About me

Hi! My name is Gean Trindade Pereira and I am a Ph.D. candidate in Computer Science at the Institute of Mathematics and Computer Science (ICMC) in the University of São Paulo (USP). Previously, I did my Master’s degree in Computer Science at the Federal University of São Carlos (UFSCar), and my Bachelor’s in Software Engineering at the Federal University of Pampa (UNIPAMPA). I have experience on the topics of Machine Learning, Deep Learning, Meta-Learning, Computer Vision, Evolutionary Computation, Multi-Objective Optimization, and Bioinformatics, with a special interest in Neural Architecture Search.

My Ph.D. research explores Meta-Learning to optimize the Neural Architecture Search (NAS) framework. Thus, my focus is to propose and investigate Meta-Learning approaches to improve NAS predictive performance, data efficiency, and to accelerate the search for novel architectures, especially for image recognition tasks. Through Meta-Learning, I am exploring the possibility of using prior knowledge to learn from fewer examples, which can significantly benefit the optimization of NAS. Besides, my research is concerned with finding neural architectures at a low cost, so my experiments involve using simple and efficient setups.

From my research findings, I was able to publish scientific papers in leading conferences on Evolutionary Computation, as the Genetic and Evolutionary Computation Conference (GECCO), and the IEEE Congress on Evolutionary Computation (CEC). Besides, I am constantly active in the Machine Learning community, having works accepted and contributing to workshops at the main conferences in the area, such as NeurIPS, CVPR, and ICCV.