Thesis

5. ISABELA N. F. DE QUEIROZ

Estimation of the saturation flux with the aid of neural networks

6. CARLOS ROBERTO PORFÍRIO JUNIOR

An approach of quantum computing in computational theory and computer simulation of these resources in conventional

Computational resources classics have faced problems of high complexity and sometimes not solvable in acceptable time intervals. Quantum computing, research area of computer science very recent, appears as one or probably the only viable alternative for solving classical algorithms that are theoretically feasible but practically not. This paper aims to summarize the basic concepts of quantum computing, framing it in computational theory, presenting the main theoretical study of algorithms already surveyed the area and mainly display a computer simulation of conventional quantum computational resources in order to provide one academic didactic means to initiate studies in this area, since the quantum phenomena are difficult to be accepted and interpreted by users freshmen in quantum computing, because these phenomena do not result in expected behaviors in its conventional form.

 

7. CANDIDO GUERRERO SALGADO

Behavior of Association Rules and Its Application for Analysis of Medical Data

Data Mining, also known as KDD (Knowledge Discovery in Databases), is the process of extracting information of interest, non-trivial, implicit, previously unknown and potentially useful, from large volumes of data. Mining Association Rules seeking interesting associations between items in a large volume of data. In this work are presented, in an introductory way, some interesting approaches to Data Mining. A comparative analysis of the various approaches is also performed and presented the results of this comparison. The following is the state of the art as regards' the Association Rules, making a presentation of its evolution to the present stage where it is. Finally, we present a case study of application of Association Rules to aid medical diagnosis.

 

8.WESLEY MARTINS TELES

Adaptive system for Web sites based on the behavior of the ant

The ant's behavior can be used as a metaphor to improve the performance of Web type systems With current technology, Internet browsing, the conditions under which the user browses pages looking for their target are very similar to the procedure of ants to find food. But, differently to the procedure of ants, web users do not have pheromone so that they can cooperate in the process of navigation. Based on these observations and the lack of a work that aims to guide the user's web emphasizing route optimization study was carried out in order to fill this gap by applying the ant's behavior we call AntWeb web. This study was divided into two parts: The evaluation of websites for AntWeb AntWeb and adaptive. Regarding AntWeb for evaluating websites was developed an evaluation methodology which was implemented in software that had as input the structure of a website and how to output a measure of performance of the structure. Regarding AntWeb adaptive, two models were developed heuristics to guide the user in a Web context adaptive hypermedia which we call model 1 and model 2. The first model was implemented and tested on the website of the Department of Computer Science, UNB. With the model 2 simulations were made which showed optimistically that the model can assist the user 2 Web The aim of this study is to report the research done with the AntWeb.

 

9. OSMAR QUIRINO DA SILVA. 30/09/2003

Modeling genetic algorithm for optimization, prediction and guidance of road traffic

This study evaluates the applicability of the  

This paper presents research on the problem of predicting the travel time public transport bus urban mass between consecutive stops, aiming to carry out a comparative study between two well-known methodologies, Artificial Neural Networks (ANN) and Model-Box Jenkins. For this, there were a few steps to reach the main focus of the work, among which may be mentioned broadly, the study of the state of the art, which represented a literature search to support the work of the research for field data collection and observation of the bus line chosen in Pilot Plan, in Brasília, Distrito Federal (the choice of a single line was to delimit the object of study). Then the data were modeled to be the need of the software, used later in the implementation phase, in the case of neural networks was used MATLAB, and the Box-Jenkins model, the SAS software. And finally, choose the best model to formalize a solution for the estimation of the arrival time of transport vehicles in urban public bus stops. The results were satisfactory and it has been found that the use of Artificial Neural Network is suitable to solve the prediction problem.

model as Genetic Algorithm technique Artificial Intelligence - AI applied to problems of prediction, optimization and guidance of road traffic. Implements a solution to the problem of choosing the best route traffic artery in the Federal District on a data set that reflects the situation on these routes in real time, based on the flow of Origin / Destination. As a consequence, leads to a dynamic adjustment of the flow of traffic, reaching an equilibrium system. The case study presents three simulations considering actual data such as distance, free stream velocity and obstacles. The experiments demonstrate the efficiency of the classification and generating the results of simulations considering the road network linking urban centers. Because it is a heuristic, the stopping criterion of the algorithm was arbitrated in 5 generations, selecting the best result among the five, in order to avoid that the algorithm presents a solution other than the minimization of the cost function. However, this solution applies to any city or similar problems. This study shows improvements in the solutions of shortest path and dynamic balancing traffic in real time, avoiding excessive points congestionamentos.Palavra Keywords: Optimization Traffic, Genetic Algorithm, Traffic Forecast. Traffic Guidance.

 

10. SORAIA SILVA PRIETCH, 13/02/04

Computational Methods for Forecast Travel Urban Bus Stops between Consecutive

  

 

11. MARCOS VINICIUS PINHEIRO DIB, 03/09/2004

Multi-Agent System using Computational Grid Synchronization and Management of Traffic Flow

This research proposes a Multi-Agent System using Grid Services for Synchronization and Management of Air Traffic Flow (ATFM). The ATFM service is designed to ensure optimum flow of air traffic to, or through, the areas within which the traffic demand at times exceeds the available capacity. The article presents the problem of ATFM and its synchronization property. To demonstrate the developed system are described the architecture of the grid, agents and negotiation mechanisms between them. As an example, it is reported a case study of tactical planning in 4 Brazilian airports. As a criterion for measuring effectiveness in reducing the number of negotiations between the agents and flight delays, is used in this analysis, the Standard Balancing Airport (PBA). The simulation shows the efficiency of the developed model and the success of its application in this case study.

 

12. MANQI WU, 15/04/2005

Cooperative navigation system for websites: approach and case study portal Interlegis

AntWeb system is a navigation aid for Websites and their approach is inspired by the behavior of ant colonies to find food. The system considers Web users as if they were artificial ants. This paper presents an extension to the AntWeb to take into account the complex behavior of human beings during the search process, with the use of techniques of mining Web usage patterns to capture visitors' navigation of the website, making it more appropriate and applicable to the reality of Web users the proposed extended model of AntWeb is to preprocess log files of the Web server, discover categories of users that exhibit similar information needs by providing pages and dynamically adapted with highlighted links that allow visitors to reach their common interests. A prototype of this model was implemented based on real data obtained from Interlegis Portal - Website Community Legislative Brazilian - with the use of free software: Zope (Z Object Publishing Environment), CMF (Content Management Framework) and the database PostgreSQL relational. To demonstrate the feasibility and potential of the proposed solution, three case studies were conducted: the first validates the applicability of Web usage mining in AntWeb, the second validates the extended model of AntWeb as a method of navigational assistance cooperative effective for Websites; and the third evaluation system performance with and without the use of AntWeb. In the case with the AntWeb, tests were made involving variation in parameters that directly or indirectly affect the calculation of the probability. The results show the sensitivity and accessibility of the system as well as benefits to users of the Website Interlegis.