Teaching

I designed two courses hosted in the ENS Cognitive Science master's program. Below you will find example syllabi from previous years.

Mathematical tools for data science, machine learning, and statistical modeling

Synopsis: This course provides a foundational overview of the core mathematical concepts that underpin modern data science, machine learning, and statistical modeling. Topics include linear algebra, probability theory, optimization, and elements of information theory and numerical methods. The emphasis is on developing both mathematical intuition and practical skills, enabling students to understand and implement common algorithms and models from first principles. While mathematical derivations will frequently be demonstrated, these will be curated to guide intuition rather than to provide an entirely rigorous foundation for the material. Some lectures may provide an overview of many different techniques (e.g., a survey of linear dimensionality reduction methods), while others will delve deep into a single fundamental concept. This course is designed for students with prior quantitative training (background in math, physics, or CS) seeking to strengthen their skills in order to engage critically with computational methods in data-driven research.

Prerequisites: A solid foundation in calculus, differential equations, linear algebra, and probability theory, as well as proficiency in Python.

Course format: Chalkboard lectures and practical coding sessions.

Assessment:

Homework assignments (3 x 20%). These will combine light analytical exercises with practical coding tasks exploring real-world datasets. These assignments are designed to reinforce core mathematical concepts and their application in data analysis and modeling.
Final project (40%). Students will prepare a journal club style presentation of a contemporary research paper. The focus will be on critically examining the use of a statistical model within the paper and explaining how it was applied to address a central scientific question.

Topics covered: Point estimation, Fisher information, Cramer-Rao bound, bias-variance decomposition, hypothesis testing, multivariate linear regression, L1/L2 regularization, SVD, dimensionality reduction, cross-validation, information criteria.

Cognitive models and artificial intelligence

Synopsis: This course explores the intersection of neuroscience, cognitive science, and artificial intelligence, with a focus on computational models that bridge brain function and intelligent behavior. Students will engage with a broad range of models spanning spatial scales—from single neurons to neural networks to social interactions between agents. Normative theories such as efficient coding, the Bayesian brain hypothesis, and reinforcement learning will also be introduced. Emphasis will be placed on understanding the assumptions behind each model, the questions it aims to address, and the experimental predictions it generates. While key equations will be discussed to build intuition, students will not be required to solve or manipulate them. The course is designed to equip students with a conceptual foundation for how computational models are used to explain brain function and inform the design of intelligent systems.

Prerequisites: None

Course format: Weekly powerpoint lectures.

Assessment:

Quizzes (15%).
Midterm (35%). These will combine light analytical exercises with practical coding tasks exploring real-world datasets. These assignments are designed to reinforce core mathematical concepts and their application in data analysis and modeling.
Final exam (50%).

Topics: Connectionism vs GOFAI, Hodgkin-Huxley model, neural population coding, neural manifolds, deep neural networks, Hopfield networks, continuous attractors, signal detection theory, drift diffusion model, Bayesian brain hypothesis, reinforcement learning.

Page updated

Google Sites

Report abuse