Multi-label classification aims at assigning more than one label to each instance. Many real-world multi-label classification tasks are high dimensional, leading to reduced performance of traditional classifiers. Feature selection is a common approach to tackle this issue by choosing prominent features. Multi-label feature selection is an NP-hard approach, and so far, some SWARM INTELLIGENCE-based strategies and have been proposed to find a near optimal solution within a reasonable time. In this paper, a hybrid INTELLIGENCE algorithm based on the binary algorithm of particle SWARM optimization and a novel local search strategy has been proposed to select a set of prominent features. To this aim, features are divided into two categories based on the extension rate and the relationship between the output and the local search strategy to increase the convergence speed. The first group features have more similarity to class and less similarity to other features, and the second is redundant and less relevant features. Accordingly, a local operator is added to the particle SWARM optimization algorithm to reduce redundant features and keep relevant ones among each solution. The aim of this operator leads to enhance the convergence speed of the proposed algorithm compared to other algorithms presented in this field. Evaluation of the proposed solution and the proposed statistical test shows that the proposed approach improves different classification criteria of multi-label classification and outperforms other methods in most cases. Also in cases where achieving higher accuracy is more important than time, it is more appropriate to use this method.

This paper presents a new approach to solve Fractional Programming Problems (FPPs) based on two different SWARM INTELLIGENCE (SI) algorithms. The two algorithms are: Particle SWARM Optimization, and Firefly Algorithm. The two algorithms are tested using several FPP benchmark examples and two selected industrial applications. The test aims to prove the capability of the SI algorithms to solve any type of FPPs. The solution results employing the SI algorithms are compared with a number of exact and metaheuristic solution methods used for handling FPPs. SWARM INTELLIGENCE can be denoted as an effective technique for solving linear or nonlinear, non-differentiable fractional objective functions. Problems with an optimal solution at a finite point and an unbounded constraint set, can be solved using the proposed approach. Numerical examples are given to show the feasibility, effectiveness, and robustness of the proposed algorithm. The results obtained using the two SI algorithms revealed the superiority of the proposed technique among others in computational time. A better accuracy was remarkably observed in the solution results of the industrial application problems.

To achieve high-quality software, different tasks such as testing should be performed. Testing is known as a complex and time-consuming task. Efficient test suite generation (TSG) methods are required to suggest the best data for test designers to obtain better coverage in terms of testing criteria. In recent years, researchers to generate test data in time-efficient ways have presented different types of methods. Evolutionary and SWARM-based methods are among them. This work is aimed to study the applicability of SWARM-based methods for efficient test data generation in EvoSuite. The Firefly Algorithm (FA), Particle SWARM Optimization (PSO), Teaching Learning Based Optimization (TLBO), and Imperialist Competitive Algorithm (ICA) are used here. These methods are added to the EvoSuite. The methods are adapted to work in a discrete search space of test data generation problem. Also, a movement pattern is presented for generating new solutions. The performances of the presented methods are compared over 103 java classes with two built-in genetic-based methods in EvoSuite. The results show that SWARM-based methods are successful in solving this problem and competitive results are obtained in comparison with the evolutionary methods.

Background and Objectives: Stock price prediction has become one of the interesting and also challenging topics for researchers in the past few years. Due to the non-linear nature of the time-series data of the stock prices, mathematical modeling approaches usually fail to yield acceptable results. Therefore, machine learning methods can be a promising solution to this problem. Methods: In this paper, a novel machine learning approach, which works in two phases, is introduced to predict the price of a stock in the next day based on the information extracted from the past 26 days. In the first phase of the method, an automatic clustering algorithm clusters the data points into different clusters, and in the second phase a hybrid regression model, which is a combination of particle SWARM optimization and support vector regression, is trained for each cluster. In this hybrid method, particle SWARM optimization algorithm is used for parameter tuning and feature selection. Results: The accuracy of the proposed method has been measured by 5 companies’ datasets, which are active in the Tehran Stock Exchange market, through 5 different metrics. On average, the proposed method has shown 82. 6% accuracy in predicting stock price in 1-day ahead. Conclusion: The achieved results demonstrate the capability of the method in detecting the sudden jumps in the price of a stock.

The internet and its various services have made users to easily communicate with each other. Internet benefits including online business and e-commerce. E-commerce has boosted online sales and online auction types. Despite their many uses and benefits, the internet and their services have various challenges, such as information theft, which challenges the use of these services. Information theft or phishing attacks are internet attacks that are major approach to success it is social engineering that the phisher has used. In these types of attacks, the attacker deceives the users and steals their valuable information by using a fake website that looks like real websites. The damage caused by fake websites and phishing attacks is so high that researchers are trying to identify these types of websites in different ways. So far, various methods have been developed to identify phishing web sites which most of them based on data-mining and learning machine are trying to identify these malicious websites. Artificial neural network is a data-mining method for identifying phishing websites which is used in most studies; however the error rate of this can be significant in detecting these websites, so learning-based optimization algorithm is used as a SWARM INTELLIGENCE algorithm to reduce its error. In the proposed method, the error rate of multi-layer artificial neural network in detecting phishing websites is considered as a target function which minimized by using learning-based optimization algorithm. In the proposed method, learning-based optimization algorithm selects weights and bias of multi-layer artificial neural network optimally to minimize the error of clssification as an objective function. The datasets used to evaluate the proposed method are Phishing Websites explaind by others. The results of evaluating phishing attack dataset indicate that the rate of error of fake website detection in the proposed method is constantly reduced by repetition. The results of our assessment also indicate that the average accuracy, sensitivity, specificity, precision of the proposed method are 93. 42, 92. 27, 93. 19 and 92. 78%, respectively. The decision tree and regression are more accurate in detecting fake websites than artificial neural network.

Most applications of sensor nodes are in hazardous areas, inaccessible or hostile environments. Therefore, the need for security in these networks is essential. Trust methods are powerful tools for diagnosing unexpected behavior of nodes (malicious nodes or failure nodes). In this paper, we have proposed TBSI trust model whose main features are low computational overhead, low energy consumption and confronting attacks in WSNs. This model is simulated and evaluated by NS-2 simulator and its behavior has been evaluated based on the results of these simulations. Examining practical results shows that energy consumption, routing overhead, and the time of death of nodes are reduced and the rate of packet delivery to the base station is increased. These desirable outcomes prove that using the method of trust to achieve a secure network is a good solution to solve security issues in wireless sensor networks.