JOURNAL OF SOFT COMPUTING AND INFORMATION TECHNOLOGY (JSCIT)
Year:2019 | Volume:8 | Issue:4 |Papers Count:9

In this paper, we consider the issue of automatic and unsupervised class-manifold selection in a multi-view multi-manifold space. General multi-manifold learning methods achieve multiple independent manifolds, so it is challenging for them to adjust the intra-class local manifold information and global inter-class discriminative structure. In this paper, we propose a multi-manifold embedding method, which can explicitly obtain multi-view multi-manifold structure while considering both intra-class compactness and inter-class separability without using the class label information. Furthermore, to the generalization of embedding to novel points, known as the out-of-sample extension problem in multi-view multi-manifold learning, we propose a supervised method for building a regularized map that provides an out-of-sample extension for general multi-view multi-manifold learning studied in the context of classification. Experimental results on face and object images demonstrate the potential of the proposed method for the classification of multi-view multi-manifold data sets and the proposed out-of-sample extension algorithm for the classification of manifold-modeled data sets.

In this paper, an improved metaheuristic algorithm based on Grey Wolf Optimizer algorithm is proposed for solve optimization problems. In the proposed algorithm, the weakest of wolves would be excluded from the population and included with other wolves from the initial population. The choice of placed wolves would be random or fitness basis. In this algorithm, the spatial fitness of the particles is studied in each iteration, and in the case of improving the fitness basis, the wolves are moving toward the target. Otherwise, they remain in the last fit state. This algorithm is designed to improve the search performance against various issues, increase convergence speed, and avoid local optimal. Simulation has been done in Matlab software and it has been implemented with 23 different optimization mathematical functions. By examining the results and comparing the results of the results obtained from the new algorithm, Grey wolf optimizer algorithm, and several other algorithms, we conclude that by adjusting the parameters, the performed improvements have a significant effect on the performance of the algorithm on different functions.

Intelligent video surveillance is one of the main applications in machine vision. People re-identification as part of these systems is of particular importance. Indeed, the accuracy in this part improves the efficiency of many types of monitoring algorithms. The re-identification task in human mind is performed consciously and is based on a prior knowledge of the 3D attributes of the human body. One of these attributes is the orientation of the body relative to the camera. In other words, a human supervisor at the matching stage uses the angle information to estimate the appearance of the person at different angles. In this research, the above process is modeled. Thus, in this research, first the body orientation is automatically extracted in the image, and accordingly, upper part of the clothing is extracted, which might be hidden at different angles. Removing or re-sampling these areas reduces the destructive effects on the matching process. For evaluation and comparison, the proposed method was used in two of the efficient re-identification algorithms. Experiments were performed on the ViPer dataset and the results show %1. 3 percent improvements in the recognition rate for 316 people.

In today's competitive world, reducing the distribution costs is an important issue that lies in the forefront of industry managers' thinking. A large percentage of the overall price of a given product belongs to distribution costs. Accordingly, eliminating unnecessary trips and optimizing traversed routes are considered to be one of the ideal solutions to reduce distribution costs. The aim of this study is to develop a dynamic distribution model in cold supply chains of dairy products by using an enhanced hybrid metaheuristic approach based on ant colony optimization. The proposed distribution model is defined according to the capacitated vehicle routing problem (CVRP) where vehicles' routes are not specified for the distribution of dairy products and depending on the volume of orders requested by each trailer on a given day, the minimum number of vehicles along with optimal distribution routes are determined. We assess the efficiency of our proposed distribution model by generating some test data inspired by the data collected from Qazvin Pegah Dairy Company in five different levels of distribution. The evaluation criterion is compared with the results of the current static distribution system. The experimental results indicate that the proposed dynamic distribution model exhibits more efficiency and flexibility than the static distribution system in terms of transportation costs, manpower and handling costs due to fewer number of vehicles employed, shorter mileage traversed and less fuel consumed.

