Deepti Mehrotra

A comparative review of approaches to prevent premature convergence in GA

Detailed discussion on various approaches for handling premature convergence in GA.Theoretical framework is presented for convergence analysis of GA.Strengths and weaknesses of each approach are provided.Summary and comparison of the approaches is given for quick review. This paper surveys strategies applied to avoid premature convergence in Genetic Algorithms (GAs). Genetic Algorithm belongs to the set of nature inspired algorithms. The applications of GA cover wide domains such as optimization, pattern recognition, learning, scheduling, economics, bioinformatics, etc. Fitness function is the measure of GA, distributed randomly in the population. Typically, the particular value for each gene start dominating as the search evolves. During the evolutionary search, fitness decreases as the population converges, this leads to the problems of the premature convergence and slow finishing. In this paper, a detailed and comprehensive survey of different approaches implemented to prevent premature convergence with their strengths and weaknesses is presented. This paper also discusses the details about GA, factors affecting the performance during the search for global optima and brief details about the theoretical framework of Genetic algorithm. The surveyed research is organized in a systematic order. A detailed summary and analysis of reviewed literature are given for the quick review. A comparison of reviewed literature has been made based on different parameters. The underlying motivation for this paper is to identify methods that allow the development of new strategies to prevent premature convergence and the effective utilization of genetic algorithms in the different area of research.

User Centric Retrieval of Learning Objects in LMS

Research and Academic Institutions own and archive a great number of documents like lesson plan, study material and research related resources, which are needed to be stored and used over for a longer period of time by lecturers and researchers. In order to achieve this it is required to convert these educational resources into Learning Objects and store them in structured & meaningful way via a learning management system (LMS) thus enriching classical teaching. Also with enhancement of e-learning environment there is a great need of managing the LMS repositories by storing information resources as Learning object, a digital entity which can be used in electronic learning environment. These learning objects are stored in repositories and are managed by Learning Management Systems. It aids teaching and learning process and helps in communications between users. Many designs of LMS are non user-centric and has limited capabilities in delivering user preferred learning material. Searching through keywords or metadata of learning material will result in display of huge quantity of information. Thus there is an earnest need to identify the techniques that can provide more efficient mechanism for information retrieval. Recommendation techniques have shown to be successful in many domains (e.g. movies, books, music, etc.). Thus there is a need to deploy a recommending system in the E-Learning domain to extend the functionality of standard-based learning management systems with providing the user based retrieval. In this paper a model is being proposed that can enhance the search and delivery of a relevant learning object based on his/her preferences and further ranking & clustering of learning objects are done through K-Mean and Self Organising Maps for a personalised learning environment.

Quality metrics a quanta for retrieving learning object by clustering techniques

E-learning today has shown exponential growth as it provides the potential to provide right information to the right people at right time and place, using the right medium. The atomic unit of any e-learning environment is a Learning object, a digital entity which can be used in electronic learning environment. These learning objects are stored in repositories and are managed by Learning Management Systems. However, the exponential availability of information leads to a difficult scenario like finding a particular educational resource for a learner, based on the context or based on his/her preferences. The searching through keywords or metadata will result in display of huge quantity of information. Thus there is an earnest need to identify the techniques that can provide more efficient mechanism for information retrieval. In this paper a model is being proposed that can enhance the search and delivery of a relevant learning object to a learner using quality metrics & clustering of learning objects through Self Organising Maps.

Genetic algorithms: concepts, issues and a case study of grammar induction

This paper discusses a case study of grammar induction. Grammar induction is the process of learning grammar from a set of training data of the positive (S+) and negative (S-) strings. An algorithm has been designed and implemented for the induction of context free grammar (CFG). Special bit mask oriented data structures have been used to apply the crossover and mutation operations. The aim is to establish the applicability of the genetic algorithms (GAs) for different engineering problems. The paper lays a concrete foundation to formulating problems in the genetic algorithm framework. In addition, the basic principles of standard genetic algorithm, such as encoding techniques, selection techniques, operators (crossover and mutation), and the issues raised in the relevant literature have been discussed to establish the applicability of the genetic algorithm.

Comparative review of selection techniques in genetic algorithm

This paper compares various selection techniques used in Genetic Algorithm. Genetic algorithms are optimization search algorithms that maximize or minimizes given functions. Indentifying the appropriate selection technique is a critical step in genetic algorithm. The process of selection plays an important role in resolving premature convergence because it occurs due to lack of diversity in the population. Therefore selection of population in each generation is very important. In this study, we have reported the significant work conducted on various selection techniques and the comparison of selection techniques.

