مجله رایانش نرم و فناوری اطلاعات
سال یکم شماره 4 (زمستان 1391)

The university course timetabling problem (UCTP) is a hard combinatorial optimization problem. Because the search space of this problem is very large، human timetabling is a sophisticated job; as a result، there have been many attentions over the recent years from computer researchers to present an automated timetabling approach for UCTP. In a general point of view، UCTP is a problem in which some courses must be assigned to valid timeslots. There are some constrains and preferences in this problem. Most constraints and preferences come from curriculum of teachers and students. If a timetable satisfies all constraints، it is called a feasible timetable. In UCTP، the outcome timetable must be a feasible timetable. The less violation preferences have، the better timetable would be created. Since UCTP is a constraint satisfaction problem (CSP)، in this paper in order to create feasible timetables، a novel CSP-based algorithm is presented. In addition، a genetic algorithm (GA) combined with local search methods is used to optimize the feasible timetables. The experimental results show that the suggested method produces appropriate results in a reasonable time considering a real data with large state space as input data.

In recent years, hybrid algorithms (HAs) have been successfully applied for solving decision and optimization problems. Nevertheless, selecting good algorithms for hybridization has been a crucial issue in HAs. In this paper, a new hybrid algorithm composed of gravitational search algorithm (GSA) and the proposed adaptive stochastic search (ASS) method is introduced. These effective search algorithmsprovide a good trade-off between exploration and exploitation. The performance of the proposed HA is evaluated in the field of numerical function optimization on 23 standard benchmark functions and also on a practical optimization problem, optimal approximation of linear systems. The results are compared with those of some well-known HAs and confirm the efficiency of the proposed method in solving various nonlinear test functions.

The computation of capacity for discrete memoryless channels can be efficiently solved by using the Arimoto-Blahut (AB) iterative algorithm. However, the extension of this algorithm to compute the capacity for channels with Side Information (SI) is not straightforward, i.e. for channel with noncausal side information at the transmitter, because generally it is hard to evaluate the rates and capacities having auxiliary random variables. In this paper, we exploit an alternative reformulation of differential evolution optimization method to compute the capacity for channels with causal side information at the transmitter, and introduce an efficient algorithm to compute the capacity of these channels. Also, we extend the AB algorithm to compute the capacity of channels with side information at the receiver.

Nowadays, steganography methods are significantly in progress. Therefore, steganalysis systems are required to be designed such that they have to recognize the smallest embedded information. Steganalysis is performed based on feature extraction and classification system. The extracted features from the images are needed to have high sensitivity for embedding and manipulation. In this paper, some appropriate features based on Markov chain process are presented. Also, based on the extracted features that often have large dimensions, a classifier is required to perform accurate and optimal classification. Support Vector Machine (SVM) classifier is used for this purpose. Also, a feature selection method based on variance is used to reduce the features dimensions. The obtained results show that the proposed method provides better performance compared to two general methods called Sdam and Spam.

IHS method for image fusion is one of the conventional methods with a simple theory. Implementation of this method is simple and effective from computational point of view. But this method causes the spectral content of the fused image to get disturbed in comparison with the spectral criteria of the main multispectral image. In this article a statistical image fusion method is combined with the conventional IHS method to improve its spectral quality and overcome its deficiency. Some spectral indices were used for evaluation of this combined method and compared with the conventional IHS and the used statistical fusion methods. This combined method improves the spectral quality of the fused image and eliminates the existing shortcomings of the spectral disturbing in fused image produced by IHS method.