Vipul K. Dabhi

Empirical modeling using genetic programming: a survey of issues and approaches

Empirical modeling, which is a process of developing a mathematical model of a system from experimental data, has attracted many researchers due to its wide applicability. Finding both the structure and appropriate numeric coefficients of the model is a real challenge. Genetic programming (GP) has been applied by many practitioners to solve this problem. However, there are a number of issues which require careful attention while applying GP to empirical modeling problems. We begin with highlighting the importance of these issues including: computational efforts in evolving a model, premature convergence, generalization ability of an evolved model, building hierarchical models, and constant creation techniques. We survey and classify different approaches used by GP researchers to deal with the mentioned issues. We present different performance measures which are useful to report the results of analysis of GP runs. We hope this work would help the reader by facilitating to understand key concepts and practical issues of GP and steering in selection of an appropriate approach to solve a particular issue effectively.

Classification of ECG signals using machine learning techniques: A survey

Classification of electrocardiogram (ECG) signals plays an important role in diagnoses of heart diseases. An accurate ECG classification is a challenging problem. This paper presents a survey of ECG classification into arrhythmia types. Early and accurate detection of arrhythmia types is important in detecting heart diseases and choosing appropriate treatment for a patient. Different classifiers are available for ECG classification. Amongst all classifiers, artificial neural networks (ANNs) have become very popular and most widely used for ECG classification. This paper discusses the issues involved in ECG classification and presents a detailed survey of preprocessing techniques, ECG databases, feature extraction techniques, ANN based classifiers, and performance measures to address the mentioned issues. Furthermore, for each surveyed paper, our paper also presents detailed analysis of input beat selection and output of the classifiers.

Hybrid Wavelet-Postfix-GP model for rainfall prediction of anand region of India

An accurate prediction of rainfall is crucial for national economy and management of water resources. The variability of rainfall in both time and space makes the rainfall prediction a challenging task. The present work investigates the applicability of a hybrid wavelet-postfix-GP model for daily rainfall prediction of Anand region using meteorological variables. The wavelet analysis is used as a data preprocessing technique to remove the stochastic (noise) component from the original time series of each meteorological variable. The Postfix-GP, a GP variant, and ANN are then employed to develop models for rainfall using newly generated subseries of meteorological variables.The developed models are then used for rainfall prediction.The out-of-sample prediction performance of Postfix-GP and ANN models is compared using statistical measures. The results are comparable and suggest that Postfix-GP could be explored as an alternative tool for rainfall prediction.

Empirical modeling using symbolic regression via postfix Genetic Programming

Developing mathematical model of a process or system from experimental data is known as empirical modeling. Traditional mathematical techniques are unsuitable to solve empirical modeling problems due to their nonlinearity and multimodality. So, there is a need of an artificial expert that can create model from experimental data. In this paper, we explored the suitability of Neural Network (NN) and symbolic regression via Genetic Programming (GP) to solve empirical modeling problems and conclude that symbolic regression via GP can deal efficiently with these problems. This paper aims to introduce a novel GP approach to symbolic regression for solving empirical modeling problems. The main contribution includes: (i) a new method of chromosome representation (postfix based) and evaluation (stack based) to reduce space-time complexity of algorithm (ii) comparison of our approach with Gene Expression Programming (GEP), a GP variant (iii) algorithms for generating valid chromosomes (in postfix notation) and identifying non-coding region of chromosome to improve efficiency of evolutionary process. Experimental results showed that empirical modeling problems can be solved efficiently using symbolic regression via postfix GP approach.

Soft computing based intelligent grid architecture

Workflow generation, resource management, and scheduling of workflow are three main components of grid based computing architecture. The components in traditional grid architecture require human intervention due to lack of intelligence and use of knowledge from historical data. We have identified the need of intelligence based automatic workflow generation, and the prediction on run-time of tasks for optimal allocation and reservation of resources. We propose soft computing based architecture of intelligent grid. We also propose the neural network based algorithmic flow for the resource matchmaking, the allocation and reservation, and the scheduling of workflow. Our proposed architecture and algorithmic flow does not require human intervention in workflow generation and in finding and allocating optimal resource for task execution. As our proposed work provides flexibility and reduces complexity in workflow generation, it saves the valuable time and cost of user.

