About (formal bio)

I am an assistant professor (tenure-track) and Hi! Paris chair holder in the center for applied math (CMAP) of the École Polytechnique, France, broadly interested in optimization, machine learning, signal processing, and control, and in particular, in the intersections of these fields. In my research, I develop and analyze tools that enable intelligent systems to extract, process, and act on information. That is,

• gather data, e.g., using sampling^{[ 1 L. F. O. Chamon and A. Ribeiro. Greedy sampling of graph signals. IEEE Trans. on Signal Process., 66[1]:34–47, 2018. , 2 L. F. O. Chamon, G. J. Pappas, and A. Ribeiro. Approximate supermodularity of Kalman filter sensor selection. IEEE Trans. on Autom. Control., 66[1]:49–63, 2021. ]} and active learning^{[ 3 L. F. O. Chamon and A. Ribeiro. Approximate supermodularity bounds for experimental design. In Conference on Neural Information Processing Systems (NeurIPS), 5403–5412. 2017. ]} techniques,
• turn this data into insights, e.g., using ML^{[ 4 L. F. O. Chamon and A. Ribeiro. Probably approximately correct constrained learning. In Conference on Neural Information Processing Systems (NeurIPS). 2020. , 5 A. Robey*, L. F. O. Chamon*, G. J. Pappas, H. Hassani, and A. Ribeiro. Adversarial robustness with semi-infinite constrained learning. In Conference on Neural Information Processing Systems (NeurIPS). 2021. (* equal contribution). , 6 A. Robey, L. F. O. Chamon, G. J. Pappas, and H. Hassani. Probabilistically robust learning: Balancing average- and worst-case performance. In International Conference on Machine Learning (ICML). 2022. , 7 L. F. O. Chamon, S. Paternain, M. Calvo-Fullana, and A. Ribeiro. Constrained learning with non-convex losses. IEEE Trans. on Inf. Theory, 69[3]:1739–1760, 2023. ]}, geometric ML^{[ 8 L. Ruiz, L. F. O. Chamon, and A. Ribeiro. Graphon neural networks and the transferability of graph neural networks. In Conference on Neural Information Processing Systems (NeurIPS). 2020. , 9 L. Ruiz, L. F. O. Chamon, and A. Ribeiro. Graphon signal processing. IEEE Trans. on Signal Process., 69:4961–4976, 2021. , 10 L. Ruiz, L. F. O. Chamon, and A. Ribeiro. Transferability properties of graph neural networks. IEEE Trans. on Signal Process., 71:3474–3489, 2023. ]}, and statistical methods^{[ 11 L. F. O. Chamon, Y. C. Eldar, and A. Ribeiro. Functional nonlinear sparse models. IEEE Trans. on Signal Process., 68[1]:2449–2463, 2020. , 12 M. Peifer, L. F. O. Chamon, S. Paternain, and A. Ribeiro. Sparse multiresolution representations with adaptive kernels. IEEE Trans. on Signal Process., 68[1]:2031–2044, 2020. , 13 D. S. Kalogerias, L. F. O. Chamon, G. J. Pappas, and A. Ribeiro. Better safe than sorry: Risk-aware nonlinear Bayesian estimation. In IEEE International Conference in Acoustic, Speech, and Signal Processing (ICASSP). 2020. , 14 B. Arzani, S. Ciraci, L. F. O. Chamon, Y. Zhu, H. Liu, J. Padhye, B. T. Loo, and G. Outhred. 007: Democratically finding the cause of packet drops. In USENIX Symposium on Networked Systems Design and Implementation (NSDI), 419–435. 2018. ]},
• and turn these insights into actions e.g., using resource allocation^{[ 15 M. Eisen, C. Zhang, L. F. O. Chamon, D. D. Lee, and A. Ribeiro. Learning optimal resource allocations in wireless systems. IEEE Trans. on Signal Process., 67[10]:2775–2790, 2019. ]}, optimal control^{[ 16 L. F. O. Chamon, A. Amice, S. Paternain, and A. Ribeiro. Resilient control: Compromising to adapt. In IEEE Control and Decision Conference. 2020. , 17 L. F. O. Chamon, S. Paternain, and A. Ribeiro. Counterfactual programming for optimal control. In Learning for Dynamics & Control (L4DC). 2020. , 18 L. F. O. Chamon, A. Amice, and A. Ribeiro. Approximately supermodular scheduling subject to matroid constraints. IEEE Trans. on Autom. Control., 67[3]:1384–1396, 2022. ]}, and reinforcement learning^{[ 19 S. Paternain, L. F. O. Chamon, M. Calvo-Fullana, and A. Ribeiro. Constrained reinforcement learning has zero duality gap. In Conference on Neural Information Processing Systems (NeurIPS), 7555–7565. 2019. , 20 S. Paternain, M. Calvo-Fullana, L. F. O. Chamon, and A. Ribeiro. Safe policies for reinforcement learning via primal-dual methods. IEEE Trans. on Autom. Control., 68[3]:1321–1336, 2023. , 21 M. Calvo-Fullana, S. Paternain, L. F. O. Chamon, and A. Ribeiro. State augmented constrained reinforcement learning: Overcoming the limitations of learning with rewards. IEEE Trans. on Autom. Control., 69[7]:4275–4290, 2024. ]} algorithms.

