Journal of Computer and Robotics
Volume:6 Issue: 2, Winter and Spring 2013

In 2D Soccer Simulation league, agents will decide based on information and data in their model. Effective decisions need to have world model information without any noise and missing data; however, there are few solutions to omit noise in world model data; so we should find efficient ways to reduce the effect of noise when making decisions. In this article we evaluate some simple solutions when making defense decisions and try to find a solution based on message-passing to coordinating agents in defense situations. Our experimental results showed that in each situation one of the agents has a better view than others, so that agent can send messages to the others and provide needed information for doing defense behavior(ex: block behavior or clear ball behavior). Finally, we implement our solution based on Agent2D, version 3 and compare that with other solutions implemented in Cyrus2014 and Marlik2013 Soccer 2D simulation teams.

By extending wireless networks and because of their different nature, some attacks appear in these networks which did not exist in wired networks. Security is a serious challenge for actual implementation in wireless networks. Due to lack of the fixed infrastructure and also because of security holes in routing protocols in mobile ad hoc networks, these networks are not protected against attacks. For example in black hole attack, an attacker catches packets and throw them away, instead of forwarding them to their destinations. By using wireless intrusion detection systems, wireless networks can be protected. In this study, we introduce a new intrusion detection system to encounter black hole attack. This system is based on a combination of anomaly based intrusion detection (ABID) and specification based intrusion detection (SBID), we also use a new intrusion response. The analysis of simulation results (with NS-2) show that our method is success by using three measures: throughput, packet loss rate and packet delivery rate in comparing with ABID and SBID.

Multi agent Markov decision processes (MMDPs), as the generalization of Markov decision processes to the multi agent case, have long been used for modeling multi agent system and are used as a suitable framework for Multi agent Reinforcement Learning. In this paper, a generalized learning automata based algorithm for finding optimal policies in MMDP is proposed. In the proposed algorithm, MMDP problem is described as a directed graph in which the nodes are the states of the problem, and the directed edges represent the actions that result in transition from one state to another. Each state of the environment is equipped with a generalized learning automaton whose actions are moving to different adjacent states of that state. Each agent moves from one state to another and tries to reach the goal state. In each state, the agent chooses its next transition with help of the generalized learning automaton in that state. The experimental results have shown that the proposed algorithm have better learning performance in terms of the speed of reaching the optimal policy as compared to existing learning algorithms.

Prediction, diagnosis, recovery and recurrence of the breast cancer among the patients are always one of the most important challenges for explorers and scientists. Nowadays by using of the bioinformatics sciences, these challenges can be eliminated by using of the previous information of patients records. In this paper has been used adaptive nero fuzzy inference system and data mining techniques for processing of input data and the educational combined algorithm for arranging of parameters of input functions.  It has used also the downward gradient algorithm for arranging of unlined input parameters and the algorithm of the least of squares for arranging of lined output parameters. It has been used the data the institute of oncology Ljubljana of Yugoslavia that contain the information of 1090 the breast cancer patients. The results show the suggesting system has 89% accuracy in the diagnosis of progressing the breast cancer, which has improved by compared with neural network classification method.

Nowadays, Hidden Markov models are extensively utilized for modeling stochastic processes. These models help researchers establish and implement the desired theoretical foundations using Markov algorithms such as Forward one. however, Using Stability hypothesis and the mean statistic for determining the values of Markov functions on unstable statistical data set has led to a significant reduction in the accuracy of Markov algorithms including Forward algorithm used in solving Evaluation problems. The model’s parameters such as the occurrence probability of observation symbol being produced by state, varies directly among the successive events. Since the probability value of the above-mentioned parameter plays an important role in the accurate Evaluation and assessment of the probability of observations’ occurrence in the Evaluation problem by Forward algorithm, the variations between events and observations generated by the States should be automatically extracted. In order to achieve this, the current paper proposes an adaptive parameter for event probability in order to match and adjust the variations in the parameter after each event during the lifetime of Forward algorithm. The results of the experiments on a real set of data indicates the superior performance of the proposed method compared to other conventional methods regarding their accuracy.

Most of the contour detection methods suffers from some drawbacks such as noise, occlusion of objects, shifting, scaling and rotation of objects in image which they suppress the recognition accuracy. To solve the problem, this paper utilizes Zernike Moment (ZM) and Pseudo Zernike Moment (PZM) to extract object contour features in all situations such as rotation, scaling and shifting of object inside the image. The proposed method consist of three steps: first step employs Line Detection with Contours (LDC) in order to find the object region based on the connected components objects inside the image. In the second step, PZM is applied on the detected object regions to extract feature vector. Regarding to investigate the effectiveness of classifier at the final stage, the SVM and KNN classifiers are employed to extract final object contours. Experimental results on Caltech-101 dataset shows that classification rate is improved to 96.46%. In comparison to the former contour detectors, that proves the ability of the proposed method to detect object boundary in the most of the contour’s changes such as rotation or scaling.