With the explosive growth in amount of information, it is highly required to utilize tools and methods in order to search, filter and manage resources. One of the major problems in text classification relates to the high dimensional feature spaces. Therefore, the main goal of text classification is to reduce the dimensionality of features space. There are many feature selection methods. However, only a few methods are utilized for huge text classification problems. In this paper, we propose a new wrapper method based on Particle Swarm Optimization (PSO) algorithm and Support Vector Machine (SVM). We combine it with Learning Automata in order to make it more efficient. This helps to select better features using the reward and penalty system of automata. To evaluate the efficiency of the proposed method, we compare it with a method which selects features based on Genetic Algorithm over the Reuters-21578 dataset. The simulation results show that our proposed algorithm works more efficiently. تفاصيل المقالة

با توجه به هزینه کمتر و انعطاف‌پذیری بیشتر، انتقال صدا از طریق پروتکل اینترنت (VoIP) به طور گسترده‌ای در ارتباطات راه دور استفاده می‌شود. تنوع پایانه‌های VoIP باعث آسیب‌پذیری آنها می‌شود. یک راه متداول برای ایمن‌سازی VoIP، شامل تشخیص نفوذ مبتنی بر یادگیری ماشین است. با توجه به تنوع ترافیک و عدم وجود برچسب کلاس برای آموزش سیستم‌های تشخیص نفوذ (IDS) در بسیاری از مواقع، بر رویکردهای خوشه‌بندی (یادگیری بدون ناظر) متمرکز شده‌اند. اما سیستم‌های خوشه‌بندی منفرد نمی‌توانند تنوع مقادیر ویژگی‌ها را به خوبی پوشش دهند و برخی از نمونه‌های ترافیک ممکن است به عنوان نقاط پرت شناسایی شوند. مدل پیشنهادی، به‌عنوان یک رویکرد تجمیعی برای حل این مسائل، روی استفاده از الگوریتم خوشه‌بندی دومرحله‌ای متمرکز شده و سعی می‌کند با ایجاد بهبودی در آن، فرآیند تشخیص نفوذ مبتنی بر خوشه‌بندی را بهبود دهد. علاوه بر این، با توجه به اهمیت فرآیند انتخاب ویژگی، ترکیبی از الگوریتم شبیه‌سازی تبرید (SA) و شبکه عصبی پرسپترون چندلایه (MLP)، برای شناسایی ویژگی‌های برتر مورد استفاده در خوشه‌بندی بسته‌های VoIP، در قالب بسته‌های عادی یا حمله انکار سرویس (DoS)، حمله کاربر به ریشه (U2R)، حمله کاربر از راه دور (R2L) و حمله پویش‌گر مورد بهره‌‌برداری قرار گرفته است. بر اساس نتایج ارزیابی بر روی مجموعه داده "آزمایشگاه امنیت شبکه– کشف دانش در پایگاه‌های داده‌ای" ( NSL-KDD)، توسط نرم‌افزار متلب، انتخاب ویژگی پیشنهادی با کاهش ویژگی‌ها به 10 و 8، زمان آموزش و آزمایش را به‌ترتیب 77 درصد و 80 درصد کاهش می‌دهد. همچنین در مقایسه با تعدادی از مطالعات قبلی، IDS پیشنهادی بهبود متوسطی معادل 34/3 درصد، 17/14 درصد و 87/32 درصد را به‌ترتیب در دقت، نرخ تشخیص و معیار F نشان می‌دهد. تفاصيل المقالة

Moving towards nanometer scales, Quantum-dot Cellular Automata (QCA) technology emerged as a novel solution, which can be a suitable replacement for complementary metal-oxide-semiconductor (CMOS) technology. The 3-input majority function and inverter gate are fundamental gates in the QCA technology, which all logical functions are produced based on them. Like CMOS technology, making the basic computational element such as an adder with QCA technology, is considered as one of the most important issues that extensive research have been done about it. In this paper, a new QCA full-adder based on coupled majority-minority and 5-input majority gates is introduced which its novel structure, appropriate design technique selection and its arrangement make it very suitable. The experimental results showed that the proposed QCA full-adder makes only 48 cells and the first output is obtained in the 0.05clock. Therefore, the presented QCA full-adder improves the number of cells and gains a speedup rate of 33% in comparison with the best previous robust QCA full-adders. In addition, temperature analysis of the QCA full-adders shows that our design is more robust compared with other suggested QCA full-adders. تفاصيل المقالة

Speaker verification is the process of accepting or rejecting claimed identity in terms of its sound features. A speaker verification system can be used for numerous security systems, including bank account accessing, getting to security points, criminology and etc. When a speaker verification system wants to check the identity of individuals remotely, it confronts problems such as noise effect on speech signal and also identity falsification with speech synthesis. In this system, we have proposed a new speaker verification system based on Multi Model GMM, called SV-MMGMM, in which all speakers are divided into seven different age groups, and then an isolated GMM model for each group is created; instead of one model for all speakers. In order to evaluate, the proposed method has been compared with several speaker verification systems based on Naïve, SVM, Random Forest, Ensemble and basic GMM. Experimental results show that the proposed method has so better efficiency than others. تفاصيل المقالة

Nowadays diabetes disease is one of the main problems of health domain and it’s known as the fourth factor of death in the world. The main problem with this dangerous disease is the late or weak diagnosis. The reason of weak diagnosis is because sometimes doctors aren’t able to select the right patterns or they can’t use the standard patterns very well, so the outcome is that the disease will be diagnosed by the patients when it has become late for controlling or curing it. Therefore, implementing a method which can help each person to have an authentic diagnosis of being or not being affected to this disease; can be an important step for prevention and controlling this special disease at the beginning of it. In this paper, a new method is presented for diagnosing diabetes disease which is able to extract the proper knowledge by helping to cluster and analyze the training patterns, after that in recognition phase it can diagnose diabetes disease precisely and fast via a fuzzy reward-penalty mechanism. For evaluating the proposed method, PIMA dataset has been used. The experimental results show that the proposed method has a better performance compared to other existing methods. تفاصيل المقالة