Research Associate

About INRIA

INRIA is a leading French research institute dedicated to advancing computer science and digital technologies. Located in the heart of the vibrant Saclay plateau, our center fosters collaboration with prestigious institutions like Paris-Saclay University and the Polytechnic Institute of Paris.

Research Opportunity: Two-Layer Neural Networks

We are seeking a highly motivated Research Associate to join our team and contribute to cutting-edge research in the field of deep learning. Under the guidance of Dr. Etienne Boursier, you will delve into the fascinating world of two-layer neural networks with ReLU activation functions.

Your Mission:

• Conduct a comprehensive literature review on the theoretical foundations of two-layer neural networks and their training dynamics.
• Develop and implement numerical simulations using Python to explore the impact of stochastic gradient descent (SGD) on network training.
• Seek theoretical insights by applying mathematical tools from non-convex optimization and stochastic differential equations to understand the behavior of these networks.

Required Skills:

• Strong understanding of neural network theory and architectures.
• Proficiency in reading and interpreting scientific literature in English.
• Excellent programming skills in Python, particularly for numerical simulations.
• Familiarity with non-convex optimization techniques and their application to machine learning.
• Basic knowledge of stochastic differential equations and their relevance to dynamical systems.