My goal is to design AI systems that can learn and adapt with minimal human intervention while ensuring that they comply with rigorous operational requirements, such as robustness^{[ 4 L. F. O. Chamon and A. Ribeiro. Probably approximately correct constrained learning. In Conference on Neural Information Processing Systems (NeurIPS). 2020. , 5 A. Robey*, L. F. O. Chamon*, G. J. Pappas, H. Hassani, and A. Ribeiro. Adversarial robustness with semi-infinite constrained learning. In Conference on Neural Information Processing Systems (NeurIPS). 2021. (* equal contribution). , 6 A. Robey, L. F. O. Chamon, G. J. Pappas, and H. Hassani. Probabilistically robust learning: Balancing average- and worst-case performance. In International Conference on Machine Learning (ICML). 2022. , 7 L. F. O. Chamon, S. Paternain, M. Calvo-Fullana, and A. Ribeiro. Constrained learning with non-convex losses. IEEE Trans. on Inf. Theory, 69[3]:1739–1760, 2023. ]}, fairness^{[ 4 L. F. O. Chamon and A. Ribeiro. Probably approximately correct constrained learning. In Conference on Neural Information Processing Systems (NeurIPS). 2020. , 7 L. F. O. Chamon, S. Paternain, M. Calvo-Fullana, and A. Ribeiro. Constrained learning with non-convex losses. IEEE Trans. on Inf. Theory, 69[3]:1739–1760, 2023. ]}, safety^{[ 19 S. Paternain, L. F. O. Chamon, M. Calvo-Fullana, and A. Ribeiro. Constrained reinforcement learning has zero duality gap. In Conference on Neural Information Processing Systems (NeurIPS), 7555–7565. 2019. , 20 S. Paternain, M. Calvo-Fullana, L. F. O. Chamon, and A. Ribeiro. Safe policies for reinforcement learning via primal-dual methods. IEEE Trans. on Autom. Control., 68[3]:1321–1336, 2023. , 21 M. Calvo-Fullana, S. Paternain, L. F. O. Chamon, and A. Ribeiro. State augmented constrained reinforcement learning: Overcoming the limitations of learning with rewards. IEEE Trans. on Autom. Control., 69[7]:4275–4290, 2024. ]}, smoothness^{[ 22 J. Cervino, L. F. O. Chamon, B. D. Haeffele, R. Vidal, and A. Ribeiro. Learning globally smooth functions on manifolds. In International Conference on Machine Learning (ICML). 2023. ]}, and invariance^{[ 23 I. Hounie, L. F. O. Chamon, and A. Ribeiro. Automatic data augmentation via invariance-constrained learning. In International Conference on Machine Learning (ICML). 2023. ]}.

For more information, you can explore my research projects, check out my CV, read a more formal bio, or learn more about some of my personal interests.

