Sharif Digital Repository

Transfer Learning is one of the most important fields in the Machine Learning area. Respect to the advances that we have seen in the Computer Science, especially in the Machine Learning area, we need a tool that can transfer learnings from different domains to each other. As data distribution varies, many statistical models require restructuring using new training data. In many applications, re-assembling training data and re-structuring models is inefficient and costly, so reducing the need for this practice seems appropriate. In these cases, knowledge transfer or learning transfer between domains may be desirable. For example, in the area of the B rain-Computer Interface, when it... 

The person identification by Electroencephalographic (EEG) signals has attracted the researchers’ great attention in recent years and lots of investigations have been developed. An identification system seeks to identify a person in a database. The advantage of using EEG signals for person identification is the difficulty in generating artificial signals for imposters. But more works need to be done to use EEG based biometric in real-life and this thesis is one of them. In this project we classify the EEG signals for person identification using AdaBoost algorithm. Adaptive boosting (AdaBoost) is a machine learning technique for pattern classification in which the performance of the weak... 

Human face detection is an important research area with several applications such as human computer interface (HCI), face recognition, surveillance systems, security systems, and content-based image retrieval (CBIR). Face detection problem can be stated as “determining whether there are human faces in the image” and if there are “returning the location of each human face in the image” regardless of its position, size, scale, orientation, and lighting condition. In this thesis, we have proposed a new face detection method which combines the AdaBoost algorithm with skin color information and support vector machine (SVM). First, a cascade classifier based on AdaBoost is used to detect faces in... 

Because of deregulation and economic issues, power systems operate closer to their stability limits than before. So, security assessment of power system has gained a lot of attention in power system studies. Transient stability and its real time assessment is a great concern in power system security. Among all different methods for transient stability assessment, Automatic Learning methods are used the most in real time assessment problems. These methods can provide fast assessment of transient stability, sensitivity analysis and means of control of the transient stability phenomena. In this paper Adaboost algorithm is used as a brand new method in transient stability assessment of a medium... 

Biometrics has been long known to recognize persons based on their physical and behavioral characteristics. Face recognition (FR) is one of such biometrics that has received a considerable attention in recent years both from the industry and research communities. As the boosting framework has shown good performance in face recognition, it has been adopted in this work. This thesis deals with pattern recognition methods such as linear discriminant analysis (LDA) and machine learning approaches such as boosting which are integrated to overcome the technical limitation of existing FR methods. However, LDA-based methods often suffer from the so-called “small-sample-size” (SSS) problem arising... 

Online voltage stability assessment is essential for preventing the failure in power systems. Voltage instability in power systems usually occurs in consequence of contingency situations or continuous load increasing or both of them. A new method for online voltage stability assessment using the data acquired from phasor measurement units (PMUs) and offline trained decision trees is presented in this thesis. The index of voltage stability margin (VSM) is used for security assessment and considering WSCC criterion, different operation points are divided in two classes: safe and unsafe. In order to obtain the voltage stability margin and generate the database, a new method which has been named... 

This thesis discusses a real-time vision-based framework for the purpose of hand region detection and hand gesture recognition. Our proposed methods include detecting hand regions in the cluttered background, based on Viola-Jones object detection algorithm and improving the classification of detected hand gestures regions in a novel contour-based framework. Our studies have demonstrated that deformability and high degree of freedom (DoF) of human hand as a non-rigid object besides diversity of skin color types, undeniable effect of cluttered background complexity, scalability and being robustness against rotation are the main reasons for considering some simplifications in visionbased...