Kelton Costa, Ph.D.
      




ACADEMIC LIFE


Graduated in Systems Analysis from the Sagrado Coração University-USC, Master's Degree in Computer Science from the Euripides de Marilia University-UNIVEM, Ph.D. from the São Paulo University-USP, Post-doctoral in Computer Networks by Campinas State University-UNICAMP and post-doctoral in Anomaly Detection in Computer Networks by São Paulo State University-UNESP. Currently is Professor of Technology College-FATEC, Professor of São Paulo State University-UNESP, Advisor Professor of the Program Master's Degree in Computer Science-UNESP. It evaluator undergraduate courses INEP-MEC and has experience in Computer Science with emphasis in Computer Systems Architecture and Distributed Systems, acting on the following topics: Management in Computer Networks, Security in Computer Networks, Anomaly Detection Systems and Signatures in Computer Networks and Data Flow, Analysis in Computer Networks.



Latest Publications (Book Chapters/Journals/Conferences >> see more about Kelton's Curriculum Lattes

Master Degree Program

Computer Science




Research Projects CNPq

2019 - current
About the Image Security Using Intelligent Techniques - CNPq Universal
CNPq PROC 429003/2018-8



Research Projects Fapesp

 
2018 - current
About Image Security Using Machine Learning

Description: FAPESP PROC 2017/22905-6: Seam carving and Watermarking techniques to the specific methods for the use in the image processing area, where they have as characteristics to obtain security in the information, the concealment of the information within the images, as well as the reduction of their size, preserving their features. Nowadays, such techniques are characterized as one of the leading problems in the area of image security, and may even damage other areas such as the detection of modified images, an area specifically studied and applied by Computational Forensic Perception. The present project proposal aims to act in two aspects, the first one in the detection of altered images using the Forest Path Pathfinder (OPF) and Convolutional Neural Networks (CNN) classifier, and the second one in the detection of Watermarking in images using the OPF classifier.
Status: Ongoing; Nature: Search.
Students involved: Graduation: (0)/Specialization:(0)/academic Master Degree: (1)/Professional Master Degree: (0)/PhD (1).
Members: Kelton Augusto Pontara da Costa
Financer: FAPESP Foundation for Research of the São Paulo State

2017 - 2018
Detection Images Resizing via Seam Carving Using Convolutional Neural Networks in Computer Forensic Context


Description: FAPESP PROC  2016/25687-7: Seam carving is currently one of the most used methods to resize images, besides being a simple algorithm that maintains the content of the image, thus providing popularity. However, one of the techniques that can be used through this procedure is the removal of objects and people from original images, which can cause problems with Computational Forensics. The present proposal of scientific initiation aims at the creation of a technique to detect this process even without having knowledge of the original image, through image characteristics and texture properties of the image acquired by the Local Binary Pattern applied to a Convolutional Neural Network, the Which will determine whether the image is original or has undergone a process of image manipulation. The method will aim to detect original images that have undergone modifications and will serve as an aid to researchers working in the area of ​​forensic and cognitive computing, as well as to compare their effectiveness in relation to existing techniques.
Status: Ongoing; Nature: Search.
Students involved: Graduation: (0)/Specialization:(0)/academic Master Degree: (0)/Professional Master Degree: (0)/PhD (2).
Members: Kelton Augusto Pontara da Costa - Advisor / Luiz Fernando da Silva Cieslak - Researcher / João Paulo Papa - Collaborator Researcher.

Financer: FAPESP Foundation for Research of the São Paulo State

2015 - 2017
About Anomaly Detection in Computer Networks Using Optimum-Path Forest: Advances and Applications in Computer Networks

Description: FAPESP PROC 2015/00801-9: Modality Research. Note, in recent years, the number of people interested in accessing unauthorized information is growing exponentially in conjunction with the different number of attacks that come at the same speed. It is considered also that certain tools, such as antivirus and firewalls are important to security in a computer network, however due to the large number of anomalies present in such environments, it becomes difficult an application has great efficiency and effectiveness in detection of these malicious codes. Thus, considering the diversification of the attacks and their complexities, companies has increased their investment in research for the development of more efficient intrusion detection systems using artificial intelligence techniques. This research project aims the study and development of anomaly detection techniques in computer networks using the Optimum-Path Forest Classifier, which has not yet been employed in this context. In addition, the project aims to contribute to a library to the OPF Classifier in order to expand its dissemination among national and international researchers.
Status: Ongoing; Nature: Search.
Students involved: Graduation: (2)/Specialization:(0)/academic Master Degree: (3)/Professional Master Degree: (0)/PhD (1).
Members: Kelton Augusto Pontara da Costa - Coordinator / João Paulo Papa - Collaborator Researcher.
Financer: FAPESP Foundation for Research of the São Paulo State
.

2016 - 2017
Features Selection Using Meta-Heuristic Approach to Anomaly Detection Computer Network

Description: FAPESP PROC 2016/03088-4: Currently a number of people interested in obtaining unauthorized access to the information contained in computer networks is growing exponentially along with all kinds of intrusions, attacks and malicious code. One of the areas that has been most studied in the context of anomaly detection is the invasion of computer networks, mainly due to the variety of attacks and their complexities. Because of this, companies have increased their investment in research for the development of more efficient intrusion detection systems, using such artificial intelligence techniques. The proposed objective scientific research to create a robust database of anomalies in computer networks and the application of a technique for selection of meta-heuristic-based features in order to maximize the hit rate of the classifiers to be applied for this database. The new database is intended to serve as an aid to researchers from security area computer networks that require large databases to improve the results of their experiments.
Status: Ongoing; Nature: Search.
Students involved: Graduation: (0)/Specialization:(0)/academic Master Degree: (0)/Professional Master Degree: (0)/PhD (1).
Members: Kelton Augusto Pontara da Costa - Advisor / Bruna de Camargo Rubio - Researcher.
Financer: FAPESP Foundation for Research of the São Paulo State
 
 

















 

Copyright © 2016 - All Rights Reserved