انجمن کامپیوتر ایران

درك سازوكار مغز به عنوان عامل هوشمندي موجودات زنده قرنها است كه ذهن بشر را مشغول كرده است. محققين در مسير پاسخ به اين سوال دو رويكرد كل به جزء و جزء به كل را پيش گرفتهاند. در اين مقاله به معرفي مدلي حاصل از اتصال كارآمدترين نظريات اين دو رويكرد خواهيم پرداخت. مدل ارائه شده تواناييهاي جالب توجهي از جمله يادگيري خودكار و پردازش بلادرنگ دنبالههاي زماني را داراست. همچنين اين مقاله نشان ميدهد كه مدل مذكور با دادههاي زيست شناختي مغز نيز انطباق قابل قبولي دارد. در انتها، كارآمدي مدل ارائه شده در يك آزمايش دنياي واقعي بررسي ميشود. مدل پيشنهادي اولين قدم به سمت ساخت سيستمهاي هوشمند خلاق و داراي قدرت سازشپذيري بالا به حساب ميآيد.

شناسايي و دسته بند ي دقيق جريانهاي ترافيكي شبكه بر اساس پروتكل توليد كننده آنها، يكي از المانهاي اصلي مديريت شبكه محسوب مي شود . تاكنون تكنيك هاي مختلفي در اين زمينه مانند شماره پورت و بررسي بخش داده اي بسته ها ارائه شده است . امروزه بسياري از برنامه ها (مانند (P2P با استفاده از شماره پورتهاي پويا و تكنيك هاي رمزنگاري سبب ناكارآمدي تكنيكهاي فوق شد ه اند . تكنيك ديگر در اين زمينه، استفاده ازاطلاعات آماري لايه انتقال جريانهاي عبوري مي باشد . در اين مقاله يك كنترلر فازي براي دسته بندي جريانهاي ترافيكي پيشنهاد شده است . در اين كار اطلاعات آماري لايه انتقال جريانهاي عبوري، بعنوان پارامترهاي دسته بندي استفاده شده است . نتايج آزمايشات نشان مي دهند كه مكانيزم پيشنهادي بطور متوسط داراي كارايي كلي 94% بوده و 98/77% جريانها را شناسايي كرده است . در حالي كه كارايي كلي الگوريتم هاي K-Means و DBSCAN به ترتيب 84% و 75/6 % مي باشد.

تخمين تعداد سيگنالهاي طيف گسترده پرش فركانسي وارده به يك آنتن يكي از مسايل مورد توجه است. روش هاي تجزيه مقادير ويژه مورد استفاده براي تخمين تعداد كاربران براي SNR هاي پايين از كارايي لازم برخوردار نيست. روش كار به اين صورت است كه ابتدا با استفاده از تحليل آنتروپي يك زير مجموعه ديتاي بدون پرش از سيگنال جدا شده و سپس با استفاده از تجزيه مقادير ويژه تعداد كاربران بدست ميآيد. در اين مقاله سعي شده است تا مشكل تخمين تعداد كاربران در SNR هاي پايين با اعمال يك بلوك حذف نويز به آنتروپي بدست آمده از سيگنال ورودي بهبود بخشيده شود.

برنامه نويسي ژنتيك، از جمله الگوريتم هاي تكاملي است كه توانايي زيادي، نسبت به اغلب روش هاي يادگيري ماشين دارا م ي باشد . مشكل اين روش، زمانبر بودن اجراي آن است . در اين مقاله، هدف استفاده از برنامه نويسي ژنتيك و تطبيق آن به منظور تصديق هويت گوينده و بررسي هويت كاربران از طريق صداي آنهاست . ايده هاي متعددي را پيشنهاد نموده ايم تا به كمك آنها بتوانيم ضمن افزايش كارايي روش برنام ه نويسي ژنتيك براي تصديق هويت گويند ه، سرعت آموزش مدلهاي گويندگان را كه معمولا در روش برنام ه نويسي ژنتيك بسيار زمانبر است را افزايش دهيم. براي اين منظور سعي شده است تا با روشهايي چون خوشه بندي به كمك روشهاي چندي سازي برداري و نيز با استفاده از توابع گوسي بدست آمده از روش مدل مخلوط گوسي GMM به جاي بردارهاي ويژگي داده هاي آموزشي ، حجم داده هاي آموزشي را كاهش دهيم و بدين ترتيب بر سرعت ساخت مدلهاي حاصل از برنامه نويسي ژنتيك بيافزائيم. نتايج بدست آمده نشان مي دهند كه استفاده از ميانگين هاي مدل هاي مخلوط گوسي حاصل از داده هاي آموزشي گوينده خودي و گويندگان غيرخودي به جاي استفاده مستقيم از داده هاي آموزشي و بطور مشابه استفاده از ميانگين هاي مدل مخلوط گوسي حاصل از داده هاي آزمايشي، منجر به دقت خوبي در تصديق هويت گويندگان و نيز افزايش سرعتي در حد 20 برابر (از 5 ساعت به 15 دقيقه) در آموزش مدلهاي گويندگان به كمك روش برنامه نويسي ژنتيك مي گردد.

