Pattern recognition systems are widely used in a host of different fields. Due to some reasons such as lack of knowledge about a method based on which the best classifier is detected for any arbitrary problem, and thanks to significant improvement in accuracy, researchers turn to ensemble methods in almost every task of pattern recognition. Classification as a major task in pattern recognition, have been subject to this transition. The classifier ensemble which uses a number of base classifiers is considered as meta-classifier to learn any classification problem in pattern recognition. Although some researchers think they are better than single classifiers, they will not be better if some conditions are not met. The most important condition among them is diversity of base classifiers. Generally in design of multiple classifier systems, the more diverse the results of the classifiers, the more appropriate the aggregated result. It has been shown that the necessary diversity for the ensemble can be achieved by manipulation of dataset features, manipulation of data points in dataset, different sub-samplings of dataset, and usage of different classification algorithms. We also propose a new method of creating this diversity. We use Linear Discriminant Analysis to manipulate the data points in dataset. Although the classifier ensemble produced by proposed method may not always outperform all of its base classifiers, it always possesses the diversity needed for creation of an ensemble, and consequently it always outperforms all of its base classifiers on average. پرونده مقاله

پیش بینی کوتاه مدت بار در بازار برق اهمیت زیادی دارد. از طرفی عوامل مهم تأثیرگذار بر پیش بینی کوتاه مدت بار به ویژگی های بار الکتریکی و آب و هوایی هر منطقه بستگی دارد، بنابراین با استفاده از داده های واقعی استان چهارمحال و بختیاری-شامل بار و دما- به پیش بینی کوتاه مدت بار الکتریکی استان پرداخته ایم. بدین منظور با استفاده از چهار روش مختلف شبکه عصبی پرسپترون (MLp < /strong>)، مجمعی از شبکه عصبی پرسپترون (MLP Ensemble)، شبکه SVM(Support Vector Machine) و مجمعی از شبکه SVM به پیش بینی کوتاه مدت بار استان چهار محال و بختیاری پرداختیم. نتایج حاصل از مقایسه این چهار روش نشان می دهد که مجمعی از شبکه عصبی پرسپترون بهترین روش به منظور پیش بینی کوتاه مدت بار می باشد. پرونده مقاله

Wireless sensor networks (WSNs) are composed of many low cost, low power devices with sensing, local processing and wireless communication capabilities. Clustering is a useful topology-management approach to improve lifetime and reduce the energy consumption in wireless sensor networks. In this paper we have proposed a new dynamic clustering method (NDCM) where clusters are created periodically and cluster head (CH) is selected based on threshold function. Unlike the LEACH protocol that clustering are static and cluster head number is fixed in the entire scenario, CHs in our method distributed in Land dimensions and the number of cluster can be dynamically adjusted based on the number of nodes. The simulation was performed in MATLAB software and it was compared with LEACH, LEACH-C, O-LEACH, LEACH-B, M-LEACH, V-LEACH AND W-LEACH algorithms. The simulation results show that proposed method have been reduced energy conservation and enhancement of network lifetime comparing with LEACH algorithm. Coverage of the number of clusters in proposed method is shown too. The results showed that in a test network life of leach protocol was 1100 rounds, whereas network life of proposed method was 3100 rounds. پرونده مقاله

Routingincomputernetworks has played a special role in recent years.The cause ofthisisthe role ofroutingina performanceof the networks.The quality ofserviceand securityis one of the most importantchallengesin routing due tolack of reliablemethods.Routers use routing algorithms to findthe best route toa particulardestination. When talkingabout the bestpath, we consider parameters likethe number of hops, change times, and communication cost of sending data packet. In this study we will try to improve the routing operations using local and global smart factors. The Ants Colony Algorithm is a multi-factor solution for optimization issues. This solution has models based on the ants’ collective intelligence and has attracted some users in computer networks through converting to an efficient technology. Although the Ant is a simple insect, but a colony of them are able to perform useful tasks such as finding the shortest path to the food source and to share this information with other ants through leaving back a chemical material called pheromone. This algorithm consists of three stages. The first phase is clustering nodes of the network to smaller colonies. This phase is conducted by using learning automata network in accordance with the need of the network; For example, putting nodes in one cluster which will have more close relations in near future. The second phase is finding the routes of the network by ants, and the third phase is sending network پرونده مقاله

Exploiting multimodal information like acceleration and heart rate is a promising method to achieve human action recognition. A semi-supervised action recognition approach AUCC (Action Understanding with Combinational Classifier) using the diversity of base classifiers to create a high-quality ensemble for multimodal human action recognition is proposed in this paper. Furthermore, both labeled and unlabeled data are applied to obtain the diversity measure from multimodal human action recognition. Any classifiers can be applied by AUCC as its base classifier to create the human action recognition model, and the diversity of classifier ensemble is embedded in the error function of the model. The model’s error is decayed and back-propagated to the basic classifiers through each iteration. The basic classifiers’ weights are acquired during creation of the ensemble to guarantee the appropriate total accuracy of the model. Considerable experiments have been done during creation of the ensemble. Extensive experiments show the effectiveness of the offered method and suggest its superiority of exploiting multimodal signals. پرونده مقاله

Nowadays, it is common to find optimal point of the dynamic problem; dynamic problems whose optimal point changes over time require algorithms which dynamically adapt the search space instability. In the most of them, the exploitation of some information from the past allows to quickly adapt after an environmental change (some optimal points change). This is the idea underlining the use of memory in the field, which involves key design issues concerning the memory content, the process of memory update, and the process of memory retrieval. With use of the Aging Best Solution and Keeping Diversity in Population, the speed convergence of algorithm can be increased. This article presents a genetic algorithm based on memory for dealing with dynamic optimization problems and focuses on explicit placement of memory schemes, and performs a comprehensive analysis on current design of Moving Peaks Benchmark (MPB) problem. The MPB problem is the most proper benchmark for simulation of dynamic environments. The experimental study show the efficiency of the proposed approach for solving dynamic optimization problems in comparison with other algorithms presented in the literature. پرونده مقاله

Faults in computer networks may result in millions of dollars in cost. Faults in a network need to be localized and repaired to keep the health of the network. Fault management systems are used to keep today’s complex networks running without significant cost, either by using active techniques or passive techniques. In this paper, we propose a novel approach based on imperialist competitive algorithm using passive techniques to localize faults in computer networks. The presented approach using end-to-end data detect that there are faults on the network, and then we use imperialist competitive algorithm (ICA) to localize faults on the network. The aim of proposed approach is to minimize the cost of localization of faults in the network. According to simulation results, our algorithm is better than other state-of-the-art approaches that localize and repair all faults in a network. پرونده مقاله

In this paper, principles and existing feature selection methods for classifying and clustering data be introduced. To that end, categorizing frameworks for finding selected subsets, namely, search-based and non-search based procedures as well as evaluation criteria and data mining tasks are discussed. In the following, a platform is developed as an intermediate step toward developing an intelligent feature selection system, involving crucial, decisive and effective factors in feature selection process. The procedure increases accuracy in classification and goodness of clusters. Finally, some of the problems and challenges facing the current and future feature selection processing are also discussed. پرونده مقاله