Various techniques are developed for detecting skin cancer. However, the type of maligned skin cancer is still an open problem. The objective of this study is to diagnose melanoma through design and implementation of a computerized image analysis system. The dataset which is used with the proposed system is Hospital Pedro Hispano (PH²). The proposed system begins with preprocessing of images of skin cancer. Then, particle swarm optimization (PSO) is used for detecting the region of interest (ROI). After that, features extraction (geometric, color, and texture) is taken from (ROI). Lastly, features selection and classification are done using a support vector machine (SVM). Results showed that with a data set of 200 images, the sensitivity (SE) and the specificity (SP) reached 100% with a maximum processing time of 0.03 sec.

Cyber criminals are trying to steal personal information from victim¡¦s machine to acquire more benefits by using malware. Antivirus is the most commonly used tool of malware identification, but the frequency of virus definition update is often less than the frequency of new type malware increase. Therefore, we need an effective and fast tool of malware identification in the current environment. In addition to antivirus, software analysis platform is currently one of the ways to identify malware. User could figure out behaviors of software in detail by the analysis report provided by software analysis platform. Most of software analysis platforms only offer an analysis report, user have to identify whether the software is malware or not by them self. This type of report is no help for user if their expertise not enough to find out these behaviors. Some of software analysis platforms which used antivirus can provide information to user about the software is malware or not, but they don¡¦t have the ability of identifying new type malware immediately. According to research and analysis report, we generalized differences in file and registry activities of normal software and malware and defined malware classification features from these differences. We adopted Support Vector Machine¡]SVM¡^as our algorithm of classification to build and test three classifiers which can identify normal software and malware. After several experimental evaluations, we confirmed that the identification rate of malware was up to 97.6%. Finally, we compared the performance of our classifiers with ThreatExpert, and the result show that the performance of our classifiers is as well as ThreatExpert.

No description available.

The problem of stress assignment is to identify which syllables are phonetically more prominent than the others in a word. It is not only of theoretical interests to linguists but also very important to Text-to-Speech systems in terms of both accuracy and naturalness of pronunciation. Besides providing an in-depth survey of existing stress assignment algorithms in the fields of linguistics and speech generation, this thesis presents a ranking approach to stress assignment for both letters and phonemes. The final system is language independent and clearly outperforms all previous systems. When combined with a current state of art Letter-to-Phoneme system, error rate in stress assignment is reduced by up to 40%

Emotions have significant importance in human life in learning, decision-making, daily interaction, and perception of the surrounding environment. Hence, it has become very essential to detect and recognize a person's emotional states and to build a connection between humans and computers. This process is called brain-computer interaction (BCI) and is a vast field of research in neuroscience. Hence, in the past few years, emotion recognition has gained adequate attention in the research community. In this thesis, an emotion recognition system is designed and analyzed using EEG signals. Several existing feature extraction techniques are studied, analyzed, and implemented to extract features from the EEG signals. An SVM classifier is used to classify the features into various emotional states. Four emotional states are detected, namely, happy, sad, anger, and relaxed state. The model is tested, and simulation results are presented with an interpretation. Furthermore, this study has mentioned and discussed the efficacy of the results achieved. The findings from this study could be beneficial in developing emotion-sensitive technologies, such as augmented modes of communication for severely disabled individuals who are unable to communicate their feelings directly.

