Welcome to my website! I'm a Computer Science Ph.D. student at UdeM (Université de Montréal) under the supervision of Prof. Margarida Carvalho and Prof. Kim Yu. My research interests span from combinatorial optimization to deep learning. I am currently focusing on Bilevel IPs and applications to game theory (Nash and Stackelberg games, Integer Programming Games). I got both my B.Sc. and my M.Sc. at the Dept. of Systems and Automation Eng. of UFSC (Federal University of Santa Catarina), Brazil. Both my thesis were done under the supervision of Prof. Eduardo Camponogara, with whom I worked on MIP with piecewise functions, learning-based heuristics for MILP, oil production optimization, nanosatellite task scheduling, and physics-informed learning with implicit models. I have also worked with deep learning for medical imaging and uncertainty estimation for segmentation models under the supervision of Prof. Danilo Silva, and hyperparameter optimization (AutoML) at Fraunhofer IPT. The best way to contact me is by email: mpacheco.bruno@gmail.com

News

[2024/04/07] From May 12th to 15th I will be at Optimization Days (JOPT2025) in Montreal, Canada, to present our recent paper on MINLP problems with multilinear interpolations.

[2024/11/22] I was awarded the DIRO excellence scholarship (bourse d'excellence DIRO)!

[2024/08/13] My master's thesis titled "Deep-learning-based Primal Heuristics for MILP: Supervised Solution-prediction Models" has been presented and approved. Thank you Prof. Danilo Silva and Prof. Teobaldo Bulhões for composing the examining board.

[2024/04/18] I have just received the offer from UdeM: I am strating my Ph.D. this fall at DIRO under the supervision of Prof. Margarida Carvalho and Prof. Kim Yu!

[2023/10/02] From October 15th to 18th I will attend SBAI 2023 and present our paper on deep learning for oil production optimizaton.

[2023/09/26] I just presented our paper on pre-training for deep-learning-based brain age at BRACIS 2023.

[2023/05/01] I have started as a B.Sc. research intern in a Petrobras project on offshore oil production optimization.

Selected Publications

For an up-to-date list including all my publications, check out my Google Scholar page.

• A relax-fix-and-exclude algorithm for an MINLP problem with multilinear interpolations
  Bruno M. Pacheco, Pedro M. Antunes, Eduardo Camponogara, Laio O. Seman, Vinícius R. Rosa, Bruno F. Vieira, Cesar Longhi
  Under review. Available here.

• Graph Neural Networks for the Offline Nanosatellite Task Scheduling Problem
  Bruno M. Pacheco, Laio O. Seman, Cezar A. Rigo, Eduardo Camponogara, Eduardo A. Bezerra, Leandro dos S. Coelho
  Under review. Available here.

• Selective Prediction for Semantic Segmentation using Post-Hoc Confidence Estimation and Its Performance under Distribution Shift
  Bruno L. C. Borges, Bruno M. Pacheco, and Danilo Silva
  PML4LRS Workshop, ICLR 2024. Available here.
  2024

• Deep-learning-based Early Fixing for Gas-lifted Oil Production Optimization: Supervised and Weakly-supervised Approaches
  Bruno M. Pacheco, Laio O. Seman, and Eduardo Camponogara
  SBAI 2023. Available here.
  2023

• Does pre-training on brain-related tasks results in better deep-learning-based brain age biomarkers?
  Bruno M. Pacheco, Victor H. R. de Oliveira, Augusto B. F. Antunes, Saulo D. S. Pedro, Danilo Silva
  BRACIS 2023. Available here.
  2023

• Automated machine learning for predictive quality in production
  Jonathan Krauß, Bruno M. Pacheco, Hanno M. Zang, and Robert H. Schmitt
  PROCEDIA CIRP. Available here.
  2023

Selected Code Repositories

These are some code repositories of which I am either the only author or a main contributor. Most of my publicly available code can be found in my GitHub.

• Deep Learning Boilerplate
  A boilerplate for deep learning projects with PyTorch, including training and evaluation scripts, data loaders, and experiment tracking (using W&B). I use it for most of my deep learning research projects.

• SatGNN
  Implementation of the experiments reported in the paper "Graph Neural Networks for the Offline Nanosatellite Task Scheduling Problem", including GNN training and heuristic building. It is also where the results of my M.Sc. thesis come from.

• Pre-training brain age deep learning models
  Implementation of the experiments reported in the paper "Does pre-training on brain-related tasks results in better deep-learning-based brain age biomarkers?".

• Physics-Informed Deep Equilibrium models
  The code of my B.Sc. thesis. PyTorch implementation of deep equilibrium models with support for physics-informed learning, i.e., second-order regularization terms. It also contains the experiments performed on the Van der Pol oscillator.

• Part counting
  A project I worked on for a course of my B.Sc. in Control and Automation Engineering. It is basically a computer vision tool (with some deep learning) to estimate the amount of parts in an RGB-Depth image.

• ImunoForce
  A simple shoot 'em up game I wrote with a colleague during high school as a class project. The game is written in C using the Allegro library.