Pedro Abranches de Carvalho

Pedro Abranches de Carvalho

Postdoc CHUV — self supervised learning for timeseries

Biography

I am a postdoc working with Prof. Henri Lorach on building a foundational model for neurostimulation, combined with active learning approaches, for personalized neurostimulation.

I did my PhD at EPFL in the Laboratory for Information and Inference Systems (LIONS) led by Prof. Volkan Cevher and Neurorestore (.NR) center led by Prof. Grégoire Courtine and Prof. Jocelyne Bloch. My main responsibilities were regarding automating the search of stimulation protocols used in the project of .NR where we deal with spinal cord injury patients, by employing intelligent decision making systems. Main work was incorporated in a patent.

Until now, my academic journey was definitely not a straight path. My background involves neuroscience, engineering and computer science. I love to understand how things work and see how deep the rabbit hole is!

Interests

  • ML for science
  • Bayesian optimization
  • Neuroscience & neurotechnologies
  • Generative models and graph neural networks

Education

  • PhD in Electrical Engineering, 2020–2025

    EPFL, Switzerland

  • MS in Computer Science, 2017–2019

    Universidade do Porto, Portugal

  • BSc in Biochemistry, 2013–2017

    Universidade de Coimbra, Portugal

Work projects

*
kSE(x, x′) = e−d22ℓ2
0.19

PhD — main project 1

Automating the search for stimulation protocols used with spinal cord injury patients at .NeuroRestore, by employing intelligent decision making systems. The main work was incorporated in a patent.

Branches & other applications
BCI Award 2025
BCI Award 2025 video 3rd place. An implantable brain–spine interface restoring lower limb movements after complete spinal cord injury.
BCI Award 2024
BCI Award 2024 video 2nd place. A brain to spinal cord digital bridge to restore voluntary arm and hand movements after a spinal cord injury.
60 Minutes
60 Minutes video Press. Television feature covering the brain–spine interface work.

Personal projects

*
All Machine Learning Foundations

Causal Brain

Trying to infer a weak causality between activations in the brain using fMRI data (project started in class in need of a final finishing blow)

Building Pandora

There is much I do not know starting my PhD journey. This is a place where I hope to explain (to myself) some foundational concepts.

Publications

Highlighted

GG-GAN: A Geometric Graph Generative Adversarial Network

Treats graph generation from a geometric perspective, associating each node with a position in space and connecting edges through a similarity function.
Igor Krawczuk*, Pedro Abranches*, Andreas Loukas*, Volkan Cevher
ML
PDFOpenReview
Other
AllMLBiochemistryNeurotech