Today, online social networks (OSNs) have gained an important role in everyday human life. With the ever Increasing use of different types of OSNs as well as the extension of social interactions, the role of trust has become significantly more important. The success of the social network depends on the correct analysis of social situations, interactions and applying appropriate approaches according to each specific situation. Trust and distrust are two considerable factors to analyze these networks. The main purpose of this research has been to improve the accuracy of calculating trust and distrust based on the theoretical foundation of social interactions as well as the decision making in the social environment. By reviewing the body of literature in the fields of sociology and psychology focusing on trust and distrust in the social environments, we have concluded that distrust information, as well as trust, plays an important role in social interactions and decision making. The construct of trust and distrust are independent but they affect one another. This independent identity means that they are counting on the basis of the attributes and related factors. The aim of this research has been to model the co-existence of trust and distrust in maintaining the independence of each identity while considering different criteria for each of them. Based on the theorem of subjective logic we have modeled the coexistence of trust and distrust. So far, the existing models have only focused on trust information and its corresponding calculations. There are other works that have focused on distrust Information. But, in these models, distrust information has been gathered directly by users or calculated based on trust information sources. Therefore, in this research, we have proposed the calculation of trust and distrust based on individual and entangled trust and distrust formation factors. These factors are used in the decision making process. The results of the performed evaluations demonstrate that the proposed model has generated more accurate outcomes in calculating trust and distrust within a trust-based decision making context compared to other existing models.

The existence of noise in image reduces its quality and hinders analysis of the image. Image noise reduction techniques are often accompanied with artifact, especially in facing with strong noise. Since sensitivity of human visual system is not alike in all areas of image, i. e. smooth and nonsmooth areas, noise removal can be performed considering the textual information of the image. The proposed approach intends to earnestly remove noise from the smooth region as it is more obvious to human visual system. Indeed, the filtered image produces less artifact in nonsoomth region, as the noise is superficially removed from the nonsmooth region. In the proposed method, the image is segmented into smooth and non-smooth regions using entropy information of the image. Then to remove the noise from each region, the diffusion filter with different parameters is used. The proposed method not only removes the noise but also preserves the edges and details of the image. The proposed method was evaluated using several noisy images and images from CSIQ and IVC databases. According to subjective and objective quality results, accuracy of the proposed method in Gaussian noise reduction is better than the previous works.

Skin cancer affects millions of people all around the world. If skin cancer is detected in the early stages, the survival rate is very high. So, computer-aided diagnosis (CAD) systems are being developed to help dermatologists in early and accurate diagnosis. Segmentation is the first and most important step in the auto diagnosis systems. The purpose of this paper is to introduce a new method based on geometric active contours that combines texture and color information to separate the lesion area from healthy skin. The innovation of this paper is the way that, color and texture information are combined together to define the speed function and the use of texture features in the form of an image. To evaluate the proposed method, two databases including dermoscopy images, were used. The ISIC2017 database (including 2750 data) and the PH2 database (including 200 data). Experimental results showed that, the proposed algorithm has the highest accuracy (97. 92% for PH2 database and 94. 78% for ISIC2017 test data), sensitivity (97. 83% for PH2 database and 90. 11% for ISIC2017 test data) and specificity (99. 45% for PH2 and 98. 53% for ISIC2017 test data) in comparison with recent state-of-the-art algorithms.

Support vector machine is one of the most powerful tools in the field of supervised machine learning to classify the existed data. In the data that the linear support vector machine does not have the required efficiency in their classification, using the kernel-based support vector machine which is based on the use of feature space instead of the original data is considered. As a result of this structure, nonlinear classification can be provided. One of the challenges in this approach is to increase the computational complexity and ultimately increase in the required time for classification. As such, it is not particularly useful for large data sets. This increasing in computational time is mainly due to the appearance of the kernel in solving the quadratic optimization problem, which we will be able to overcome this problem using the presented low rank approximation in this paper. In this technique, using a truncated Mercer series of the kernel, the quadratic optimization problem in the kernel-based support vector machine is replaced with a much simpler optimization problem. In the new presented approach, the required vector computations and matrix decompositions will be much faster such that these changes lead to faster resolution of the quadratic optimization problem and increase efficiency. Finally, the results of experiments show that using a low rank kernel-based approximation of support vector machine, while keeping the classification performance in an acceptable range, the computational time has been significantly reduced.

proper choice for descripting images captured by ordinary optic sensors. In order to cover all spectrum and extracting better features filter banks are usually used. Although there is different scales and orientations in filter bank, but using proper values for other parameters such as maximum frequency, filters’ dimension and length of arc can effectively impact on final result. In this paper Meta-heuristic methods are used to estimate optimum values for these parameters. According to obtained results, in identification using Optimum Arc-Gabor Filter Bank (OAGFB) trained by Improved Gravitational Search Algorithm, the average of 1st Rank identification rate is increased from 79. 43 to 95. 71% and in verification by optimizing proposed filter bank using Simulated Annealing the average of Equal Error Rate is decreased from 8. 84 to 5. 12%.