Adrien Dorise

Adrien Dorise holds a position as Research Engineer at CNES, France. He received the Engineering degree from INSA CVL in 2018, and his PhD in computer science and embedded systems from INSA Toulouse in 2022. His research interests include artificial intelligence, tiny machine learning, computer vision, explainable AI and space radiation.

He is the founder, developer and composer of Law Tech Productions, an indie game studio, and aims to develop this activity into the creation of serious games, combining his interest in interactive games with his research work.

What’s new?

Thesis

  1. Adrien Dorise, “Embedded anomaly detection – Machine learning based radiation hardening of space electronics”. Computer Science and Embedded Systems. INSA de Toulouse, 02/12/2022, hal: tel-03997861

Papers

  1. CVPR / AI4SPACE 2026; “Rethinking Satellite Image Restoration for Onboard AI: A Lightweight Learning-Based Approach“; Adrien Dorise, Marjorie Bellizzi, Omar Hlimi
    doi: CVPR 2026 Open Access Repository
  2. EvoAPP 2026, “Assessing Evolving and Learning-Based Controllers for Efficient Cursor Control in Human–Computer Interaction“; Joao Martin–Saquet, Adrien Dorise et al.
    doi: https://doi.org/10.1007/978-3-032-23604-3_15
  3. European Data Handling and Processing Conference (EDHPC) 2025; “Explaining raw data complexity to improve satellite onboard processing“, Adrien Dorise, Marjorie Bellizzi, Adrien Girard, Benjamin Francesconi, Stéphane May
    doi: Explaining raw data complexity to improve satellite onboard processing | IEEE Conference Publication | IEEE Xplore; arXiv: https://arxiv.org/abs/2510.06858
  4. Artificial Intelligence Revolutions (AIR), 2024, pp. 117-123; “Sequential Decision-Making in Atari 2600 Games: Comparing Temporal Features,”; J. Martin-Saquet, A. Dorise, E. Villain, S. Sanchez and D. Panzoli
    doi: 10.1109/AIR63653.2024.00027, hal: https://hal.science/hal-05069376/
  5. Radiation and its Effects on Components and Systems (RADECS) 2022; “Advanced machine learning for the detection of single event effects” ; Adrien Dorise, Audine Subias, Louise Travé-Massuyès, Corinne Alonso
    hal: hal-03789895
  6. IFAC Safeprocess 2022; “Dyd²: Dynamic Double anomaly Detection: Application to on-board space radiation faults”; Adrien Dorise, Louise Travé-Massuyès, Audine Subias, Corinne Alonso
    doi: 10.1016/j.ifacol.20, hal: hal-03609573v2.
  7. Radiation and its Effects on Components and Systems (RADECS) 2021; Machine learning as an alternative to thresholding for space radiation fault detection”; Adrien Dorise, Corinne Alonso, Audine Subias, Louise Travé-Massuyès, François Vacher, Leny Baczkowsky
    doi: 10.1109/RADECS53308.2021.9954582, hal: hal-03331030v1

Workshops

  1. European Conference of Non-Destructive Testing (ECNDT) 2023; “Experimental analysis of planar/volumetric defects in ultrasonics NDT: standardization of evaluation metrics using symbiosis of TOFD and TR-NEWS methods“; Serge Dos Santos, Giuseppe Nardoni, Adrien Dorise, Parnian Hemmati, Marco Feroldi, Pietro Nardoni
    doi: 10.13140/RG.2.2.16285.90087
  2. ePicture this 2023; “Machine learning for computer vision: a case study in human-machine interaction“; Boris Lenseigne, Julia Cohen, Adrien Dorise, Edouard Villain
    doi: 10.13140/RG.2.2.36695.44964.
  3. European Space Components Conference (ESCCON) 2021; “Creation of a database for Single events latch-up detection on Atmel SAM3X microcontroller”; Adrien Dorise, Louise Travé-Massuyès, Corinne Alonso, Audine Subias, François Vacher, Leny Baczkowsky
    hal: hal-03330093v1.
  4. ESA Workshop on Avionics, Data, Control and Software Systems (ADCSS) 2020; “Anomaly Detection for Radiation Hardening of Space Electronics – Application of Machine Learning Algorithms on an Atmel SAM3X Microcontroller”; Adrien Dorise, Louise Travé-Massuyès, Corinne Alonso, Audine Subias, François Vacher, Leny Baczkowsky
    hal: hal-03330033v1.

Posters

  1. ANITI meeting 2022, “Dynamic Double anomaly Detection through evolving clustering: Application to on-board space radiation faults“; Adrien Dorise, Louise Travé-Massuyès, Audine Subias, Corinne Alonso
    hal: hal-03622285.

Selected Projects

  • DragonflAI
    • DragonflAI is a novel way to implement, train and test artificial intelligence models. Similar to Keras for Tensorflow, it is thought to be an additional layer to the PyTorch API.
    • Experiment-centric API: Encapsulate data, model, training, testing, and visualization in one class you control.
    • Extensible hooks — Override key methods for training/testing/visualisation to fit simple to complex workflows.
  • Dynamic Double Anomaly Detection (DyD²)
    • DyD² is a machine learning method to perform embedded anomaly detection on dynamic systems.
    • DyD² has been designed to detect single effects in space missions.
  • Sound2Clip (On going)
    • Sound2Clip is a machine-learning research program that creates custom video clips from audio input.
    • At its core, it uses the DragonflAI API.
    • First results show that AI models easily interpret audio signals in the form of Fourier transforms.
  • Fluid Image Processing
    • Program analysing videos of a movig fluid to extract its charactertics (speed, plume’s height, plume’s diameter…).
    • It extracts a velocity vector field.

Teachings

Gema / Nexa

Ynov Campus / Université Champollion

  • Master level: Artificial Intelligence & Non-Playable Behaviours.

National Institute of Applied Sciences (INSA)

  • Master level: Machine Learning.
  • Master level: Deep Learning.
  • Master level: Data Visualization.
  • Bachelor level: Computer Science (C).
  • Bachelor level: Computer Science (ADA).

University of Toulouse

  • Bachelor level: Discrete Control.
  • Bachelor level: Linear Control.
Adrien_Dorise_resume_2026Download