International Journal of Intelligent Engineering Informatics (16 papers in press)
Empirical study to predict the understandability of requirements schemas of data warehouse using requirements metrics
by Tanu Singh, Manoj Kumar
Abstract: Information quality of data warehouse is assessed by its data model quality. Various authors have proposed metrics for data models, that are designed to capture physical, conceptual, logical and requirements views of data warehouse. These metrics were validated not only formally but also empirically to assess quality of the respective data models. However, very less work was seen in the literature to assess quality of requirements model. Therefore, in this paper, an empirical validation of requirements metrics are performed to predict the understandability of requirements schemas of data warehouse using machine learning techniques (random forest and artificial neural network). Result shows that, artificial neural network technique performed better than random forest technique. In this way, effect of requirements metrics on understandability of schemas has been assessed, thus, good quality of requirements schema may be identified and help to the designers for producing better quality of conceptual schema.
Keywords: Artificial neural network; Data warehouse; Requirements engineering; Requirements metrics; Requirements schemas understandability; Random forest.
Application of Metaheuristic Techniques in Software Quality Prediction: A Systematic Mapping Study
by KIRTI LAKRA, Anuradha Chug
Abstract: This paper focuses on the systematic review of various metaheuristic techniques employed for analyzing different software quality aspects, including fault proneness, defect anticipation, change proneness, maintainability prediction, and software reliability prediction. It is observed that machine learning algorithms are still popular models, but metaheuristic algorithms are also gaining popularity in the field of software quality measurement. This is due to the fact that metaheuristic algorithms are more efficient in solving real-world, search-based, and optimization problems. Initially, 90 papers were considered and analyzed for conducting this study from 2010 to 2020, and 55 studies were shortlisted based on predesigned quality evaluation standards. Resultantly, Particle Swarm Optimization, and Genetic Algorithms came out as the most prominently used metaheuristic techniques for developing software quality models in 36.3% and 27.2% of the shortlisted studies, respectively. The current review will benefit other researchers by providing an insight into the current trends in software quality domain.
Keywords: metaheuristic techniques; object-oriented metrics; software quality; software fault proneness; software defect prediction; software change prediction; software reliability prediction; software maintainability prediction; software quality improvement.
Multiclassifier Learning for the Early Prediction of Dementia disease progression from MCI
by Rohini M
Abstract: Recently many machine learning and deep learning prediction models have been proposed for early detection and classification of Alzheimers Disease (AD). Before the actual onset of AD, there occurs several structural changes in brain. This stage of pathology causes Mild Cognitive Impairment (MCI). The proposed study intends to develop machine learning model that utilizes relevant subset of predictors to diagnose the progression of disease. The conversion from MCI to Stable MCI (sMCI) or Progressive MCI (pMCI) is identified at early stage of onset of symptoms. The quality of existing research works lies in more early identification of disease that greatly affects subjects recovery. The Open Access Series of Imaging Studies (OASIS-1) cross-sectional MRI data were analyzed. This study utilized Mini-mental state exam (MMSE), Clinical Dementia Rating(CDR), Estimated Total Intracranial Volume, Normalize Whole Brain Volume and Atlas Scaling Factor for constructing randomized trees and thus predicting the progression of disease stages from MCI to Alzheimers disease that causes Dementia. These features are also used for constructing SVM models that produced corresponding performance without significant accuracy loss. The proposed model proved to give good robust classification results that is sufficient for future clinical implementation.rn
Keywords: Alzheimer’s disease; Mild Cognitive impairment; stable Mild Cognitive impairment; progressive Mild Cognitive impairment.
Detection of Induction Motor Broken Rotor Bar Faults under No Load Condition by using Support Vector Machines
by Hayri Arabaci, Mohamed Ali Mohamed
Abstract: Induction motors are the most popular electric machines used domestically and in multiple industries. They are compact, low in cost and easy to maintain. Nonetheless, induction motors experience faults during operation as a result of stresses or manufacturing errors. These faults are grouped into stator, rotor, bearing and eccentricity-related faults. An important fault is the broken rotor bar fault. Many techniques have been proposed for the detection of the rotor faults. However, the traditional techniques like motor current signature analysis have difficulty in detecting the rotor faults at no load condition due to low slip. In this study, an algorithm which uses fast Fourier transform, principal component analysis and intelligent classifiers is proposed. The proposed algorithm was able to accurately detect the rotor faults of different severity levels at low slip. Experiments were carried out with three submersible induction motors. Four different rotor faults and healthy motor conditions were investigated for each motor. The motors were loaded different load levels to test the proposed algorithm. The best results were achieved with Medium Gaussian Support Vector Machine. The condition of having any faulted bar in the motor was obtained with 100% accuracy and faults classification carried out by 92.2% accuracy.
