ML Safety Researcher
& Assistant Professor
University of the Basque Country
(UPV/EHU)

Jon Vadillo

I am an Assistant Professor at the University of the Basque Country (UPV/EHU), where I conduct research on adversarial machine learning in the Intelligent Systems Group.

I earned a PhD in Artificial Intelligence in 2023 from UPV/EHU under the supervision of Dr. Roberto Santana and Prof. Jose A. Lozano, investigating how adversarial attacks can manipulate deep learning models. Before that, I obtained an MSc in Computational Engineering and Intelligent Systems and a BSc in Computer Science from UPV/EHU. I was also a visiting researcher at the University of Oxford in 2022, under the supervision of Prof. Marta Kwiatkowska.

My current research focuses on the vulnerabilities and trustworthiness of deep learning models in adversarial scenarios, with a particular interest in privacy attacks and the implications of adversarial examples in explainable models.

Publications (Google Scholar)

Adversarial Attacks in Explainable Machine Learning: A Survey of Threats Against Models and Humans.
WIREs Data Mining and Knowledge Discovery, e1567. DOI: 10.1002/widm.1567.
Vadillo, J., Santana, R. and Lozano, J.A. (2024).

Extending Adversarial Attacks to Produce Adversarial Class Probability Distributions.
Journal of Machine Learning Research, 24(15). pp. 1-42. Featured at ICML 2024.
Vadillo, J., Santana, R. and Lozano, J.A. (2023).

Uncertainty-Aware Explanations Through Probabilistic Self-Explainable Neural Networks.
Preprint arXiv:2403.13740.
Vadillo, J., Santana, R., Lozano, J.A., Kwiatkowska, M. (2024).

Analysis of Dominant Classes in Universal Adversarial Perturbations.
Knowledge-Based Systems, 236. pp. 107719. Elsevier. DOI: 10.1016/j.knosys.2021.107719.
Vadillo, J., Santana, R. and Lozano, J.A. (2022).

On the Human Evaluation of Audio Adversarial Examples.
Computers & Security, 112. pp. 102495. Elsevier. DOI: 10.1016/j.cose.2021.102495.
Vadillo, J. and Santana, R. (2022).

Adversarial Perturbations for Evolutionary Optimization.
Machine Learning, Optimization, and Data Science (LOD). LNCS, 13164. pp. 408-422. Springer. DOI: 10.1007/978-3-030-95470-3_31.
Garciarena, U., Vadillo, J., Mendiburu, A. and Santana, R. (2022).

When and How to Fool Explainable Models (and Humans) with Adversarial Examples.
Preprint, arXiv:2107.01943.
Vadillo, J., Santana, R. and Lozano, J.A. (2021).

Universal Adversarial Examples in Speech Command Classification.
Proceedings of the XIX Conference of the Spanish Association for Artificial Intelligence (CAEPIA). pp. 642-647
Vadillo, J. and Santana, R. (2021).

Exploring Gaps in DeepFool in Search of More Effective Adversarial Perturbations.
In Machine Learning, Optimization, and Data Science (LOD). LCNS, 12566. pp. 215-227. Springer. DOI: 10.1007/978-3-030-64580-9_18.
Vadillo, J., Santana, R. and Lozano, J.A. (2020).

I4TSPS: a Visual-Interactive Web System for Industrial Time-Series Pre-processing.
IEEE International Conference on Big Data (Big Data), pp. 2012-2018. IEEE. DOI: 10.1109/BigData.2018.8621887.
Villalobos, K., Vadillo, J., Diez, B., Calvo, B., & Illarramendi, A. (2018).

Education

PhD - Computer Science
- Title: "Broadening the Horizon of Adversarial Attacks in Deep Learning"
- Grade: Cum Laude
- Four-month research stay at the University of Oxford (UK), supervised by Prof. Marta Kwiatkowska.

M. Sc. - Computational Engineering and Intelligent Systems
- Average grade: 9.81
- Final project passed with honours
- 3 subjects passed with honours (out of 10)

Bachelor's Degree in Computer Sciences
- Extraordinary Award to the Best Academic Record
- Average grade: 9.52
- Final project passed with honours
- 24 subjects passed with honours (out of 36)

Other courses

3rd ACM Europe Summer School on Data Science - Athens (Greece).

Grants and Awards

Award to the Best Ph. D. Project

Doctoral Consortium of the XIX Conference of the Spanish Association for Artificial Intelligence (CAEPIA 20/21).

Special Award to the Best Academic Record - Bachelor's Degree on Computer Sciences

ACM International Collegiate Programming Contest (ICPC)
- Honorable mentions in the South Western Europe Regional Contest (SWERC) in 2016 and 2017.

Granted collaboration with a university research group - Ikasiker Scholarship Plan
- Interoperable Databases Group - University of the Basque Country (UPV/EHU)

Dissemination

Do You Trust Your Speech Recognition System? Fooling Deep Learning with (Inaudible) Audio Perturbations.
SIAM News. Vadillo, J., Santana, R. and Lozano, J.A. (2022).

Even Robots Can Be Fooled, but They're Getting Smarter.
SYFY Wire. Rayne, E. (2022).

Contact

Feel free to contact me for potential collaborations or question!