noth | Geraldin Nanfack

Biography

I come from Nkongsamba, a farming town in Cameroon (Africa).
I am a Postdoctoral Researcher at Conocordia University and at Mila under the supervision of Prof. Eugene Belilovsky. We work on robustness of interpretability methods and debiasing mechanisms of models/data.
Previously, I obtained my PhD (in Belguim) in November 2022 under the supervision of Benoît Frénay and Paul Temple. My PhD focused on the enforcement of constraints on machine learning models to make them more easily explainable and trustworthy.

Interests
  • Robust Mechanistic Interpretability
  • Constraint for interpretability, explainability and trustworthiness
  • Statistical guarantees in machine learning
  • Optimisation for machine learning
Education
  • PhD in Machine Learning, 2022

    University of Namur, Belgium

  • MSc in Data Science, 2018

    Ecole Nationale Supérieure d'Informatique et d'Analyse des Systèmes, Morocco

  • $\equiv$ BEng in Computer Engineering (2 years of Maths & Physics + 1 year of Computer Engineering), 2016

    Ecole Nationale Supérieure Polytechnique de Yaoundé, Cameroon

Publications

Quickly discover relevant content by filtering the full list of publications.
(2024). Adversarial Attacks on the Interpretation of Neuron Activation Maximization. In AAAI 2024.

PDF Cite

(2023). Learning Customised Decision Trees for Domain-knowledge Constraints. Pattern Recognition.

PDF Code Project

(2023). Decision trees: From efficient prediction to responsible AI. Frontiers in Artificial Intelligence, section Machine Learning and Artificial Intelligence.

PDF Project Link

(2022). Constraint Enforcement on Decision Trees: a Survey. ACM Computing Surveys.

PDF Cite Project

(2022). An Experimental Investigation into the Evaluation of Explainability Methods. AIMLAI workshop at ECML-PKDD.

PDF Code Project

Recent & Upcoming Talks

In progress…