AccomplishmentsDevelopment of Support Markov Model for the Prediction of Disulfide-bonding State of Cysteines o Grade: 30/30 L. o Duration: 4 months. o Place: University of Bologna. o Abstract: A hybrid predictor is developed to recognize the disulfide-bonding state of cysteines using evolutionary information in the form of multiple alignment profiles. Support Vector Machine (SVM) is adopted to predict the bonding state of cysteines in local context while Hidden Markov Model (HMM) is employed to predict the bonding state in global context. Some statistics were also computed to present more information about our dataset which comprises 5079 non-redundant protein resolved as three dimensional structures, containing in total 20027 cysteine residues. This integrated system attained a 89% and 85% prediction accuracy per cysteine and per protein, respectively, and is considered quite successful compared to previously published predictors. It’s done during the Laboratory of Bioinformatics II course. Full version is available upon request.
Support Vector Machines to Enhance the Performance of Pedotransfer Functions for Predicting the Water Retention Properties of Calcareous Soils. o Duration: not finished yet. o Place: Collaboration with Ghent University in Belgium o Abstract: Knowledge of soil hydraulic properties is indispensable for land management in arid and semiarid areas. The most important properties are the soil-water retention curve (SWRC) and hydraulic conductivity characteristics. Direct measurement of the SWRC is difficult, tedious to accomplish and expensive. Pedotransfer functions (PTFs) utilize data mining tools to predict SWRC. Nowadays, modern data mining techniques has become crucial in enabling high accuracy and good generalization to novel data. In this study we explore the use of Support Vector Machines (SVMs), a novel type of learning algorithm based on statistical theory, for predicting SWRC from more easily and cheaply measured properties. 72 undisturbed soil samples have been collected from different agro-climatic zones of Syria. The soil water contents at eight matric potentials were determined and selected as output variables. While two different input groups were used to compare their performances. A brief overview of the theoretical background of this fairly new technique and the use of specific kernel functions are presented. Then, the model parameters were optimized with cross-validation and grid-search method. The performance of the SVM-based PTFs was analyzed using the coefficient of determination, root mean square error (RMSE) and mean error (ME). This study shows that SVMs has the potential to be a useful and practical tool for predicting the SWRC. Full version is available upon completion.
ALEPESDD o Grade: 91% o Duration: 6 months. o Place: University of Aleppo
Source Code: http://www.m4ai.com/Prototype.rar ADT o Grade: 82% o Duration: 5 months. o Place: La Trobe University
Developing and producing an interactive engine for the design, manipulation and storage of aircraft designs, providing an intuitive function to facilitate the production of engineering reports, and being structured in such a way that users must log in to the program. AOCR o Grade: 80% o Duration: 10 months. o Place: University of Aleppo.
Scanning images that contain Arabic text, then extracting the Arabic text by detecting the tokens existed in the page. Finally, segmenting each token into letters and recognizing these letters using Neural Networks and Patterns Correlation techniques. E-GRID DRAWER o Duration: 2 months. o Place: Al-Aqsa Company.
Creating graphics to be displayed on an electronic grid that is composed of LEDs. VBULLETIN SMS BRIDGE o Duration: 3 months. o Place: Al-Aqsa Company.
A web-based application for sending short messages to clients’ or friends mobiles. Compatible with VBulletin Systems. |