JUAN CERVINO




jcervino@seas.upenn.edu

Google Scholar

Resume

Github / Linkedin / Twitter

Ph.D. student at the University of Pennsylvania

3401 Walnut Street, Rm 405B, Philadelphia, PA



About me

I am a PhD student in Electrical and System Engineering (ESE) at the University of Pennsylvania, my advisor is Alejandro Ribeiro. I am interested in seeing machine learning through the lens of optimization theory. My interests include optimization, machine learning, and control. Recently, I was named a rising star in signal processing at ICASSP 2023. In summer 2023, I interned on Google's Geometric AI team led by Ameesh Makadia.

I was born in Montevideo, Uruguay where I graduated from Universidad de la Republica in 2018 as an Electrical Engineer. My industry working experience are more than 2 years in Industrial Control and Robotics. I did my undergraduate's internship in medical electronics at Integer.


Bio

Juan received the B.Sc. degree in electrical engineering from the Universidad de la Republica Oriental del Uruguay, Montevideo, in 2018. He is now a PhD student in the Department of Electrical and Systems Engineering at the University of Pennsylvania, supervised by Professor Alejandro Ribeiro. Juan was named a rising star in signal processing in 2023. Juan's current research interests are in machine learning, optimization and control.


Research

2024

FastSample: Accelerating Distributed Graph Neural Network Training for Billion-Scale Graphs
Hesham Mostafa, Adam Grabowski, Md Asadullah Turja, Juan Cervino, Alejandro Ribeiro, Nageen Himayat
to appear at DAC 2024.[PDF]


2023

Learning Globally Smooth Functions on Manifolds
Juan Cervino, Luiz F. O. Chamon, Benjamin D Haeffele, Rene Vidal, Alejandro Ribeiro
presented at ICML 2023.[PDF]


Training Graph Neural Networks on Growing Stochastic Graphs
Juan Cervino, Luana Ruiz, Alejandro Ribeiro
presented at ICASSP 2023.[PDF]


Multi-task Bias-Variance Trade-off Through Functional Constraints
Juan Cervino, Juan Andres Bazerque, Miguel Calvo-Fullana, Alejandro Ribeiro
presented at ICASSP 2023.[PDF]


Learning by Transference: Training Graph Neural Networks on Growing Graphs
Juan Cervino, Luana Ruiz, Alejandro Ribeiro
published at IEEE Transactions on Signal Processing.[PDF]


Intrinsically motivated graph exploration using network theories of human curiosity
Shubhankar P. Patankar, Mathieu Ouellet, Juan Cervino, Alejandro Ribeiro, Kieran A. Murphy, Dani S. Bassett
presented at Learning on Graphs Conference 2023.[PDF]


2022

An Agnostic Approach to Federated Learning with Class Imbalance
Zebang Shen, Juan Cervino, Hamed Hassani, Alejandro Ribeiro
presented at ICLR 2022.[PDF] [Code]


Federated Representation Learning via Maximal Coding Rate Reduction
Juan Cervino, Navid NaderiAlizadeh, Alejandro Ribeiro
available online.[PDF] [Code]


Training stable graph neural networks through constrained learning
Juan Cervino, Luana Ruiz, Alejandro Ribeiro
presented at ICASSP 2022.[PDF] [Code]


2021

Multi-task Reinforcement Learning in Reproducing Kernel Hilbert Spaces via Cross-learning
Juan Cervino, Juan Andres Bazerque, Miguel Calvo-Fullana, Alejandro Ribeiro
published at IEEE Transactions on Signal Processing.[PDF]


2020

Multi-task Supervised Learning via Cross-learning
Juan Cervino, Juan Andres Bazerque, Miguel Calvo-Fullana, Alejandro Ribeiro
presented at the European Signal Processing Conference (EUSIPCO).[PDF]


Parameter Critic: a Model Free Variance Reduction Method Through Imperishable Samples
Juan Cervino, Harshat Kumar, Alejandro Ribeiro
available online.[PDF]


2019

Meta-learning through coupled optimization in reproducing kernel hilbert spaces
Juan Cervino, Juan Andres Bazerque, Miguel Calvo-Fullana, Alejandro Ribeiro
presented at the American Control Conference (ACC).[PDF]



Personal

On a personal note, I am a stereotypical uruguayan: I drink mate every day, I play soccer every week, and I take the opposite side on every discussion. I was born and raised in the beautiful city of Montevideo, a couple of blocks away from the sign that can be seen in the image below. One of my favourite books is the novel 'La Tregua' from Mario Benedetti, which narrates a 'truce' in the tough life of another stereotypical uruguayan. Among my favourite painters, there is Jose Cuneo who painted bucholic moons, and Jose Gurvich whose museum in Montevideo is a must-visit. The incredibly talented Jorge Drexler, not only has wonderful songs, amazing lyrics, but also a very interesting TEDx about 'Poetry, music and identity'.

Montevideo


Miscellaneous

I once wrote an op-ed on a Uruguayan newpaper (La Diaria) in regards to a tax increase: Es una suba de impuestos de aca a Delaware. The setting is the following: consider there is a country (Uruguay) where the value added tax is 22%, and you only pay 19% if you pay with a card (as opposed to cash). If the government removes the benefit of paying with card, are they increasing the tax or not? My answer is: yes of course.


Hosted on Eniac in the beautiful city of Philadelphia. Last updated February 2024.