Evaluation of Genetic Algorithm’s Selection Methods

The focus of this paper is towards analyzing the performance of various selection methods in genetic algorithm. Genetic algorithm, a novel search and optimization algorithm produces optimum response. There exist different selections method available—plays a significant role in genetic algorithm performance. Three selection methods are taken into consideration in this study on travelling salesman problem. Experiments are performed for each selection methods and compared. Various statistical tests (F-test, Posthoc test) are conducted to explain the performance significance of each method.

An IOT by information retrieval approach: Smart lights controlled using WiFi

In recent years, the rapid development of Internet of Technology (IOT) makes the intelligent home come true as people expect. The intelligent home system creates the more comfortable, safer, humane and intelligent living environment. It can resolve the problems facing by the people who have busy schedules and get a very less amount of time to spend at home which is increasing rapidly around the world. For the solution of this problem, user can depend on the automated machines and gadgets like smart phones. These smart gadgets are using cloud computing which sends and receives signal o the cloud. The data that is of our use can be fetched by matching some key values using the concept of information retrieval. The key objective of this paper is to create a full-fledged application which could let user to operate the lights of their house from any remote location. The user have a list of options to select which light is to be on and when. The only requirement is to have working wifi at home to which the lights are connected. The is developed in Lua Language by using the Esplorer Integrated Development Environment (IDE). We have also used the micro controller chip ESP 8266 to build our board.

Usability evaluation of website using combined weighted method: fuzzy AHP and entropy approach

The usability of the website is a topic of great interest for researchers. The fuzzy MCDM approaches are widely adapted to measure and rank the usability of the website. These measures incorporate human input towards website are subjective in nature. Along with these measures, design dimension and other support features that encourage the user to reach desired information in stipulated time are equally important and should be incorporated in the proposed metric. These functions are objective in nature and hence suitable mathematical model or formulae can be used to evaluate its crisp value. In this paper a novel metric is proposed integrating objective and subjective approach, evaluating usability using fuzzy AHP and entropy method respectively. The proposed metric can be considered more accurate and complete as it includes expert input, experience of end user, design dimension and support features.

Delivery of learning knowledge objects using fuzzy clustering

Abstracte-Learning industry is rapidly changing and the current learning trends are based on personalized, social and mobile learning, content reusability, cloud-based and talent management. The learning systems have attained a significant growth catering to the needs of a wide range of learners, having different approaches and styles of learning. Objects delivered by these systems should provide a variety of learning content to satisfy different learners and should also have a pedagogical value than simple course content to empower learning. The Knowledge Objects of Knowledge Management Systems can be combined and delivered with existing Learning Objects of Learning Management System to provide better and more holistic user experience. Choosing a suitable object in accordance with learner category is a complex task. The paper encompasses data mining approach, fuzzy clustering technique to combine Learning and Knowledge objects based on attributes of metadata. These objects are further mapped with various learning styles and an appropriate set of objects are delivered to the learners. Thus, a personalized and more authentic learning experience is achieved emphasizing the content reusability.

Grammar induction using bit masking oriented genetic algorithm and comparative analysis

A background on theory of grammar induction is presented.The effect of premature convergence is discussed in detail.Proposed a system for grammar inference by utilizing the mask-fill reproduction operators and Boolean based procedure with minimum description length principle.Comparative analysis, discussion and observation of obtained results are given in an effective manner.Statistical tests (F-test and post hoc test) are conducted. This paper presents bit masking oriented genetic algorithm (BMOGA) for context free grammar induction. It takes the advantages of crossover and mutation mask-fill operators together with a Boolean based procedure in two phases to guide the search process from ith generation to (i+1)th generation. Crossover and mutation mask-fill operations are performed to generate the proportionate amount of population in each generation. A parser has been implemented checks the validity of the grammar rules based on the acceptance or rejection of training data on the positive and negative strings of the language. Experiments are conducted on collection of context free and regular languages. Minimum description length principle has been used to generate a corpus of positive and negative samples as appropriate for the experiment. It was observed that the BMOGA produces successive generations of individuals, computes their fitness at each step and chooses the best when reached to threshold (termination) condition. As presented approach was found effective in handling premature convergence therefore results are compared with the approaches used to alleviate premature convergence. The analysis showed that the BMOGA performs better as compared to other algorithms such as: random offspring generation approach, dynamic allocation of reproduction operators, elite mating pool approach and the simple genetic algorithm. The term success ratio is used as a quality measure and its value shows the effectiveness of the BMOGA. Statistical tests indicate superiority of the BMOGA over other existing approaches implemented.