Survey of multi objective evolutionary algorithms

Multi-objective optimization aims at simultaneously optimizing two or more objectives of a problem. Multi-objective evolutionary algorithms (MOEAs) are widely accepted and useful for solving real world multi-objective problems. When we have two or more conflicting objectives of a problem then we can apply MOEA. MOEA generates a set of non-dominated solutions at the end of run, which is called Pareto set. The Pareto front contains set of Pareto solutions. Any MOEA aims to improve (i) convergence of population towards true Pareto front and (ii) diversity of solutions belonging to Pareto set. Generally, an external archive is used by MOEAs to maintain a set of non-dominated Pareto set solutions. Sometimes, Pareto set contains more number of solutions than the size of archive. This paper presents survey of various methods used by different MOEAs for reducing the size of Pareto set while maintaining solutions diversity. It presents comparison of these methods along with their advantages and disadvantages. The paper concludes by giving limitation of crowding distance based method in various scenarios.

A survey on semantic document clustering

Clustering is the process of partitioning a set of data objects into subsets. It is commonly used technique in data mining, information retrieval, and knowledge discovery for finding hidden patterns or objects from a data of different category. Text clustering process deals with grouping of an unstructured collection of documents into semantically related groups. A document is considered as a bag of words in traditional document clustering methods; however, semantic meaning of word is not considered. Thus, more informative features like concept weight are important to achieve accurate document clustering and this can be achieved through semantic document clustering because it takes meaningful relationship into account. This paper highlights major challenges in traditional document clustering and semantic document clustering along with brief discussion. This paper identifies five major areas under semantic clustering and presents a survey of 17 papers that has studied, covering major significant works. Moreover, this paper also provides a survey of tools, ontology databases, and algorithms, which help in applying and evaluating document clustering. The presented survey is used in preparing the proposed work in the same direction. This proposed work uses the concept weight for text clustering system which is to be developed based on a Hierarchical Agglomerative Clustering, Bisecting k-means algorithm, and Self Organized Map Neural Network in accordance with the principles of WordNet ontology as a background knowledge.

A survey on job scheduling algorithms in Big data processing

Scheduling problem has been an active area of research in computing systems since their inception. The Apache Hadoop framework has emerged as most widely adopted framework for distributed data processing because of open source and allowing use of commodity hardware. Job scheduling has become an important factor to achieve high performance in Hadoop cluster. Several scheduling algorithms have been developed for Hadoop-MapReduce model which vary widely in design and behavior, handling different issues such as locality of data, user share fairness and resource awareness. This paper highlights fundamental issues in job scheduling, presents classification of Hadoop schedulers, and discusses presented survey of existing scheduling algorithm. Moreover paper also discusses features, advantages, and limitations of the scheduling algorithms. This paper also discusses about how various resource monitoring tools or frameworks help in achieving better result from MapReduce. It also discusses customized MapReduce frameworks used for improving the performance. This paper would be useful to beginners and researchers for understanding the state-of-the-art on scheduling in Big data processing.

An improved SPEA2 Multi objective algorithm with non dominated elitism and Generational Crossover

Multi-objective Optimization Problem (MOP) is an essential and challenging area for scientific research of real life problem. Multi-objective Optimization Problem (MOP) can be effectively solved by Multi-objective Evolutionary Algorithm (MOEA). In this paper, enhancements to a renowned Multi-objective Evolutionary algorithm SPEA2 are proposed. The proposed enhancements are useful to improve convergence and diversity simultaneously. In present study, for better convergence, Generational Crossover and Non-dominated solutions (based on SPEA2 fitness) based elitism are used. K-nearest neighbor density estimation technique is used to maintain diversity among solutions. The proposed algorithm is tested on widely used test problems of ZDT family. Simulation results suggest that proposed algorithm outperforms SPEA2 and gives better or comparative performance with other preeminent Multi-objective algorithms. The proposed Generational Crossover concept is generic and can be used with other MOEAs as well.

Cutting stock problem: A survey of evolutionary computing based solution

The cutting stock problem (CSP) is an important problem in class of combinatorial optimization problems because of its NP-hard nature. Cutting of the required material from available stock with minimum wastage is a challenging process in many manufacturing industries such as rod industry, paper industry, textile industry, wood industry, plastic and leather manufacturing industry etc. The objective of this paper is to present a comparison of various meta-heuristic techniques for solving cutting stock problem based on accuracy and speed of convergence. It also focuses on new chromosome representation using co-operative co-evolutionary Genetic Algorithm (CCEGA) for multiple stock size cutting stock problem (MSSCSP). The main challenge in solving cutting stock problem is to develop chromosome representation for MSSCSP in GA. Moreover, this paper also presents detailed study of the existing chromosome representations for CSP problem with their loopholes and summary of the referred papers based on meta-heuristic techniques. To overcome these loopholes, we finally present the cutting pattern based chromosome representation using co-evolutionary GA.