This thesis begins with providing a basic introduction of electricity requirements for small developing country communities serviced by small scale generating units (focussing mainly on small wind turbine, small Photo Voltaic system and Micro-Hydro Power Plants). Scenarios of these small scale units around the world are presented. Companies manufacturing different size wind turbines are surveyed in order to propose a design that suits the most abundantly available and affordable turbines. Different Maximum Power Point Tracking (MPPT) algorithms normally employed for these small scale generating units are listed along with their working principles. Most of these algorithms for MPPT do not require any mechanical sensors in order to sense the control parameters like wind speed and rotor speed (for small wind turbines), temperature and irradiation (for PV systems), and water flow and water head (for Micro-Hydro). Models for all three of these systems were developed in order to generate Maximum Power Point (MPP) curves. Similarly, a model for Permanent Magnet Synchronous Generators (PMSGs) has been developed in the d-q reference frame. A boost rectifier which enables active Power Factor Correction (PFC) and has a DC regulated output voltage is proposed before implementing a MPPT algorithm. The proposed boost rectifier works on the principle of Direct Power Control Space Vector Modulation (DPC-SVM) which is based on instantaneous active and reactive power control loops. In this technique, the switching states are determined according to the errors between commanded and estimated values of active and reactive powers. The PMSG and Wind Turbine behaviour are simulated at various wind speeds. Similarly, simulation of the proposed PFC boost rectifier is performed in matlab/simulink. The output of these models are observed for the variable wind speeds which identifies PFC and boosted constant DC output voltage is obtained. A buck converter that employs the MPPT algorithm is proposed and modeled. The model of a complete system that consists of a variable speed small wind turbine, PMSG, DPC-SVM boost rectifier, and buck converter implementing MPPT algorithm is developed. The proposed MPPT algorithm is based upon the principle of adjusting the duty ratio of the buck converter in order reach the MPP for different wind speeds (for small wind turbines) and different water flow rates (Micro-Hydro). Finally, a prototype DPC-SVM boost rectifier and buck converter was designed and built for a turbine with an output power ranging from 50 W-1 kW. Inductors for the boost rectifier and buck DC-DC converter were designed and built for these output power ranges. A microcontroller was programmed in order to generate three switching signals for the PFC boost rectifier and one switching signal for the MPPT buck converter. Three phase voltages and currents were sensed to determine active and reactive power. The voltage vectors were divided into 12 sectors and a switching algorithm based on the DPC-SVM boost rectifier model was implemented in order to minimize the errors between commanded and estimated values of active and reactive power. The system was designed for charging 48 V battery bank. The generator three phase voltage is boosted to a constant 80 V DC. Simulation results of the DPC-SVM based rectifier shows that the output power could be varied by varying the DC load maintaining UPF and constant boosted DC voltage. A buck DC-DC converter is proposed after the boost rectifier stage in order to charge the 48 V battery bank. Duty ratio of the buck converter is varied for varying the output power in order to reach the MPP. The controller prototype was designed and developed. A laboratory setup connecting 4 kW induction motor (behaving as a wind turbine) with 1kW PMSG was built. Speed-torque characteristic of the induction motor is initially determined. The torque out of the motor varies with the motor speed at various motor supply voltages. At a particular supply voltage, the motor torque reaches peak power at a certain turbine speed. Hence, the control algorithm is tested to reach this power point. Although the prototype of the entire system was built, complete results were not obtained due to various time constraints. Results from the boost rectifier showed that the appropriate switching were performed according to the digitized signals of the active and reactive power errors for different voltage sectors. Simulation results showed that for various wind speed, a constant DC voltage of 80 V DC is achieved along with UPF. MPPT control algorithm was tested for induction motor and PMSG combination. Results showed that the MPPT could be achieved by varying the buck converter duty ratio with UPF achieved at various wind speeds.

Electronic mail has become cast and embedded in our everyday lives. Billions of legitimate emails are sent on a daily basis. The widely established underlying infrastructure, its widespread availability as well as its ease of use have all acted as catalysts to such pervasive proliferation. Unfortunately, the same can be alleged about unsolicited bulk email, or rather spam. Various methods, as well as enabling architectures are available to try to mitigate spam permeation. In this respect, this dissertation compliments existing survey work in this area by contributing an extensive literature review of traditional and emerging spam filtering approaches. Techniques, approaches and architectures employed for spam filtering are appraised, critically assessing respective strengths and weaknesses. Velocity, volume and variety are key characteristics of the spam challenge. MapReduce (M/R) has become increasingly popular as an Internet scale, data intensive processing platform. In the context of machine learning based spam filter training, support vector machine (SVM) based techniques have been proven effective. SVM training is however a computationally intensive process. In this dissertation, a M/R based distributed SVM algorithm for scalable spam filter training, designated MRSMO, is presented. By distributing and processing subsets of the training data across multiple participating computing nodes, the distributed SVM reduces spam filter training time significantly. To mitigate the accuracy degradation introduced by the adopted approach, a Resource Description Framework (RDF) based feedback loop is evaluated. Experimental results demonstrate that this improves the accuracy levels of the distributed SVM beyond the original sequential counterpart. Effectively exploiting large scale, ‘Cloud’ based, heterogeneous processing capabilities for M/R in what can be considered a non-deterministic environment requires the consideration of a number of perspectives. In this work, gSched, a Hadoop M/R based, heterogeneous aware task to node matching and allocation scheme is designed. Using MRSMO as a baseline, experimental evaluation indicates that gSched improves on the performance of the out-of-the box Hadoop counterpart in a typical Cloud based infrastructure. The focal contribution to knowledge is a scalable, heterogeneous infrastructure and machine learning based spam filtering scheme, able to capitalize on collaborative accuracy improvements through RDF based, end user feedback. MapReduce based RDF Assisted Distributed SVM for High Throughput Spam Filtering

