Hugo Penedones

About me

I am a Machine Learning researcher and engineer, co-founder of Inductiva Research Labs, on a mission to blend Scientific Computing and Machine Learning (we are hiring!). Most recently, I worked at Google DeepMind, in London and Zurich. Prior to that, I worked in the Query Formulation team at Microsoft Bing and did research in Machine Learning and Computer Vision at Idiap Research Institute and École Polytechnique Fédérale de Lausane, both in Switzerland. I did my undergraduate studies in Informatics and Computing Engineering at FEUP, in Portugal, where I am originally from.

Talks

Scientific Machine Learning - guest speaker at the Deep Learning for NIR Chemometrics workshop, University of Algarve, 2024.

AI4Science - round table discussion with Cláudia Soares, Joel Arrais and Inês Dutra, organized by Deep Learning Sessions Portugal, 2023.

Tackling Scientific Problems with Deep Learning - hosted by The Science Circle, by Jousef Murad, 2022.

Deep Learning is Alive - round table discussion with Mário Figueiredo and Pedro Bizarro, organized by Deep Learning Sessions Portugal, 2022.

Reinforcement Learning and Long-Term Value Prediction at JNation 2020.

Real-World Reinforcement Learning - Challenges and Opportunities at DSPT day 2019.

Deep Reinforcement Learning in 25 minutes, for non-experts. This is talk I gave at Commit Porto 2018 for an audience of software developers, where I assumed no background in Machine Learning.

Publications

See my Google Scholar profile for an up to date list.

Computational Biology

[DeepMind's AlphaFold Nature paper] "Improved protein structure prediction using potentials from deep learning", Andrew W Senior, Richard Evans, John Jumper, James Kirkpatrick, Laurent Sifre, Tim Green, Chongli Qin, Augustin Žídek, Alexander WR Nelson, Alex Bridgland, Hugo Penedones, Stig Petersen, Karen Simonyan, Steve Crossan, Pushmeet Kohli, David T Jones, David Silver, Koray Kavukcuoglu, Demis Hassabis - In Nature journal. 15 Jan 2020. (read online)

Reinforcement Learning

"Adaptive Temporal-Difference Learning for Policy Evaluation with Per-State Uncertainty Estimates", Hugo Penedones, Carlos Riquelme, Damien Vincent, Hartmut Maennel, Timothy Mann, André Barreto, Sylvain Gelly, Gergely Neu - at NeurIPS - December 2019. (pdf)

"Temporal Difference Learning with Neural Networks - Study of the Leakage Propagation Problem", H. Penedones, D. Vincent, Hartmut Maennel, S. Gelly, T. Mann, A. Barreto - July 2018. (arXiv , pdf)

"Adaptive Lambda Least-Squares Temporal Difference Learning", T. Mann, H. Penedones, S. Mannor, T. Hester - December 2016. (arXiv , pdf)

Computer Vision

"Improving Object Classification using Pose Information", H. Penedones, R. Collobert, F. Fleuret, D. Grangier - Idiap Research Report 2011 (pdf)

"Playground Learning for Object Detection", H. Penedones, F. Fleuret - Idiap Internal Research Report 2009 (pdf)

Time Series

"Predicting Venus Express Thermal Power Consumption", H. Penedones, B. Sousa, A. Donati, J. Martinez-Heras - SpaceOps 2008 Conference Proceedings - Heidelberg, Germany. (pdf)

Book

In 2011, I wrote a book in Portuguese, expressing my views of the world. In a collection of very small chapters, I touch a wide range of topics, such as: democracy, economy, education, entrepreneurship, transparency, sustainability, peace, etc.

If you are interested, you can read the full book in digital formats for free, or order a paperback copy.

Source Code

Check my github profile to see whether I am working on something open source lately.

• luacnn - Convolutional Neural Network for hand digit recognition using Torch7 and Lua
• metrics - a Torch7 package to compute some metrics, such as area under ROC
• lakeml - some Machine Learning algorithms implemented in C++, e.g. AdaBoost, K-means, EM for (diagonal) Gaussian Mixture Models
• partitracker - a simple Particle Filter implemented in C++ and OpenCV
• sudoku - a Sudoku puzzle solver in pure C
• kdtree - a K-d tree C++ implementation, for Nearest Neighbours search
• likewise - a simple Python webserver for image similarity search, using the embeddings from a pre-trained convnet
• fractals - Zoomable Mandelbrot fractal renderings using PyGame
• acm_utils - Data structures and algorithms useful for programming competitions (trie, balanced tree, big ints, suffix arrays, etc.) and other small C or C++ pearls.

Collaborators

• Luís Sarmento - co-founder at Inductiva Research Labs
• Clara Gonçalves - co-founder at Inductiva Research Labs
• Andrew Senior - Google DeepMind
• Timothy Mann - Google DeepMind
• Rich Evans - Google DeepMind
• Piotr Mirowski - Google DeepMind
• David Grangier - Facebook Artificial Intelligence Research
• Ronan Collobert - Facebook Artificial Intelligence Research
• Francois Fleuret - Idiap Research Institute, EPFL
• Jose Martinez Heras - European Space Agency
• Alexessander Alves - Imperial College London

See my Google Scholar profile for a more complete list of scientific collaborators.

Inspiration

"Courage is not the absence of fear. It is acting in spite of it." - Mark Twain

"Live as if you were to die tomorrow. Learn as if you were to live forever." - Mahatma Gandhi

"Truth is what stands the test of experience." - Albert Einstein

"Write programs that do one thing and do it well. Write programs to work together." - Doug McIlroy