Jules Dedieu

About me

I am a MS student in Industrial Engineering and Operations Research at University of California Berkeley. I also perform student research as part of AutoLab (part of BAIR) in which I work under the supervision of Professor Ken Goldberg. My main work and research interests are in data science, computer vision - including applications to healthcare- and natural language processing. I also conducted theoretical reinforcement learning research.

Prior to starting my Master, I did a 6-months research scientist internship at Therapanacea, in which I built deep-learning pipelines to predict personalized radiotherapy treatment plans relying solely on patient's MRI information. I then completed a 4-months research scientist internship at Tribun-Healthcare in which I built computer vision models to detect rare mitotic cells among large (10 GB) biopsies digitalized by different scanners. My research was focusing on getting the most accurate predictions while enforcing strong generalization compatibilities on unseen scanners.

Prior to joining UC Berkeley, I earned an MS in Computer Science from Ecole CentraleSupélec (Paris Saclay) . During that time, I performed research on statistics on a joined project between Servier and CentraleSupélec's MICS lab , where we aimed to build a mixed model simulating the action of a diabetes drug, and to perform statistical inference under partial differential equations.

---

Journal articles

• Robust Mitosis Detection Using a Cascade Mask-RCNN Approach With Domain-Specific Residual Cycle-GAN Data Augmentation. [Preprint]
  Springer LNCS, 2021. Jules Dedieu, Gauthier Roy, Capucine Bertrand, Alireza Moshayedi, Ali Mammadov, Stéphanie Petit, Saima Ben Hadj, Rutger H.J. Fick.

Conference articles

• Monte Carlo Augmented Actor-Critic for Sparse Reward Deep Reinforcement Learning from Suboptimal Demonstrations.
  Accepted, NeurIPS 2022. Albert Wilcox, Ashwin Balakrishna, Jules Dedieu, Daniel Brown, Wyame Benslimane, Ken Goldberg
• End-to-End VMAT Pelvis Dose Prediction & Treatment Planning Inference.
  ESTRO, 2020. Jules Dedieu, Kumar Shreshtha, Aurélien Lombard, Norbert Bus, Sonia Martinot, Rutger Fick, Nikos Paragios

Course Projects

• Reinforcement Learning Using Difference-of-Value Functions. Jules Dedieu, Dilara Aykanat.
• Mathematical Modelling of Diabetic Drugs (fr). Jules Dedieu, Alexandra Augusti, Gregoire Baverez.