Student Number : 0213068F - MSc Dissertation - School of Computer Science - Faculty of Science / Drug resistance of the HI virus due to its fast replication and error-prone mutation is a key factor in the failure to combat the HIV epidemic. For this reason, performing pre-therapy drug resistance testing and administering appropriate drugs or combination of drugs accordingly is very useful. There are two approaches to HIV drug resistance testing: phenotypic (clinical) and genotypic (based on the particular virus’s DNA). Genotyping tests HIV drug resistance by detecting specific mutations known to confer drug resistance. It is cheaper and can be computerised. However, it requires being able to know or learn what mutations confer drug resistance. Previous research using pattern recognition techniques has been promising, but the performance needs to be improved. It is also important for techniques that can quickly learn new rules when faced with new mutations or drugs. A relatively recent addition to these techniques is the Support Vector Machines (SVMs). SVMs have proved very successful in many benchmark applications such as face recognition, text recognition, and have also performed well in many computational biology problems where the number of features targeted is large compared to the number of available samples. This paper explores the use of SVMs in predicting the drug resistance of an HIV strain extracted from a patient based on the genetic sequence of those parts of the viral DNA encoding for the two enzymes, Reverse Transcriptase or Protease, which are critical for the replication of the HIV virus. In particular, it is the aim of this reseach to design the model without incorporating the biological knowledge at hand to enable the resulting classifier accommodate new drugs and mutations. To evaluate the performance of SVMs we used cross validation technique to measure the unbiased estimate on 2045 data points. The accuracy of classification and the area under the receiver operating characteristics curve (AUC) was used as a performance measure. Furthermore, to compare the performance of our SVMs model we also developed other prediction models based on popular classification algorithms, namely neural networks, decision trees and logistic regressions. The results show that SVMs are a highly successful classifier and out-perform other techniques with performance ranging between (94.13%–96.33%) accuracy and (81.26% - 97.49%) AUC. Decision trees were rated second and logistic regression performed the worst.

Traffic Control Signs or destination boards on roadways offer significant information for drivers. Regulation signs tell something like your speed, turns, etc; Warning signs warn drivers of conditions ahead to help them avoid accidents; Destination signs show distances and directions to various locations; Service signs display location of hospitals, gas and rest areas etc. Because the signs are so important and there is always a certain distance from them to drivers, to let the drivers get information clearly and easily even in bad weather or other situations. The idea is to develop software which can collect useful information from a special camera which is mounted in the front of a moving car to extract the important information and finally show it to the drivers. For example, when a frame contains on a destination drive sign board it will be text something like "Linkoping 50",so the software should extract every character of "Linkoping 50", compare them with the already known character data in the database. if there is extracted character match "k" in the database then output the destination name and show to the driver. In this project C++ will be used to write the code for this software.

The biological function of a protein is mainly maintained by its three-dimensional structure. Protein folds support the three-dimensional structure of a protein, and then the inter-residue contacts in the protein impact the formation of protein folds and the stability of its protein structure. Therefore, the protein contact plays a critical role in building protein structures and analyzing biological functions. In this thesis, we propose a methodology to predict the residue-residue contacts of a target protein and develop a new measurement to evaluate the accuracy of prediction. With three prediction tools, the support vector machine (SVM), the k-nearest neighbor algorithm (KNN), and the penalized discriminant analysis (PDA), we compare these classifiers based on the self-testing of the training set, which are derived from representative protein chains from PDB (PDB-REPRDB), and apply the best (SVM) to predict a testing set of 173 protein chains derived from previous study. The experimental results show that the accuracy of our prediction achieves 24.84%,15.68%, and 8.23% for three categories of different contacts, which greatly improves the result of random exploration (5.31%, 3.33%, and 1.12%, respectively).