Adrien Chan Hon Tong

I am a researcher in machine learning at ONERA since 2014 in the IVA team.

I have worked from 2014 to 2018 on the development of object detection algorithm working on satellite images and/or UAV videos (for military purpose).

Since 2018, I focus on safety issues raised by machine learning on critical plateform (autonomous driving, lethal autonomous weapon systems): performance evaluation on operationnal condition, robustness to adversarial attack and data poisoning

Somes publications

Computer vision

• OFF THE SHELF DEEP LEARNING PIPELINE FOR REMOTE SENSING APPLICATIONS
  • 2017
  • http://www.boulch.eu/files/2017_bids_esa_shelf-dl.pdf
• Object Detection in Remote Sensing Images with Center Only.
  • 2018
  • https://doi.org/10.1109/IGARSS.2018.8517860

Machine learning

• An Algorithm for Generating Invisible Data Poisoning Using Adversarial Noise That Breaks Image Classification Deep Learning
  • 2019
  • https://doi.org/10.3390/make1010011
• Learning to Understand Earth Observation Images with Weak and Unreliable Ground Truth
  • 2019
  • https://doi.org/10.1109/IGARSS.2019.8898563