International Journal of Swarm Intelligence (9 papers in press)

Regular Issues

• A comprehensive review on recent intelligent metaheuristic algorithms  
  by S. Rajalakshmi, S. Kanmani 
  Abstract: Metaheuristics is an interesting research area with significant advances in solving problems with optimisation. Substantial advancements in metaheuristic are being made, and various new algorithms are being developed every day. The analyses in this area will undoubtedly helpful for future improvements. This papers main objective is to conduct a literature review of some recent algorithms motivated by nature to compare their features. This paper reviews some recently published nature inspired algorithms such as squirrel search algorithm (SSA), improved squirrel search algorithm (ISSA), grey wolf optimiser (GWO) algorithm, random walk grey wolf optimiser (RW_GWO) algorithm, sailfish optimiser (SAO) algorithm, sandpiper optimisation algorithm (SOA), search and rescue operations (SRO) algorithm, slime mould optimisation (SMO) algorithm, grasshopper optimisation algorithm (GOA) and opposition based learning grasshopper optimisation algorithm (OBLGOA). This paper focuses on a brief introduction of these algorithms and key concepts involved in formulation of swarm intelligence. Finally, this work outlines the directions for conducting effective future research.
  Keywords: metaheuristics; optimisation; swarm intelligence; improved metaheuristics.
  DOI: 10.1504/IJSI.2021.10037176
   
  

Special Issue on: ICEODS-2019 Recent Advances in Engineering Optimisation and Data Science for Sustainable Future Development

• Optimisation of EDM process parameters by application of genetic algorithm  
  by Sumit Sharma, Love Kishore Sharma, Dheeraj Joshi, Mohammad Israr, Ritesh Mathur, Deepak Sharma 
  Abstract: The EDM efficiency can be strengthen by suitably defining the critical factors and their values for obtaining desired responses. In present research work, response surface methodology (RSM) with face centred cubic (FCC) approach is implemented for correlating the response, i.e., material removal rate (MRR) with control factors, i.e., voltage, peak current, pulse on time. These factors provide maximum MRR of 230.6 gm/min. To obtain the optimal solution, genetic algorithm (GA) is coupled with the obtained mathematical model. The dielectric fluid used is kerosene oil. ANOVA and F-test are used to check for model validation. R2 and adjusted R2 which confirms the validity of proposed model. The work piece material used in research work is AISI 4140 steel.
  Keywords: response surface methodology; RSM; face centred cubic; FCC; material removal rate; MRR; genetic algorithm; ANOVA.
  DOI: 10.1504/IJSI.2021.10037199
   
  

Special Issue on: PCCDS-2020 Building Intelligent Applications Using Machine Learning

• Hybrid ARIMA-deep belief network model using PSO for stock price prediction  
  by Shaikh Sahil Ahmed, Mahesh Kankar, Nagaraj Naik, Biju R. Mohan 
  Abstract: Forecast analysis is in very high demand in many fields for improving sales and operation planning in various industries and enterprises. So, accuracy is a significant factor in forecasting stock market prices. We already know there are existing deep learning models for stock market prediction such as gated recurrent unit (GRU), support vector machine (SVM), multilayer perceptron (MLP), etc. This paper enhanced the prediction of stock prices using series hybrid models over single deep learning models. The models we used are autoregressive integrated moving average (ARIMA), deep belief network (DBN), long short-term memory (LSTM), and performed analysis on hybrid models in comparison with single models. We have chosen a model as ARIMA, LSTM, and hybrid as ARIMA-DBN and ARIMA-LSTM. For finding the best fit parameter for ARIMA and DBN, the particle swarm optimisation (PSO) technique is used. We compared the various models based on performance errors like MSE, RMSE, MAPE, etc. As already existing ARIMA and LSTM is not good enough for forecasting and so we worked over the ARIMA-DBN model to overcome the limitations of other models. After research, we found out that series hybrid ARIMA-DBN is effectively better than other single models for stock market prediction.
  Keywords: deep learning; time series forecasting; autoregressive integrated moving average; ARIMA; linear and nonlinear models; particle swarm optimisation; PSO.
  DOI: 10.1504/IJSI.2021.10036080
   
  
• Analysis of malware by integrating API extracted from dynamic and memory analysis  
  by Nishant Kumar, Lokesh Yadav, Deepak Singh Tomar 
  Abstract: Nowadays, malware is being developed and implemented on a large-scale which poses a critical security threat to digital devices. Therefore, effective analysis of malware is an important concern for security experts. Malware software exploits security vulnerabilities of the device and compromises the security of computing settings. Static analysis is a time-consuming approach and also requires a lot of manual effort. To overcome this limitation, dynamic analysis is carried in this paper by performing malicious code execution and which is enough capable in identifying multi-functional malware. But sometimes dynamic analysis is unable to handle obfuscated malware due to its API hooking capability. Hence, an approach is applied to combine dynamic analysis technique with memory analysis technique to provide an effective and efficient method for analysing malware using API calls. This approach is performed in a safe and isolated environment to capture the behaviour of the malware. This study shows a noteworthy improvement in accuracy, i.e., 98.62% and reduction in false positive rate, i.e.,1.3%.
  Keywords: malware; malware analysis; memory dumps; dynamic analysis; API calls; machine learning.
  DOI: 10.1504/IJSI.2021.10036458
   