Keywords: Current measurement; Fault detection; Feature extraction; Induction motors; Spectral analysis;Support Vector Machines.
Optimization of thresholding techniques in de-noising of ECG signals
by Shivani Saxena, Ritu Vijay, Pallavi Pahadiya
Abstract: Wavelet based thresholding procedure is most popular approach to obtained de-noised ECG signal. Traditional thresholding techniques classified into soft and hard thresholding, have their own limitations. Former one involves large data size leading to longer computational time and second one involving compression of data results in loss of information sometimes crucial. The work in this paper optimized the features of both thresholding methods in which elimination of noisy wavelet coefficients is performed and rest coefficients are modified using features of hard and soft thresholding respectively. Computed Mean Square Error is minimum in proposed method among hard thresholding, soft thresholding and proposed thresholding technique. In addition, histogram plot is used to identify frequency distribution in de-noising ECG signals.
Keywords: ECG; Baseline Wander Noise; Power line Interference Noise; Thresholding; Wavelet transform.
Special Issue on: MIDAS-2020 Machine Learning Algorithms and Applications in Industry 4.0
Recovery of a single link failure in all-optical networks based on the cuckoo search algorithm
by DINESH KUMAR, Neeru Sharma, Rajiv KUMAR
Abstract: This paper presents a problem of a link failure in all types of optical networks. Here, the concept of monitoring path (MPs) and monitoring cycles (MCs) has been introduced for the identification of single-link failures in optical networks. MPs and MCs need to go through one or more monitoring locations. These monitoring locations are established such that any single failure in the optical networks is the combination of MPs and MCs, which pass through the different monitoring locations. For a single monitoring location in the network, the three-edge connectivity is a sufficient and essential condition for establishing the MPs and MCs to distinctively identify the single link failure in the network. The average bandwidth blocking probability (BBP) for dedicated path protection (DPP), shared path protection (SPP), and proposed survivable scheme (PSS) are 0.3539, 0.399, and 0.0073 respectively. The average recovery time for DPP, SPP, and PSS are 3.8212, 12.526, and 8.6695. Similarly, the average bandwidth provisioning ratio (BPR) for DPP, SPP, and PSS are 3.5160, 1.8011, and 1.7575 respectively. In this work, the proposed cuckoo search algorithm reduces the recovery time for single link failure in all types of optical networks. The proposed system is validated in MATLAB software for testing this system.
Keywords: Optical networks; Shared path protection; Dedicated path protection; Cuckoo search algorithm.
A MCDM-Based Performance of classification algorithms in breast cancer prediction for imbalanced datasets
by Monika Lamba, Geetika Munjal, Yogita Gigras
Abstract: The choice of a suitable classification algorithm is extremely important to worry about in different areas involving cancer diagnosis. It associates more than one benchmark for selecting the correct classification algorithm, so the appropriate choice requires it to plan as a Multiple Criteria Decision Making (MCDM) barrier. Several techniques of MCDM adjudicator classifiers against distant potential aspects and resulting in the production of conflict ranking of numerous classifiers. The main motive of the manuscript is to propose a technique for determining leading classification algorithms utilizing the MCDM techniques. Three MCDM techniques are checked with the assistance of 20 classification algorithms on seven micro-array gene expression datasets in accordance with the principle of estrogen receptor (ER) status utilizing 12 performance benchmarks in an experimental investigation. Firstly, primary ranking generated are non-conclusive. Secondly, utilizing the spearsons rank correlation coefficient to settle the differences in secondary ranking shows top classifiers as IBK, RBF_Network, Naive Bayes, and Filtered Classifier.
Keywords: Multi-criteria decision making (MCDM); VIKOR; TOPSIS; GRC; breast cancer; estrogen receptor; SRCC.
A Novel Approach for Identification and Classification of Verbs in Dogri Language
by Shubhnandan S. Jamwal, Parul Gupta, Vijay Singh Sen
Abstract: Morphological Analyzer, POS data, Stemmer etc. are the basic tools required for any NLP tasks, which are not available for Dogri language which recently has been declared as an official language of J&K, UT. Because of the unavailability of the basic tools, it remains a very low resourced language. In this paper, we have presented a sub task for the development of morphological analyzer for Dogri language. We have identified the morphological behavior of the verbs and implemented the automatic process of the identification of the verbs in Dogri language using paradigm approach. The various forms of verb taken into consideration are specifically transitive, intransitive, non-finite, gerund and infinitives. The average accuracy attained in the process of identification of transitive, intransitive, non-finite, gerund and infinitive verbs is is 80%, 83%, 76%, 93%, 88% respectively.
Keywords: Morphology; Dogri; Inflections; Paradigm; Verb; Non-finite; Gerund; Transitive; Intransitive.
Special Issue on: IC_ASET’2020 Intelligent and Advanced Control Methods in Robotic Applications
Adaptive Iterative Learning-based Gait Tracking Control for Pediatric Exoskeleton during Passive-assist Rehabilitation
by Jyotindra Narayan, Mohamed Abbas, Santosha K. Dwivedy
Abstract: The design of a robust control scheme is considered a benchmark problem to address the uncertain dynamic parameters and un-modeled disturbances of the exoskeleton system. This work proposes a robust adaptive iterative learning (AIL) control (AILC) scheme for a pediatric exoskeleton system. Primarily, the mechanical description of the exoskeleton system is briefly presented along with the input parameters. Thereafter, the dynamic relation, invoking kinetic and potential energy of the system, is formulated via the Euler-Lagrange principle. The stability of the AILC scheme is ascertained using the Lyapunov analysis. The robustness is validated by incorporating the parametric uncertainties (varied mass) and un-modeled disturbances (trigonometric and random noises). Thereafter, the controllers performance is compared with classical iterative learning control (ILC) and exponential reaching law- sliding mode control (ERL-SMC) schemes. Finally, it is observed from simulation runs that the AIL controller has enough potential to track the desired gait trajectory accurately.
Keywords: robust control; adaptive iterative learning; pediatric exoskeleton; Lyapunov; parametric uncertainties; un-modeled disturbances.
Path Planning Strategy for Unmanned Aerial Vehicles Based on a Grey Wolf Optimizer
by Raja Jarray, Soufiene Bouallègue
Abstract: The path planning problems for Unmanned Aerial Vehicles (UAVs) can be considered as Large Scale Global Optimization (LSGO) problems. Collision-free and smoother flyable paths require increased sequences of flight waypoints acting as the decision variables of the formulated hard optimization problem. In this paper, an intelligent path planning strategy based on the partition of the work area into multiple sub-environments and a parameters-free Grey Wolf Optimizer (GWO) metaheuristic is proposed for a UAV drone. For each formulated planning sub-problem of reduced dimension, a collision-free with shorter length sub-path is optimized under operational constraints of obstacles avoidance and paths straightness limitation. A cubic spline technique is thus used to smooth the generated flight route and make the planned path more suitable for the UAV. The effects of the partitions size of the flight 3D static environment are investigated and discussed through demonstrative numerical simulations and nonparametric statistical analyzes. A comparative study is carried out to show the effectiveness and superiority of the proposed GWO-based planning technique compared to other homologous metaheuristics from families of swarm intelligence and evolutionary algorithms, i.e. Water Cycle Algorithm (WCA), Salp Swarm Algorithm (SSA), Particle Swarm Optimization (PSO), and Differential Evolution (EA). The conducted results are satisfactory and very encouraging in the aim of a future practical implementation using the real-world prototype Parrot AR. Drone 2.0 and the associated MATLAB/Simulink software toolkit.
Keywords: Unmanned aerial vehicles; path planning; large-scale optimization problems; global metaheuristics; grey wolf optimizer.
Modelling, control and robustness analysis of a 2-DoF exoskeleton-upper limb system
by Sana Bembli, Nahla Khraief Haddad, Safya Belghith
Abstract: This paper presents a new Adaptive Gain Terminal Sliding mode with Gravity Compensation control of an exoskeleton-upper limb system. The treated system is a two-degree-of-freedom robot (2-DoF) in interaction with an upper limb used for rehabilitation. The objective is to control the flexion/ extension movement of the shoulder and the elbow in presence of matched disturbances. First, the modelling, the control and the stability study of the considered system using different laws were made. Next, a comparison study between these tested algorithms was done based on Monte Carlo robustness analysis method. Finally, the performance and the effectiveness of the Adaptive Gain Terminal Sliding mode algorithm combined with the Gravity Compensation in presence of uncertainties are provided by simulation results.
Keywords: Exoskeleton- upper limb system; dynamic model; Adaptive Gain Terminal Sliding Mode; Gravity Compensation; stability study; Monte Carlo analysis; matched disturbances.
Improved filter design in internal model control: Application to hybrid feed drive mechatronic system
by Nahla Touati, Imen Saidi, Dhaou Soudani
Abstract: In this paper, the internal model control is proposed for overactuated systems for high frequency bands. To deal with redundancy, the method of virtual outputs is considered to square the system and design the controller obtained by a specific inversion technique. An improved low-pass filter is then inserted in the internal model control structure in order to attenuate the sensitivity of the controller, improve the system performance and the robustness of the structure towards disturbances and uncertainties. The proposed filter design aims to increase the bandwidth of the system. Furthermore, to evaluate its efficiency, an application to control a mechanical linear system, namely hybrid feed drive system, is described. The obtained simulation results are shown to be satisfactory by considering a series of different scenarios.
Keywords: internal model control; overactuated system; virtual outputs; hybrid feed drive; low-pass filter; robustness.
Gait Stabilization of an Underactuated Bipedal Walker on Steep Slopes
by Surbhi Gupta
Abstract: In this paper, we have demonstrated that by using optimized design parameters and appropriate reference trajectory, a bipedal model can be underactuated to stably walk on steep-sloped surfaces. Underactuated bipeds can walk stably on level-ground and slopes of up to 20
Keywords: Underactuation; Switched Controller; Optimal Control; Hybrid Dynamics; Non-linear Dynamics.
Special Issue on: ICCIS-2020 Applications of Nature Inspired Algorithms in Optimisation
Fully Informed ABC Algorithm for Large Scale Job Shop Scheduling problem
by Kavita Sharma, P.C. Gupta
Abstract: The large-scale job-shop scheduling problem (LSJSSP) is a complex scheduling problems. Previously, although the nature-inspired algorithm, specially the swarm intelligence (SIA) based algorithms have been efficiently applied to solve the LSJSSP, finding the best solution for LSJSSP instances remains a challenging task. Therefore, in this paper, a novel SIA is applied to solve the 105 LSJSSP instances. The selected SIA is Fully Informed Artificial Bee Colony (FABC) algorithm. The FABC algorithm is a variant of the ABC algorithm in which position update process is inspired from the GABC. In the FABC, onlooker bee process of the ABC strategy is modified and designed such that the new position of the solution search agent is obtained while learning from all the nearby agents. The results obtained by the FABC is compared with the strategies available in the literature. The results analysis shows that the proposed approach to solving LSJSSP is competitive in the field of SIA.
Keywords: Fully Informed Learning; Swarm Intelligence; Artificial bee colony; large-scale JSS Problem.
Multi-Objective Tunicate Search Optimization Algorithm for Numerical Problems
by Isha Sharma, Vijay Kumar
Abstract: In this paper, a multi-objective version of recently developed Tunicate Swarm Algorithm (TSA) is proposed. The multi-objective TSA (MOTSA) utilizes the external archive to store the non-dominated solutions. The concept of roulette wheel mechanism is also incorporated in MOTSA for selection of non-dominated solutions. To demonstrate the effectiveness of MOTSA, it is evaluated on the well-known benchmark test functions. The proposed MOTSA is compared with four well-renowned multi-Objective optimization algorithms and quantitatively analysed by using the performance measures. The experimental results reveal that the proposed MOTSA outperforms the existing techniques in terms of performance measures.
Keywords: Tunicate search optimizer; Optimization; Swarm intelligence; Multi-Objective optimization.
Economic load dispatch problem using spider monkey optimization algorithm
by Ajay Sharma, Harish Sharma, Annapurna Bhargava, Nirmala Sharma
Abstract: The coal reserves are lessening, and subsequently fuel prices are escalating now a days. Therefore, an appropriate schedule of generating units for thermal energy is indispensable to convince this rising load demand at minimum price by reducing generation cost is termed as economic load dispatch problem (ELDP). Spider monkey optimization algorithm (SMOA) is implied to execute ELDP. The proposed methodology considers transmission losses, prohibited operating zones, and multiple fuels constraints. The experiments were executed over a variety of arrangements of a different number of generating units. The significance of the applied strategy to solve ELDP is analyzed while assessing the reported results with other renowned methods.
Keywords: Swarm Intelligence; Economic Load Dispatch Problem; Spider Monkey Optimization; Nature Inspired Algorithm.