Gabriel Leivas Oliveira
Brief Bio:
I am a Machine Learning Research Team Lead at Borealis AI Vancouver - Borealis AI
From September 2014 to February of 2019 was a Ph.D. student from University of Freiburg supervised by Professors Thomas Brox and Wolfram Burgard. Mainly research deep learning approaches to robotics perception.
I hold two masters degrees, both in Computer Vision and Robotics, from the University of Minnesota and the Federal University of Minas Gerais (Brazil), respectively.
About Borealis AI Vancouver Center
Watch this:
Latest News:
Paper - Wonho Bae, Yi Ren, Mohamed Osama Ahmed, Frederick Tung, Danica J. Sutherland, Gabriel L. Oliveira, AdaFlood: Adaptive Flood Regularization, TMLR 2024, (paper, code)
Paper - Vani, A; Tung, F; Oliveira G; Sharifi H. ”Forget Sharpness: Perturbed Forgetting of Model Biases Within SAM Dynamics”, International Conference on Machine Learning (ICML) 2024. (paper, code)
Paper - Rahimian, E.; Javadi, G.; Rahimian, E; Tung, F.; Oliveira, G. "DynaShare: Task and Instance Conditioned Parameter Sharing for Multi-task Learning", Efficient Deep Learning for Computer Vision Workshop at CVPR, 2023. (paper, code).
Paper - Wonho Bae, Mohamed Osama Ahmed, Frederick Tung, Gabriel L. Oliveira "Meta Temporal Point Processes" International Conference on Learning Representations (ICLR), 2023 (paper, code).
Interests:
Foundation Models
Robust Machine Learning
Deep Learning for Robotics Perception
Multi-task learning
Semantic Segmentation
Model compression for deep learning
Computer Vision
Click here for my publications
Education
Doctor of Philosophy in Computer Science, 2014 - 2019, University of Freiburg (Freiburg), Freiburg, Germany
Master of Science in Computer Science, 2014, University of Minnesota (UMN), Minnesota, MN
Master of Science in Computer Science, 2012, Federal University of Minas Gerais (UFMG), Belo Horizonte, MG, Brazil
Bachelor of Science in Computer Engineering, 2009, Federal University of Rio Grande (FURG), Rio Grande, RS, Brazil
Invited Talks
"Robust Machine Learning" was presented at Federal University of Rio Grande, Brazil, Rio Grande, September 2024.
"Robust Machine Learning" was presented at University of Freiburg, Germany, Freiburg, July 2024.
"DynaShare: Task and Instance Conditioned Parameter Sharing for Multi-task Learning" was presented at Simon Fraser University, Canada, Vancouver, June 2023.
"Heterogeneous Multi-task Learning with Expert Diversity" was presented at University of Freiburg, Germany, Freiburg, Dec 2021.
"Encoder-decoder methods for semantic segmentation: efficiency and robustness aspects" was presented at Federal University of Rio Grande, Brazil, Rio Grande, Jan 2017.
Contact:
Gabriel Leivas Oliveira
gabrielleivas [at] gmail [dot] com
gabriel [dot] oliveira [at] rbc [dot] com