems (MOP). We have chosen five multi-objective benchmark functions called ZDT test functions that are grouped by characteristics like convex, non-convex, non-uniform, discrete and low density pareto fronts. We used the DE variants of type DE/rand/1/*, DE/rand/2/*, DE/best/1/*, DE/best/2/* and DE/rand-to-best/1/*, where * represents the binomial/exponential crossover operation, to test the five ZDT functions. The performance analysis among these variants on multi-objective functions are performed based on the convergence and diversity nature of the solutions and analyzed using metrics called Convergence Metric (Cm) and Diversity Metric (Dm). The results show that the DE variants rand/1/bin and best/1/bin have the better performance in terms of Convergence and Diversity in the solution space in solving the above mentioned test functions. It is also possible to do the further analysis on these variants by applying them in parallel (i.e. more than one variant/algorithm is used to solve the problem) and by exchanging the information among them, to improve the solution.

More »»

2015

Dr. Thangavelu S. and Dr. Shunmuga Velayutham C., “Combining different differential evolution variants in an island based distributed framework–an investigation”, Advances in Intelligent Systems and Computing, vol. 320, pp. 593-606, 2015.[Abstract]


This paper proposes to combine three different Differential Evolution (DE) variants viz. DE/rand/1/bin, DE/best/1/bin and DE/rand-to-best/1/bin in an island based distributed Differential Evolution (dDE) framework. The resulting novel dDEs with different DE variants in each islands have been tested on 13 highdimensional benchmark problems (of dimensions 500 and 1000) to observe their performance efficacy as well as to investigate the potential of combining such complementary collection of search strategies in a distributed framework. Simulation results show that rand and rand-to-best strategy combination variants display superior performance over rand, best, rand-to-best as well as best, rand-to-best combination variants. The rand and best strategy combinations displayed the poor performance. The simulation studies indicate a definite potential of combining complementary collection of search characteristics in an island based distributed framework to realize highly co-operative, efficient and robust distributed Differential Evolution variants capable of handling a wide variety of optimizations tasks. © Springer International Publishing Switzerland 2015.

More »»

2015

Dr. Thangavelu S. and Dr. Shunmuga Velayutham C., “An investigation on mixing heterogeneous differential evolution variants in a distributed framework”, International Journal of Bio-Inspired Computation, vol. 7, pp. 307-320, 2015.[Abstract]


This paper attempts a preliminary investigation to gain insight about the cooperative dynamics of mixing the four classical differential evolution (DE) variants viz. DE/rand/1/bin, DE/best/1/bin, DE/rand/2/bin and DE/best/2/bin in an island-based distributed framework. The exhaustive combinations of the above said four DE variants in an island size of 4, resulting in 35 distributed DE variants, have all been implemented and tested on 14 unconstrained test functions with diverse features grouped by their modality and decomposability. Simulation results show that the rand-best variants' mixing, display a better cooperative characteristics than rand-rand and best-best variants' mixing. This insight motivated for further investigations on mixing DE/rand-to-best/1/bin (a variant which intrinsically employs rand and best strategies) with DE/rand/1/bin and DE/best/1/bin in the distributed framework. Simulation results reiterated the observations about the cooperative characteristics of rand-best variants' combinations with the latter mixing showing still better cooperative characteristics both in terms of probability of convergence and convergence rate. Copyright © 2015 Inderscience Enterprises Ltd.

More »»

2015

Dr. Thangavelu S., Dr. Jeyakumar G., and Dr. Shunmuga Velayutham C., “Population variance based empirical analysis of the behavior of differential evolution variants”, Applied Mathematical Sciences, vol. 9, pp. 3249-3263, 2015.[Abstract]


Differential Evolution (DE) is a simple but efficient Evolutionary Algorithm (EA) for stochastic real parameter optimization. With various types of mutation and crossover applicable to DE, there exist many variants of DE. The empirical comparisons between the performances of these variants on chosen benchmarking problems are well reported in literature. However, attempts to analyze the reason for such identified behavior of the variants are scarce. As an attempt in this direction, this paper empirically analyzes the performance as well as the reason for such performance of 14 classical DE variants on 4 benchmarking functions with different modality and decomposability. The empirical analysis is carried out by measuring the mean objective function values (MOV), success rate (Sr), probability of convergence (Pc), quality measure (Qm) and empirical evolution of the variance of the population (Evar). The study also includes reporting evidences for the variants suffering with stagnation and/or premature convergence. © 2014 S. Thangavelu, G. Jeyakumar and C. Shunmuga Velyautham.

More »»

2010

Dr. Thangavelu S. and Dr. Shunmuga Velayutham C., “Taguchi method based parametric study of generalized generation gap genetic algorithm model”, Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 6466 LNCS, pp. 344-350, 2010.[Abstract]


In this paper, a parametric study of Generalized Generation Gap (G3) Genetic Algorithm (GA) model with Simplex crossover (SPX) using Taguchi method has been presented. Population size, number of parents and offspring pool size are considered as design factors with five levels. The analysis of mean factor is conducted to find the influence of design factors and their optimal combination for six benchmark functions. The experimental results suggest more experiments on granularity of design factor levels for better performance efficacy. © 2010 Springer-Verlag. More »»

Publication Type: Conference Proceedings

Year of Publication Title

2015

Dr. Thangavelu S., Dr. Jeyakumar G., Balakrishnan, R. M., and Dr. Shunmuga Velayutham C., “Theoretical Analysis of Expected Population Variance Evolution for a Differential Evolution Variant”, Computational Intelligence in Data Mining (In Smart Innovation, Systems and Technologies), vol. 32. Smart Innovation, Systems and Technologies, Springer, pp. 403–416, 2015.[Abstract]


In this paper we derive an analytical expression to describe the evolution of expected population variance for Differential Evolution (DE) variant—DE/current-to-best/1/bin (as a measure of its explorative power). The derived theoretical evolution of population variance has been validated by comparing it against the empirical evolution of population variance by DE/current-to-best/1/bin on four benchmark functions.

More »»
Faculty Research Interest: