Thiago F. Costa, Mateus Tymburibá
BRACIS 2022 - Nov. 2022 - Campinas, Brazil
Award: Honorable Mention, Best Papers Awards
Estimates how likely a software vulnerability is to be exploited, using interconnected neural networks whose settings are tuned by a genetic algorithm. Presented at BRACIS 2022 in Campinas, where it won an Honorable Mention in the Best Papers Awards.
DOIGoogle Scholar
Thiago F. Costa, Mateus Tymburibá
SINCONF 2022 - 15th IEEE International Conference on Security of Information and Networks - Nov. 2022 - Hammam Sousse, Tunisia
An open-source tool that ranks software security patches by the estimated probability that each vulnerability is exploited, not only by its severity. It uses interconnected neural networks optimised by genetic algorithms, and on 20 yearly evaluations from 2000 to 2020 it reached a higher F1-score than every CVSS-based strategy. Published at SINCONF 2022, the 15th IEEE International Conference on Security of Information and Networks.
DOIGoogle Scholar
Thiago F. Costa, André R. da Cruz
OL2A: International Conference on Optimization, Learning Algorithms and Applications 2022 - Oct. 2022 - Bragança, Portugal
Stock-Pred is an LSTM neural network that predicts stock prices from historical data. Published at OL2A 2022, the International Conference on Optimization, Learning Algorithms and Applications.
DOIGoogle Scholar
Flávio R. S. Coutinho, Ismael S. Silva, Glívia A. R. Barbosa, Thiago F. Costa, Luis G. D. Carvalho, Cassiano B. Andrade, João P. S. Pereira, Reynaldo S. G. Pinto, Jonathan H. Silva
XXI SBGames - Computing Track - Oct. 2022 - Natal, Brazil
Move2Play is an open exergaming platform that needs only commodity hardware, a computer, a screen and one smartphone per player, to run single and multiplayer sessions, and it is open to games from third parties. Four demonstration exergames were built to show it working. Presented at SBGames 2022 (XXI SBGames, Computing Track) in Natal.
DOIGoogle Scholar
Thiago F. Costa, Elizabeth F. Wanner, Flávio V. C. Martins, André R. da Cruz
I BWAIF - XLII CSBC - Aug. 2022 - Niterói, Brazil
A two-step method to design an LSTM stock-price predictor: iRace finds robust hyperparameters, then NSGA-II refines the network by trading off price error against seven-day trend accuracy. Tested on US and Brazilian stocks, it reached a MAPE of 1.279% for BVSP, 1.564% for IBM and 2.047% for AAPL. Presented at the I BWAIF, held with the XLII CSBC in 2022.
ViewPDFDOIGoogle Scholar
Thiago F. Costa, André R. da Cruz
XXII Encoinfo - Nov. 2020 - Palmas, Brazil
A systematic literature review of offensive hacking strategies such as penetration testing, aimed at making networks safer by removing the vulnerabilities they find. It covers how these attacks are avoided and simulated, using only open-source tools. Presented at the XXII Encoinfo in 2020.
PDFGoogle Scholar
Lucas D. M. dos Santos, Thiago F. Costa, Gabriel S. Brandão, Ana Luisa Pimenta, André R. da Cruz
XIX SBGames - Education Track - Nov. 2020 - Recife, Brazil
Award: Best Paper, 3rd place, Education Track
The design and creation of an educational mobile RPG-adventure for sixth graders, covering the Brazilian national curriculum (BNCC) and mixing competition and cooperation between schools. The project ranked among the best in a 2018 national educational-app marathon, and its first full version spent a year and a half in an app store. It took 3rd place for Best Paper in the Education Track of SBGames 2020.
ViewPDFGoogle Scholar
Thiago F. Costa, Tiago A. Mendonça, Bernard M. da Costa, Flávio R. S. Coutinho, André R. da Cruz
XVIII SBGames - Artes & Design - Design Cases - Oct. 2019 - Rio de Janeiro, Brazil
The design of AVENTURA DISCRETA, an open-source platform game that follows a computer student through five stages, from technician degree to a career as a teacher. It reports the game design and the process, built with Scrum and Trello, in C++ with OpenGL, freeGLUT and OpenAL. Presented at SBGames 2019 (XVIII SBGames, Arts & Design, Design Cases).
ViewPDFGoogle Scholar
Thiago F. Costa, Ismael S. Silva, Glívia A. R. Barbosa, Flávio R. S. Coutinho
XVII SBGames - Computing Track - Nov. 2018 - Foz do Iguaçu, Brazil
A cross-platform architecture that turns a smartphone into a motion-detecting controller for computer games, so exergames no longer need expensive proprietary hardware. A prototype showed it effective, efficient and scalable in the messages exchanged between its components. Presented at SBGames 2018 (XVII SBGames, Computing Track, short papers).
ViewPDFGoogle Scholar