The event will take place from May 5 to 7 at the FGV EMAp headquarters in Rio de Janeiro and will bring together applied mathematicians, data scientists and industry professionals involved in the area of ​​Optimization

Optimization is of fundamental importance in Applied Mathematics, as it deals with the search for the best possible solutions to real-world problems. Today, this branch of Mathematical Sciences is found in practically all sectors of society. Aiming to bring academic studies and industry closer together in theory and practice, the School of Applied Mathematics of the Getulio Vargas Foundation (FGV EMAp) is organizing the 1st edition of the Rio de Janeiro Workshop on Optimization and Applications (CariOPT).

CariOPT 2025 will discuss current trends in Operations Research Optimization from mathematicians, data scientists and industry professionals | Photo: ENVATO

The event will be held at the FGV EMAp headquarters in Rio de Janeiro from May 5th to May 7th, and is sponsored by the Coordination for the Improvement of Higher Education Personnel Foundation (CAPES) and the Carlos Chagas Filho Foundation for Research Support of the State of Rio de Janeiro (FAPERJ). The goal of the workshop is to bring together applied mathematicians, data scientists and professionals from regional, national and international industries who involve the area of ​​Optimization in their research.

In addition to fundamental research, CariOPT 2025 will prioritize application areas with strong intersection with Optimization. According to Philip Thompson, coordinator of the FGV EMAp initiative, the idea for the event arose in 2024 as a way to integrate the traditional Brazilian Workshop on Continuous Optimization (BrazOpt), which takes place every two years.

“The development of the Optimization area directly impacts areas that depend on algorithms for decision-making, and the objective becomes even more challenging when decisions must be made under uncertainty, such as training Machine Learning models or solving industrial problems in logistics and electricity distribution”, explains the FGV EMAp researcher.

In view of this, the conference will discuss emerging trends in Optimization in theoretical, practical and computational aspects, such as:

• Privacy in machine learning;
• Planning machine learning rates;
• Economic equilibrium problems;
• Cryptocurrency markets;
• Nonlinear programming;
• Algorithms for factorization of matrices and tensors;
• Optimization in manifolds and variational analysis.

For the 1st CariOPT, FGV EMAp invited references, researchers and enthusiasts in Machine Learning and Data Science, in addition to reserving space for industry professionals. Among the highlights of the plenary sessions are Aaron Defazio, researcher at Meta AI, owner of platforms such as WhatsApp, Instagram and Facebook, and Cristóbal Guzmán, scientist at the Faculty of Mathematics of the Pontifical Catholic University of Chile (PUC-Chile) and member of Google Research of the North American online services and software company.

Philip Thompson, from FGV EMAp, is the coordinator of CariOPT 2025 | Photo: FGV EMAp

For several years, Defazio has been working on adaptive stochastic gradient algorithms, the main type used to train neural networks. “Training a large neural network is an art in itself, and calibrating the optimal learning rate is usually a trial and error task. Adaptive algorithms such as ‘AdaGrad’ or ‘Adam’ are the most widely used in the industry because they reduce the calibration effort, one of the most important areas in Machine Learning Optimization”, says Thompson about the topic of the Meta researcher’s talk.

In turn, Guzmán also works with the same arrangement of algorithms, but with special attention to data privacy, a modern area of ​​Machine Learning that seeks to protect data from being exposed to unwanted third parties. “Privacy is also closely related to the area of ​​Robust Machine Learning, which seeks to make adequate predictions even when the data may be corrupted”, adds the FGV EMAp researcher.

In addition, the School will take advantage of the 1st CariOPT to celebrate the 75th birthday of Alfredo Iusem, Full Professor at FGV EMAp and specialist in Continuous Optimization, and thank him for his strong contribution to the segment at national and international levels.

Iusem is also part of the event's Scientific Committee, alongside Thompson, Roberto Imbuzeiro de Oliveira, a researcher at the Institute of Pure and Applied Mathematics (IMPA), and João Pereira, a mathematician at the University of Georgia in the United States, who will be offering a mini-course on tensor decomposition. 

Three other professors from FGV EMAp - in addition to Iusem and Thompson - make up the Organizing Committee of the 1st CariOPT. They are: Vincent Guigues, Diego Mesquita and Bernardo Costa, all experts in Optimization, Data Science, Deep Learning and Robust Machine Learning, Machine Learning and Artificial Intelligence.

Continuous Optimization Specialist Alfredo Iusem will be honored by FGV EMAp at CariOPT 2025 for his 75th birthday | Photo: FGV EMAp

Approximately 50 participants are expected to participate in the three-day event, with interaction in high-quality lectures and poster sessions, in which FGV EMAp offers postgraduate students, post-doctorates and young researchers the opportunity to present their work.

Registration

Undergraduate students interested in participating in CariOPT 2025 can register for a fee of R$100.00. For graduate students, the fee is R$200.00. Registration for Brazilian researchers at other levels costs R$400.00 and for foreign researchers, R$1,200.00. Registration can be done directly on the event website.

CariOPT 2025

When: May 5th - May 7th, 2025

Where: FGV Cultural Center - Praia de Botafogo, 190 - Rio de Janeiro - RJ

Important links: Registration | Schedule | Official event website

Questions? Contact us by email at cariopt@fgv.br