JOURNAL OF ADVANCES IN COMPUTER RESEARCH
Year:2013 | Volume:4 | Issue:3 (13)|Papers Count:8

Evolutionary algorithms are some of the most crucial random approaches to solve the problems, but sometimes generate low quality solutions. On the other hand, Learning automata are adaptive decision-making devices, operating on unknown random environments, So it seems that if evolutionary and learning automaton based algorithms are operated simultaneously, the quality of results will increase sharply and the algorithm is likely to converge on best results very quickly.This paper contributes an algorithm based on learning automaton to improve the evolutionary algorithm for solving a group of NP problems. It uses concepts of machine learning in search process, and increases the efficiency of evolutionary algorithm (especially genetic algorithm). In fact, the algorithm is prevented from being stuck in local optimal solutions by using learning automaton. Another positive point of the hybrid algorithm is its noticeable stability since standard division of results, which is obtained by different executions of algorithm, is low; that is, the results are practically the same. Therefore, as the proposed algorithm is used for a set of well-known NP problems and the results are very suitable it can be considered as a precise and reliable technique to solve the problems.

Web Usage Mining (WUM) is the automatic discovering of hidden information of user access pattern from the web log data. Frequent pattern discovery is one of the main techniques in WUM that can be used to implement recommender systems, forecast user’s navigational behavior, and personalize web sites. Many algorithms have been suggested on obtaining frequent user navigation patterns. This paper presents PD-FARM (Pattern Discovery based on Fuzzy Association Rule Mining) algorithm to extract the web usage patterns, based on Fuzzy Association Rule Mining (FARM). Fuzzy Frequent Pattern-Growth (FFP-Growth) algorithm is used to FARM. Finally, an illustrated example is given for a complete description of the proposed algorithm.

This paper presents a fuzzy approach to the prediction of highly nonlinear time series.The optimized Mamdani-type fuzzy system denoted SQP-FLC is applied for the input-output modeling of measured data. In order to tune fuzzy membership functions, a sequential quadratic programming (SQP) method is employed. The proposed method is evaluated and validated on a highly complex time series, daily gold price data. The time series is primarily investigated for its chaotic properties.Correlation dimension and autocorrelation function (ACF) for the time series are discussed. Accordingly, time delay and embedding dimension are computed. Month selection in each stage is based on computed correlation coefficients. Thus, for the proposed fuzzy predictor, 3, 5, and 7 dynamics are selected and the time series are verified. The simulation results for one-step-ahead prediction of daily gold price in 2010, compared with methods of ANFIS and GA-FLC, demonstrate comparably better performance of the proposed SQP-FLC until the higher significant dynamics of the chaotic trend is taken into account.

Distributed generation (DG) sources are becoming more prominent in distribution systems due to the incremental demands for electrical energy. Locations and capacities of DG sources have great impacts on the system losses in a distribution network. This paper presents a study aimed for optimally determining the size and location of distributed generation units in distribution systems with different load models. The objective is to minimize network power losses. The impacts of DG model on locating and sizing of DG are also presented considering different voltage dependent load models. Also, different types of customers such as industrial, residential, and commercial loads are considered for load modeling. The optimization problem has been solved using genetic algorithm. For simulation purpose, this algorithm has been executed on 33-bus and 69-bus test systems.Results show that type of DG modeling and load modeling has considerable effect on determination of the optimum siting and sizing of DGs. Also, DGs installation in optimum size and location has considerable effect on loss reduction and voltage improvement of distribution system.

Job Shop scheduling problem has significant importance in many researching fields such as production management and programming and also combined optimizing. Job Shop scheduling problem includes two sub-problems: machine assignment and sequence operation performing. In this paper combination of particle swarm optimization algorithm (PSO) and gravitational search algorithm (GSA) have been presented for solving Job Shop Scheduling problem with criteria of minimizing the maximal completion time of all the operations, which is denoted by Make span. In this combined algorithm, first gravitational search algorithm finds best mass with minimum spent time for a job and then particle swarm Optimization algorithm is performed for optimal processing all jobs. Experimental results show that proposed algorithm for solving job shop scheduling problem, especially for solving larger problem presents better efficiency. Combined proposed algorithm has been named GSPSO.

Internet applications spreading and its high usage popularity result in significant increasing of cyber-attacks. Consequently, network security has become a matter of importance and several methods have been developed for these attacks.For this purpose, Intrusion detection systems (IDS) are being used to monitor the attacks occurring on computer networks. Data mining Techniques, Machine Learning, Neural networks, Collective Intelligence, Evolutionary algorithms and Statistical methods are some of algorithms which have been used for classification, training and reviewing detection accuracy with analysis based on the standard datasets in Intrusion Detection Systems. In this Paper, the hybrid algorithm is introduced based on decision tree and support vector machine (SVM) using feature selection and decision rules to apply on IDS. The main idea is to use the strengths of both algorithms in order to improve detection, enhance the accuracy and reduce the rate of error detection of the results. In this algorithm, the best features are selected by SVM, afterwards decision tree is used to make decisions and define rules. The results of applying proposed algorithm are analyzed on the standard dataset KDD Cup99. The proposed method guarantees high detection rate which is proved by simulation results.

Diabetic Retinopathy is one of the most important reasons of blindness which causes serious damage in the retina. The aim of this research is to detect one lesions of the retina, named Exudates automatically with Image processing techniques.Preprocessing is the first step of proposed algorithm. After preprocessing, the optic disc was detected and removed from the retinal image due to the same color of OD and the exudates. Next, the HSV format of image has been used where the H and V channels, standard deviation on green channel of retinal image and the background removal features were used as input of the system. The Fuzzy C-mean algorithm is used for classification. In this research the databases were Diaretdb0 and Diaretdb1. The results show 88.86% for Sensitivity and 99.98% for Specificity.Also, the result for PPV was 95.66% and the Accuracy was 99.90%.

Artificial Intelligence (AI) techniques offer powerful objective algorithms for analysis of multimodal and high-dimensional data. Recently, these techniques have become a reliable tool in the medical domain. This paper describes an efficient technique for building an application that is capable of forecasting and classifying healthcare information using machine learning as a subfield of AI methods. The algorithm predicts a label for each sample. The sample is a single set of feature data and the label is what category the sample falls into. The algorithm takes many of these samples as the training set, builds an internal model and finally predicts the labels of other samples, called the testing set. We apply this methodology to the breast cancer staging and also to forecast the myocardial infarction and examine the risk assessment using fuzzy clustering and Framingham heart study. The results show that the proposed technique obtains credible outputs that could be integrated in an application to be used in the health care field.