Johanna Rondony
PhD student — Bio-inspired computer vision & complex-valued neural networks (CVNNs)
Contact
E-mail:
johanna.rondony[at]centralesupelec.fr
Links
Affiliation
I am a first-year PhD student in computer science at SONDRA, CentraleSupélec. My thesis is co-supervised by Jean-Philippe Ovarlez (SONDRA CentraleSupélec, DEMR ONERA) and Benoit Cottereau (IPAL CNRS @ NUS Singapore, CerCo CNRS).
Research interests
Computer vision · Complex-valued neural networks · Bio-inspired learning and neural synchrony · Unsupervised segmentation.
Current research
I study how object segmentation can emerge in a neural network from the synchrony between its neurons, rather than from explicit supervision. Using complex-valued and bio-inspired models, I run controlled experiments and ablations to pin down what actually drives this emergent binding. My results so far suggest it does not come from a specific architecture, but from a more global synchronization mechanism whose conditions remain surprisingly narrow and fragile.
Research visits
- Visiting researcher — IPAL, CNRS @ NUS, Singapore. [19th May 2025 – 20th June 2025] — one-month stay.
- Visiting researcher — IPAL, CNRS @ NUS, Singapore. [1st June 2026 – 4th July 2026] — one-month stay.
Publications
In preparation. My current work centres on understanding and characterizing emergent binding in complex-valued and bio-inspired networks.
Education
- PhD, Computer Science — SONDRA, CentraleSupélec, France. Oct. 2025 – present
- M.Eng. + M.Sc., General Engineering — ISAE-SUPAERO, France. Sep. 2023 – Aug. 2025 (Neuro-AI, Telecoms & Networks)
- B.E., Engineering — ENSTA, France. Sep. 2022 – Jun. 2023
- Preparatory class for engineering schools — Lycée Arago, France. Sep. 2019 – Jun. 2022
Hosted on GitHub Pages — Original theme by orderedlist