Education

• 2021: PhD, Université de Lyon, France

• 2017: Master’s Degree, Electronics & Computer Engineering, INSA Rennes, France

• 2015: Preparatory Cycle (scientific common core), INSA Rennes, France

Professional Experience

• 2022-pres.: Post-doctoral researcher, Universitat Pompeu Fabra, Spain

• 2021-2022: Research scientist, Orange Labs, France (industry)

• 2019-2021: Temporary teaching researcher assistant, ESIR, France (≈30 hours)

• 2018-2021: Doctoral researcher, Orange-INSA Lyon, France (joint industry-academia)

• 2018: Research intern, Orange Labs, France (industry)

• 2017: Research intern, Tallinn University of Technology, Estonia

Research Projects

Participation in funded research projects (collaborative researcher)

• ALIEN (2022-2026): ERC Advanced Grant. PI: Marco Baroni, Universitat Pompeu Fabra.

• REMEMBER (2020-2024): ANR funded by the French government. PI: Chistian Wolf, INSA Lyon.

Research fellowships

• CIFRE doctoral fellowship (joint industry-academia) (2018-2021). Funded by Orange and ANRT, France.

Publications

Articles in scientific journals

2021

Jaunet, T., Kervadec, C., Antipov, G., Baccouche, M., Vuillemot, R., and Wolf, C. (2021). VisQA: X-raying vision and language reasoning in transformers. In IEEE VIS, 2021.

2019

Ahmed, F., Kervadec, C., Le Moullec, Y., Tamberg, G., and Annus, P. (2019). Autonomous wireless sensor networks: Implementation of transient computing and energy prediction for improved node performance and link quality. The Computer Journal.

Articles in refereed conference proceedings

2025

Cheng, E., Doimo, D., Kervadec, C., Macocco, I., Yu L., Laio A., Baroni, M.. Emergence of a High-Dimensional Abstraction Phase in Language Transformers. In Proceedings of the 13th International Conference on Learning Representations (ICLR), 2025.

2023

Kervadec, C., Franzon, F., Baroni, M.. Unnatural language processing: How do language models handle machine-generated prompts? In Findings of the Association for Computational Linguistics: EMNLP 2023.

Cheng, E., Kervadec, C., Baroni, M.. Bridging Information-Theoretic and Geometric Compression in Language Models. In Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing (EMNLP).

2021

Kervadec, C., Wolf, C., Antipov, G., Baccouche, M., Vuillemot, R., Nadri, M.. Supervising the transfer of reasoning patterns in VQA. In Advances in Neural Information Processing Systems (NeurIPS), 2021.

Kervadec, C., Antipov, G., Baccouche, M., Wolf, C.. Roses are red, violets are blue… but should VQA expect them to? In IEEE Computer Vision and Pattern Recognition (CVPR), 2021.

Kervadec, C., Jaunet, T., Antipov, G., Baccouche, M., Vuillemot, R., Wolf, C.. How transferable are reasoning patterns in VQA? In IEEE Computer Vision and Pattern Recognition (CVPR), 2021.

2020

Kervadec, C., Antipov, G., Baccouche, M., Wolf, C.. Weak supervision helps emergence of word-object alignment and improves vision-language tasks. In European Conference on Artificial Intelligence (ECAI), 2020.

2019

Kervadec, C., Vielzeuf, V., Pateux, S., Jurie, F.. The Many Variations of Emotion. In IEEE International Conference on Automatic Face and Gesture Recognition (FG), 2019.

Articles in refereed workshop proceedings

2024

Leybzon, D., Kervadec, C.. Learning, forgetting, remembering: Insights from tracking LLM memorization during training. In Proceedings of the 7th BlackboxNLP Workshop: Analyzing and Interpreting Neural Networks for NLP, 2024.

2021

Kervadec, C., Jaunet, T., Antipov, G., Baccouche, M., Vuillemot, R., Wolf, C.. Biases and Reasoning Patterns in Visual Question Answering. In Visual Question Answering Workshop (CVPR), 2021.

2018

Kervadec, C., Vielzeuf, V., Pateux, S., Lechervy, A., Jurie, F.. Cake: Compact and accurate kdimensional representation of emotion. In Image Analysis for Human Facial and Activity Recognition (BMVC workshop), 2018.

Vielzeuf, V., Kervadec, C., Pateux, S., Lechervy, A., Jurie, F.. An occam’s razor view on learning audiovisual emotion recognition with small training sets. In International Conference on Multimodal Interaction (ICMI), 2018.

Open Resources

• GQA-OOD, Leader: Benchmark for out-of-distribution evaluation of VQA models. Freely available at https://github.com/gqa-ood/GQA-OOD

• VISQA, Co-Leader: Visualization tool for transformer-based VQA models. Freely available at https://visqa.liris.cnrs.fr/

Invited Presentations

• Bias and Reasoning in Visual Question Answering. LaP/RoM Seminar, University of Gothenburg. (20/10/23)

• Unnatural Language Processing. REST-CL, Barcelona, Spain. (01/07/23)

• Exploring the Unnaturalness of Large Language Models. NLP meetup, Barcelona, Spain. (14/03/23)

• Planes Don’t Flap Their Wings: Insights from Neurosciences. Orange Datacamp, Lannion, France. (08/09/22)

• Learning High-Level Reasoning in Vision, Language, and Robotics. INRIA, France. (28/05/21)

• Roses Are Red, Violets Are Blue… but Should VQA Expect Them To? GdR ISIS, France. (07/04/21)

Supervision and Teaching

Teaching

• Image processing. Digital Imaging Master at ESIR, France. (2019, 2021, ≈30 hours)

Master’s thesis supervision

• Co-supervisor, Paula Rivera, Universitat Pompeu Fabra. (pres.)

• Supervisor, Eder Ruiz, Universitat Pompeu Fabra. (2024)

• Supervisor, Danny Leybzon, Universitat Pompeu Fabra. (2024)

• Co-supervisor, Fernando Da Silva, Universitat Pompeu Fabra. (2023)

• Co-supervisor, Pierre Marza, INSA Lyon. (2021)

Master’s thesis committees

2-3 MA thesis committees per year since 2023.

Evaluation

Reviewing journals

• TPAMI (2023, 2024)

Reviewing conferences

• 2025: COLM

• 2024: NeurIPS, ICLR

• 2023: LIMO, CVPR

• 2022: CVPR (outstanding reviewer), ECCV (outstanding reviewer)

• 2021: NeurIPS, ICCV (outstanding reviewer), BMVC

Other Merits

• 2023: Marie Skłodowska-Curie’s Seal of Excellence. UNLACE: A project on LM Interpretability (not funded)

Languages

• French: native

• English: proficient

• Spanish (Castilian): proficient

• Catalan: intermediate

Programming

• Languages: Python, C/C++, Linux

• ML frameworks: Pytorch, Tensorflow