PhD Student, |
I am a 4th year PhD student at TTI-Chicago, a philanthropically endowed computer science research institute located in the University of Chicago campus, and I am advised by Prof. Michael Maire and Prof. David McAllester. I received my Bachelor's from the Federal University of Rio de Janeiro in Brazil, and my Master's from Columbia University. Before joining TTIC, I had the pleasure to do research under the supervision of Prof. Daniel Ratton on topics such as graph embedding and neural networks.
I am generally interested in designing simple and principled methods to make machine learning applications more efficient in terms of data and compute. In particular, I am enthusiastic about unsupervised learning, generative modelling, sparsity in deep networks, and deep learning applications in computer vision.
Growing Efficient Deep Networks by Structured Continuous Sparsification
Xin Yuan, Pedro Savarese and Michael Maire
ICLR 2021 (Oral) PDF
Winning the Lottery with Continuous Sparsification
Pedro Savarese*, Hugo Silva* and Michael Maire
*: equal contribution
NeurIPS 2020 PDF | Code
Learning Implicitly Recurrent CNNs Through Parameter Sharing
Pedro Savarese and Michael Maire
ICLR 2019 PDF | Poster | Code
Residual Gates: A Simple Mechanism for Improved Network Optimization
Pedro Savarese and Daniel Figueiredo
Technical Report 2017 PDF
Learning Identity Mappings with Residual Gates
Pedro Savarese, Leonardo Mazza and Daniel Figueiredo
arXiv 2016 PDF
Domain-independent Dominance of Adaptive Methods
Pedro Savarese, David McAllester, Sudarshan Babu, Michael Maire
CVPR 2021 PDF | Code
Kernel and Rich Regimes in Overparametrized Models
Blake Woodworth, Suriya Gunasekar, Pedro Savarese, Edward Moroshko, Itay Golan, Jason Lee, Daniel Soudry and Nathan Srebro
COLT 2020 PDF
Convergence of Gradient Descent on Separable Data
Mor Shpigel Nacson, Jason Lee, Suriya Gunasekar, Pedro Savarese, Nathan Srebro and Daniel Soudry
AISTATS 2019 PDF
How do Infinite Width Bounded Norm Networks look in Function Space?
Pedro Savarese, Itay Evron, Daniel Soudry and Nathan Srebro
COLT 2019 PDF | Poster | Code
Information-Theoretic Segmentation by Inpainting Error Maximization
Pedro Savarese, Sunnie Kim, Michael Maire, Greg Shakhnarovich and David McAllester
CVPR 2021 PDF
From Monte Carlo to Las Vegas: Improving Restricted Boltzmann Machine Training Through Stopping Sets
Pedro Savarese, Mayank Kakodkar and Bruno Ribeiro
AAAI 2018 PDF | Poster | Code
struc2vec: Learning Node Representations from Structural Identity
Leonardo Ribeiro, Pedro Savarese and Daniel Figueiredo
KDD 2017 PDF | Talk | Code
Building a Massive Corpus for Named Entity Recognition using Free Open Data Sources
Daniel Menezes, Pedro Savarese, Ruy Milidiu
BRACIS 2019 PDF | URL
Teaching Assistant, Introduction to Computer Vision, UChicago, Winter 2021
Teaching Assistant, Fundamentals of Deep Learning, TTIC, Fall 2020
Teaching Assistant, Introduction to Statistical Machine Learning, TTIC, Fall 2018
Lecturer, Introduction to Machine Learning, TTI-Japan, Summer 2018
Teaching Assistant, Introduction to Machine Learning Summer School, TTIC, Summer 2018
Teaching Assistant, Algorithmic Techniques for Massive Data, Columbia University, Fall 2015
Teaching Assistant, Introduction to Computational Learning Theory, Columbia University, Fall 2015
Teaching Assistant, Algorithms for Data Science, Columbia University, Spring 2015
Reviewer NeurIPS (2019, 2020), ICML (2021), ICLR (2021), CVPR (2020, 2021), ECCV (2020), AAAI (2020), ICCV (2019), TPAMI (2020), JMLR (2021)
Undergrad Thesis Advisor, Pedro Bandeira de Mello Martins, 2017 Application of Generative Adversarial Networks for Black and White Image Colorization
Undergrad Thesis Advisor, Renan Araujo Lage, 2016 Sentiment Lexicon Generation with Continuous Polarities for Portuguese using Logistic Regressions and Semantic Modifiers