My background

I did my Ph.D. at the University of Pennsylvania immediately followed by a postdoc at the Simons Institute of the University of California, Berkeley. During this time, I developed

• near-optimal selection methods for experimental design^{[ 3 L. F. O. Chamon and A. Ribeiro. Approximate supermodularity bounds for experimental design. In Conference on Neural Information Processing Systems (NeurIPS), 5403–5412. 2017. ]}, actuator scheduling^{[ 18 L. F. O. Chamon, A. Amice, and A. Ribeiro. Approximately supermodular scheduling subject to matroid constraints. IEEE Trans. on Autom. Control., 67[3]:1384–1396, 2022. ]}, and sampling^{[ 1 L. F. O. Chamon and A. Ribeiro. Greedy sampling of graph signals. IEEE Trans. on Signal Process., 66[1]:34–47, 2018. , 2 L. F. O. Chamon, G. J. Pappas, and A. Ribeiro. Approximate supermodularity of Kalman filter sensor selection. IEEE Trans. on Autom. Control., 66[1]:49–63, 2021. ]};
• the theoretical and algorithmic foundations of sparse functional programs, showing that sparsity is tractable on the continuum^{[ 11 L. F. O. Chamon, Y. C. Eldar, and A. Ribeiro. Functional nonlinear sparse models. IEEE Trans. on Signal Process., 68[1]:2449–2463, 2020. ]} and can be used for nonlinear line spectral estimation^{[ 11 L. F. O. Chamon, Y. C. Eldar, and A. Ribeiro. Functional nonlinear sparse models. IEEE Trans. on Signal Process., 68[1]:2449–2463, 2020. ]}, fitting multi-resolution kernel models^{[ 12 M. Peifer, L. F. O. Chamon, S. Paternain, and A. Ribeiro. Sparse multiresolution representations with adaptive kernels. IEEE Trans. on Signal Process., 68[1]:2031–2044, 2020. ]}, and learn Gaussian Processes from data^{[ 24 L. F. O. Chamon, S. Paternain, and A. Ribeiro. Learning Gaussian processes with Bayesian posterior optimization. In Asilomar Conference on Signals, Systems and Computers, 482–486. 2019. ]};
• a graphon signal processing framework (in collaboration with Luana Ruiz)^{[ 9 L. Ruiz, L. F. O. Chamon, and A. Ribeiro. Graphon signal processing. IEEE Trans. on Signal Process., 69:4961–4976, 2021. ]} that we used to demonstrate the transferability of graph neural networks^{[ 8 L. Ruiz, L. F. O. Chamon, and A. Ribeiro. Graphon neural networks and the transferability of graph neural networks. In Conference on Neural Information Processing Systems (NeurIPS). 2020. , 25 L. Ruiz, L. F. O. Chamon, and A. Ribeiro. Transferable graph neural networks on large-scale stochastic graphs. In Asilomar Conference on Signals, Systems and Computers. 2021. ]};
• a new risk-aware estimator (together with Dionysios Kalogerias) that received the best paper award at ICASSP 2020;
• constrained learning and reinforcement learning theory and algorithms to enable the design of intelligent systems that satisfy requirements such as robustness^{[ 4 L. F. O. Chamon and A. Ribeiro. Probably approximately correct constrained learning. In Conference on Neural Information Processing Systems (NeurIPS). 2020. , 5 A. Robey*, L. F. O. Chamon*, G. J. Pappas, H. Hassani, and A. Ribeiro. Adversarial robustness with semi-infinite constrained learning. In Conference on Neural Information Processing Systems (NeurIPS). 2021. (* equal contribution). , 6 A. Robey, L. F. O. Chamon, G. J. Pappas, and H. Hassani. Probabilistically robust learning: Balancing average- and worst-case performance. In International Conference on Machine Learning (ICML). 2022. , 7 L. F. O. Chamon, S. Paternain, M. Calvo-Fullana, and A. Ribeiro. Constrained learning with non-convex losses. IEEE Trans. on Inf. Theory, 69[3]:1739–1760, 2023. ]}, fairness^{[ 4 L. F. O. Chamon and A. Ribeiro. Probably approximately correct constrained learning. In Conference on Neural Information Processing Systems (NeurIPS). 2020. , 7 L. F. O. Chamon, S. Paternain, M. Calvo-Fullana, and A. Ribeiro. Constrained learning with non-convex losses. IEEE Trans. on Inf. Theory, 69[3]:1739–1760, 2023. ]}, safety^{[ 19 S. Paternain, L. F. O. Chamon, M. Calvo-Fullana, and A. Ribeiro. Constrained reinforcement learning has zero duality gap. In Conference on Neural Information Processing Systems (NeurIPS), 7555–7565. 2019. , 20 S. Paternain, M. Calvo-Fullana, L. F. O. Chamon, and A. Ribeiro. Safe policies for reinforcement learning via primal-dual methods. IEEE Trans. on Autom. Control., 68[3]:1321–1336, 2023. , 21 M. Calvo-Fullana, S. Paternain, L. F. O. Chamon, and A. Ribeiro. State augmented constrained reinforcement learning: Overcoming the limitations of learning with rewards. IEEE Trans. on Autom. Control., 69[7]:4275–4290, 2024. ]}, smoothness^{[ 22 J. Cervino, L. F. O. Chamon, B. D. Haeffele, R. Vidal, and A. Ribeiro. Learning globally smooth functions on manifolds. In International Conference on Machine Learning (ICML). 2023. ]}, and invariance^{[ 23 I. Hounie, L. F. O. Chamon, and A. Ribeiro. Automatic data augmentation via invariance-constrained learning. In International Conference on Machine Learning (ICML). 2023. ]}.