  
• Hyperparameter tuning and comparison of k nearest neighbour and decision tree algorithms for cardiovascular disease prediction  
  by Preeti Bhowmick, Sachin Gajjar, Shital Chaudhary 
  Abstract: This work aims to do hyperparameter tuning and comparison of k nearest neighbour (kNN) and decision tree algorithms for cardiovascular disease (CVD) prediction using Framingham dataset. Hyperparameter tuning is done to find optimal value of k using Euclidean, Manhattan and Chebyshev distance metric in kNN. Hyperparameter tuning is done in decision tree, to find optimal value of the depth of the tree using Gini index and Information gain attribute selection method. The algorithms are compared on the basis of confusion matrix, accuracy, error rate, specificity, recall, precision, F1 score, execution time and ROC-AUC. The results show the accuracy of the decision tree is 2% less than kNN but decision tree is 46.36% more time efficient. The AUC value of kNN is 0.613 and decision tree is 0.588. Decision tree is more appropriate for predicting CVD, as it predicted ten more true positives in confusion matrix.
  Keywords: cardiovascular disease prediction; machine learning; hyperparameter tuning; k nearest neighbour; kNN; decision tree.
  DOI: 10.1504/IJSI.2021.10036482
   
  
• Solving bulk transportation problem using a modified particle swarm optimisation algorithm  
  by Gurwinder Singh, Amarinder Singh 
  Abstract: Particle swarm optimisation (PSO) algorithm is renowned for its ability to deal with a wide variety of real life complex problems. The adaptive ability of PSO makes it applicable to continuous as well as discrete optimisation problems. The bulk transportation problem is one such discrete optimisation problem wherein the objective is to minimise the transportation cost while satisfying the bulk demand for each destination. In this article, the PSO is modified to integrate additional modules to resolve both the infeasible intermediate solution and the non-integral variables. The proposed algorithm also maintains the condition of bulk purchase of the product from a single source. To validate the proposed PSO, different test problems have been taken and it is found that the proposed method is quite effective in its convergence capability and quality of solution.
  Keywords: discrete optimisation; bulk transportation problem; swarm intelligence; global best solution.
  DOI: 10.1504/IJSI.2021.10037054
   
  
• Automatic speaker verification system using three dimensional static and contextual variation-based features with two dimensional convolutional neural network  
  by Aakshi Mittal, Mohit Dua 
  Abstract: Automatic speaker verification (ASV) systems are being used as potential alternatives for authentication in security systems. This paper discusses the development of an ASV system trained by logical access (LA) and physical access (PA) sets of ASVspoof 2019 dataset. ASV systems have two parts frontend and backend. The frontend part of the proposed system includes the extraction of 30 static, 30 first orders and 30 second order constant Q cepstral coefficients (CQCC) features from each frame of an audio. These features are reshaped in three dimensional (3D) tensors of two dimensional (2D) slices with the chosen fix number of frames. A two dimensional convolutional neural network (2D CNN) is trained in the backend with these features. The proposed system achieves 0.055 equal error rate (EER) and 0.101 tandem detection cost function (tDCF) for LA set and 0.062 EER and 0.122 tDCF for the PA set of the taken dataset.
  Keywords: contextual variation; three dimensional features; CQCC features; 2D CNN.
  DOI: 10.1504/IJSI.2021.10037055
   
  
• Identification of female genital tuberculosis in infertility using textural features  
  by Varsha Garg, Anita Sahoo, Vikas Saxena 
  Abstract: The effect of female genital tuberculosis (FGTB) on fertility of women is a topic of discussion in the medical fraternity but has still not made inroads to computer aided detection. The travails of an infertile woman could be alleviated if a non-invasive method such as transvaginal ultrasound (TVUS) could provide early insights to FGTB detection. In this paper, a novel effort has been made towards effective classification of FGTB as normal and abnormal in infertility using TVUS image analysis. Real-time TVUS images of female visiting for infertility treatments have been collected from medical centres in consultation with medical experts in India. The identification of FGTB in infertility is done in four stages; image augmentation, grey level co-occurrence matrix-based textural feature extraction, two-phased feature selection based on mutual information-based ranking followed by sequential forward selection and classification. Experiments were conducted with different classifiers, where maximum accuracy is obtained by support vector machine (SVM). The testing results show that SVM effectively classifies the dataset in hand showing a mean accuracy of 83.41%. The two-phased feature selection method is able to reduce the dimensionality of textural feature vectors by 76.19%.
  Keywords: genital tuberculosis; ultrasound image processing; textural feature; feature selection; mutual information; classification.
  DOI: 10.1504/IJSI.2021.10037151
   
  
• Predicting movie genre from plot summaries using Bi-LSTM network  
  by Prakhar Srivastava, Pankaj Srivastava 
  Abstract: Movie plot summaries are highly indicative of the genre to which they belong. Depending upon the words present in the plot summaries, we can easily decide which emotion is being portrayed in the movie. In this paper, we predict the movie genres by feeding the plot summaries to our proposed model. For making word representations that can be understood by our model, we use Facebooks fasttext library. Our model uses a Bi-LSTM network and a ranking system depending upon posterior probability scores to determine the movie genre. We split the plot summary into sentences and predict the genre associated with each sentence using our model. We then fuse the decision from all the sentences to make a collective decision for a particular plot summary. We use the majority voting algorithm for making this decision. We try document-level and sentence-level approaches for predicting the movie genres. Post comparison of results, we found sentence-level approach using Bi-LSTM network performs better than the document-level approach using the same network. For the baseline models, we used recurrent neural networks (RNN) and logistic regression (LR) and compared the results with our proposed model.
  Keywords: NLP; movie genre prediction; Bi-LSTM; word embeddings; recurrent neural networks; fasttext; majority voting algorithm.
  DOI: 10.1504/IJSI.2021.10037658