Franz I. S. Ko

Empirical Evaluation Of Classifiers For Software Risk Management

Software development involves plenty of risks, and errors exist in software modules represent a major kind of risk. Software defect prediction techniques and tools that identify software errors play a crucial role in software risk management. Among software defect prediction techniques, classification is a commonly used approach. Various types of classifiers have been applied to software defect prediction in recent years. How to select an adequate classifier (or set of classifiers) to identify error prone software modules is an important task for software development organizations. There are many different measures for classifiers and each measure is intended for assessing different aspect of a classifier. This paper developed a performance metric that combines various measures to evaluate the quality of classifiers for software defect prediction. The performance metric is analyzed experimentally using 13 classifiers on 11 public domain software defect datasets. The results of the experiment indicate that support vector machines (SVM), C4.5 algorithm, and K-nearest-neighbor algorithm ranked the top three classifiers.

THE FUZZY EVALUATION OF THE ICT PROJECTS IN STRATEGIC ENVIRONMENT (CASE STUDY: IRAN TELECOMMUNICATION RESEARCH CENTER)

Many criteria should be considered for project selection in Information and Communication Technology (ICT) sectors; while, the resources are limited. Besides, the uncertainty of the selection and evaluation process increases the difficulty of ICT project selection. Therefore, a fuzzy system in combination with fuzzy integer linear programming, fuzzy entropy, and fuzzy numbers have been developed; it smoothes the process of ICT project selection. This fuzzy system is flexible in the project selection for Iran ICT sectors, especially in (Research and Development) R&D sectors. It enhances the accuracy of the evaluation process as it resolves the uncertainty of ICT project selection.

Hybrid Genetic Algorithm for Solving Traveling Salesman Problem with Sorted Population

The performance of genetic algorithms (GA) is affected by various factors such as parameters, genetic operators and strategies. The traditional approach with random initial population is efficient however the whole initial population may contain infeasible solutions, which causes GA take longer to converge. The GA was modified in various ways to achieve faster and better convergence and it was particularly recognized by the researchers that initial population greatly affects the performance of GA. This study proposes modified GA with sorted initial population and applies it to solving travelling salesman problem (TSP). Normally the bigger the initial the population the more computationally expensive the calculation becomes with each generation. New approach will allow reducing the size of the initial problem while preserving a better fit population and thus leading to faster and better convergence. GA is based on theory of natural selection and new approach sorts out initial population so that only those with higher fitness are retained. It gives grounds to assume that a better solution can be obtained and in a shorter time. Moreover, new approach proposes that genetic algorithm has less probability of being stuck in the local optimum and rather produces a result close to global optimum. The proposed approach is tested on a simulator built using object-oriented approach and the test results prove the validity of the proposed method.

Novel OTP System Design Using Homomorphic Graph of Fingerprints

Abstract Security of internet is becoming one of the latest important concerns along with extensive application of the Internet. OTP is the first security medium for strengthening the stability of electronic financial transactions. In this paper, we propose the method of one-time password key generation of OTP using changed location and angle of fingerprint features and new OTP system. Fingerprint is one of the powerful personal authentication factors, and it can create variable password key for one-time use of information of fingerprint features. And we performed a simulation for proposed random password key generation method and efficiency analysis.

Reinforcement learning coordination with combined heuristics in multi-agent environment for university timetabling

University timetabling is a constraint-satisfaction problem where the participants usually have conflicting requirements. Multi-agent environment can solve timetabling problem through representing each participating party with agent. We propose to apply reinforcement learning approach to coordinate agents in a university timetabling system. The system simulation is built on JADE Framework.

Secure and Fast Multimedia Digital Content Distribution System Based on a Partial Encryption and Group-Key

Security and fast multimedia distribution has become an important research topic due to ever growing opportunities on the Internet and improving network capability. Faster and more secure multimedia distribution can be achieved through Web caching and encryption. Due to specific characteristic of multimedia content such as its size, it is important to reconsider encryption technique used for multimedia content. In order to reduce user perceived latency, it is proposed to use partial encryption based on the multimedia type. Moreover, in order to guarantee that only authorized users have access to secure multimedia, group key management is introduced into the proposed system. This enables to ensure backward and forward secrecy on authorized group. Using these techniques will guarantee a more secure and faster multimedia content distribution.