After that, I spent two years as the ELLIS–SimTech independent research group leader of the University of Stuttgart, Germany.

Prior to moving to the US, I received my bachelor and master degrees in electrical engineering from the University of São Paulo, Brazil, where I am originally from. My masters thesis was on combinations of adaptive filters, but I also worked on innovative digital design projects (with David Lamb), one of which was patented by Analog Devices. As an undergraduate, I studied acoustics during an exchange at the École Centrale de Lyon and INSA-Lyon, France.

Before starting my graduate studies, I spent a semester teaching certifying courses and consulting on nondestructive testing at INSACAST Formation Continue in Lyon, France, and worked as a signal processing researcher and statistics consultant on a project with EMBRAER.

Personal interests

In my spare time, I like to write and play music. Mostly the acoustic guitar and the piano, but I sometimes pretend to play the violin, the flute, and typical Brazilian percussion instruments, such as the zabumba and the triangle. I used to play the accordion in a forró band named Quaraçá, meaning sun ray or sun light in a Brazilian indigenous language. Forró is a folk music genre from the Northeast of Brazil that is surprisingly widespread worldwide: I've managed to find some underground Forró dance party in almost every city I've lived in.

Back in Brazil, I worked in many recording studios (most small, some even smaller) and was involved in the production of international theater and art exhibitions, notably Bob Wilson's "Quartett" (with Théâtre de l’Odéon, France) and Wajdi Mouawad's "The three sisters" (with Théâtre du Trident, Canada). This is actually how I got interested in electrical engineering and later, signal processing. Amidst the 2020 pandemic, I went back to recording in what served as my bedroom/office/gym/homestudio and released my first solo EP, Philathina, in July 2022. You can check it out on my music website or any major music streaming service (Spotify, Apple music, Amazon music, YouTube, Tidal...).

While I don't play live anymore, I've made a few guest appearances on Tau-zeta's YouTube channel, the first of which singing (if you can call that "singing") one of my favorite Greek songs.

 

 

Besides music, I like learning languages and discovering cultures. I speak a few languages to varying degrees of success, the latest addition being German with degree unsuccessful. I've also been pretending to speak Greek for a couple of years now. I don't always pretend very well.

During my Ph.D., I was part of a group of soccer lovers that founded the Philadelphia Open Soccer (or on Facebook), a program that taught soccer to kids in underprivileged public schools of West Philadelphia and in a Northeast Philadelphia community composed largely of immigrants and refugees.

Although I don't take soccer as seriously as you'd expect from a Brazilian, I do like watching the Seleção (Brazilian national team). And I understand if right now the thought of mentioning the 2014 "7-1" fiasco against Germany crossed your mind. I'll get back to you on that as soon as your national team also manages to win five World Cups...