General information
Organisation
The French Alternative Energies and Atomic Energy Commission (CEA) is a key player in research, development and innovation in four main areas :
• defence and security,
• nuclear energy (fission and fusion),
• technological research for industry,
• fundamental research in the physical sciences and life sciences.
Drawing on its widely acknowledged expertise, and thanks to its 16000 technicians, engineers, researchers and staff, the CEA actively participates in collaborative projects with a large number of academic and industrial partners.
The CEA is established in ten centers spread throughout France
Reference
2026-42001
Description de l'unité
The Laboratory of Memory and Computing Devices (LDMC) at CEA-Leti develops emerging memory technologies and explores their use for data storage and computing. Its research combines device design, fabrication and experimental characterization with the investigation of new hardware architectures, particularly for artificial intelligence and in-memory computing.
By joining us, you will contribute to the development of energy-efficient hardware solutions, working with multidisciplinary research teams and industrial partners.
Position description
Category
Micro and nano technologies
Contract
Internship
Job title
M2 Internship – Perturbation-based Bayesian learning with resistive memories
Subject
Bayesian neural networks assess the uncertainty of their predictions by exploring multiple sets of weights, the parameters that determine their output. However, training these networks requires costly computations. Resistive memories (ReRAM) offer a way to simplify them: they can store these weights, perform computations directly in memory and provide random variations.
This internship aims to exploit these variations for two purposes: estimating how to adjust the weights to improve predictions, and exploring different sets of weights to represent uncertainty. The goal is to reduce the computational demands of learning by exploiting the physical properties of memory devices. You will develop and evaluate this approach through simulations, progressively incorporating realistic ReRAM device characteristics.
Contract duration (months)
5 to 6 months
Job description
Within the Laboratory of Memory and Computing Devices at CEA-Leti in Grenoble, you will contribute to the development of Bayesian learning methods suited to resistive memories (ReRAM). You will build on the team’s previous work, which has already demonstrated the use of these devices to train Bayesian neural networks.
Existing approaches still require costly numerical operations or become less efficient as models grow. This internship aims to explore learning strategies that exploit the physical properties of memory devices to simplify these computations while preserving prediction quality and uncertainty estimation.
The work will initially focus on algorithm development and simulation. Starting from reference methods, you will evaluate different strategies and progressively incorporate realistic ReRAM device characteristics. You will assess their performance and the constraints associated with hardware implementation.
Depending on progress, the approach may then be evaluated using experimental data or existing memory arrays.
Applicant Profile
You are a final-year Master’s or engineering student specializing in electrical engineering, microelectronics, applied physics or computer science.
You are comfortable with programming and applied mathematics, and interested in neural networks and their hardware implementation. Some experience in machine learning would be an asset. Prior knowledge of emerging memory technologies is not required.
Position location
Site
Grenoble
Job location
France, Auvergne-Rhône-Alpes, Isère (38)
Location
Grenoble
Candidate criteria
Languages
English (Intermediate)
Prepared diploma
Bac+5 - Diplôme École d'ingénieurs
PhD opportunity
Oui
Requester
Position start date
01/02/2027
Create candidate area