In wireless sensor network often micro-battery with very limited power provides the energy of sensor nodes. Since sensors are usually utilized in remote or hostile environments, recharging or replacing the battery of the sensors is something quite undesirable or even impossible. Thus long system lifetime is a must. Sleep scheduling is a mechanism in wireless sensor network to save energy. In this paper, we propose an energyefficient distributed scheduling method considering mobile target tracking also called dynamic target coverage. The algorithm is based on cellular learning automata. In this algorithm, each node is equipped with a learning automaton which will learn (schedule) the proper on and off times of that node based on the movement nature of a single moving target. To evaluate the proposed method it is tested under straight with constant velocity movement model of target. The results of experimentations have shown that the proposed scheduling algorithm outperforms two existing dynamic target coverage scheduling methods.

Grid environment is being a service oriented infrastructure in which many heterogeneous resources participate for providing the high performance computation. On of bug issue in the grid environment is the vagueness and uncertainty between advertised resources and requested resources. In this work we propose a solution for the vagueness and uncertainty problems based on rough set theory. Here you can see how the rough set theory is developed to deal with the problem. We also report the result of the solution obtained from the simulation in Gridsim simulator. The comparison has been made between the proposed method and UDDI and OWL-S combined method. Rough set theory shows much better precision for the cases with vagueness and uncertainty.

Asynchronous protocols exhibit various noise robustness and when used in GALS NoC links, they can directly affect the signal integrity. In this paper we study the noise robustness of two well-known asynchronous protocols, namely Dual-Rail (DRP) and Bundled-Data (BDP) in the GALS NoC links, and subsequently confirm our claims through simulations. We apply an enhanced version of BDP and DRP to 32/64 parallel line links, show results in terms of noise robustness using global interconnect features, specified in the ITRS roadmap for 32nm technology. The simulation results for two thousand random generated inputs show that the number and the amplitude of noise glitches over ‘0’ state lines as well as the required threshold voltage needed for avoiding errors in BDP link are much lower than in DRP's. Therefore, BDP links can present better signal integrity features and have less overhead compared to DRP's, employing only some simple noise reduction techniques and more timing adjustment effort.

Discriminative subspace analysis is a popular approach for a variety of applications. There is a growing interest in subspace learning techniques for face recognition. Principal component analysis (PCA) and eigenfaces are two important subspace analysis methods have been widely applied in a variety of areas. However, the excessive dimension of data space often causes the curse of dimensionality dilemma, expensive computational cost, and sometimes the singularity problem. In this paper, a new supervised discriminative subspace analysis is presented by encoding face image as a high order general tensor. As face space can be considered as a nonlinear submanifold embedded in the tensor space, a decomposition method called Tucker tensor is used which can effectively decomposes this sparse space. The performance of the proposed method is compared with that of eigenface, Fisherface, tensor LPP, and ORO4×2 on ORL and Weizermann databases. Conducted experimental results show the superiority of the proposed method.

Being able to predict the coming seizure can impressively improve the quality of the patients' lives since they can be warned to avoid doing risky activities via a prediction system. Here, a locally linear neuro fuzzy model is used to predict the EEG time series. Subsequently, this model is utilized in accompany with Singular Spectrum Analysis for prediction. Afterward, an information theoretic criterion is used to select a reliable subset of input variables which contain more information about the target signal. Comparison of three mentioned methods on one hand shows that SSA enables our prediction model to extract the main patterns of the EEG signal and highly improves the prediction accuracy. On the other hand, applying the method of channel selection to the model yields more accurate prediction. It is shown that fusion of some certain signals provides more information about the target and considerably improves the prediction ability.

In this paper we propose several methods for partitioning, the process of grouping members of population to different memeplexes, in a shuffled frog leaping algorithm. These proposed methods divide the population in terms of the value of cost function or the geometric position of members or quite random partitioning. The proposed methods are evaluated on several low and high dimensional benchmark functions. The obtained results on low dimensional functions demonstrate that geometric partitioning methods have the best success rate and the fastest performance. Also on high dimensional functions, however using of the geometric partitioning methods for the partitioning stage of the SFL algorithm lead to a better success rate but these methods are more time consuming than other partitioning methods.

In this paper, we proposed a new feature subset selection approach. In proposed approach first, the entire dataset are classified and the best number of clusters over it are found according to silhouette value. Then according to this value, each feature is alone classified with the same cluster number and accordingly the proposed entropy fuzzy measure is found for them. We examine our method on some traditional datasets. The results show a good performance of proposed method.

Uncertainty reasoning has been identified as an important and challenging issue in the database research. Many logic frameworks have been proposed to represent and reason about uncertainty in deductive databases. On the basis of the way in which uncertainties are associated with the facts and rules in programs, the approaches of these frameworks have been classified into “annotation based (AB)” and “implication based (IB).” When extending both frameworks with certainty constraints, they become equivalent in terms of expressive power. In this paper, we propose a uniform environment to evaluate and experiment with logic programs in AB and IB frameworks at the same time. We also extend the existing query processing to handle certainty constraints and we carry out experiments to evaluate its performance. Our experiments and results indicate that the proposed techniques yield tools that are capable to reason with uncertainty.