‹ Back to the directory

Corinne Ancourt Le Quellenec

Corinne Ancourt Le Quellenec

Research Director

Center · CRI

Discipline(s)
Computer Science
Topic(s)
Computer Science

Biography

Corinne Ancourt-Le Quellene is a researcher specializing in code optimization and parallel compilation, with a strong expertise in polyhedral techniques and the application of machine learning to performance optimization. Her work focuses notably on improving loop transformations, such as *tiling*, to exploit data locality and parallelism by integrating advanced methods like *machine learning* to refine the parameterization of transformations. She has also contributed to the development of innovative solutions for detecting nested objects, particularly in the field of beekeeping, by combining neural architectures like Mask R-CNN with multi-dataset approaches. Her research extends to the design of configurable tools for polyhedral optimization, such as PolyTOPS, adaptable to heterogeneous architectures like NPUs, as well as the automation of benchmark generation for evaluating optimization techniques.

Publication(s)

PhD supervision

  • 2025 IMPROVEMENT OF DATA TRANSFERS AND PERFORMANCE ON HETEROGENEOUS ARCHITECTURES SAULEAU Luc
  • 2025 Large language models for sequential decision-making ATTIA EL HILI Youssef
  • 2024 Optimization of the placement of convolutional neural networks on hybrid architectures under resource constraints VIENS Arthur
  • 2023 Static cost models of compilers for optimizing the energy performance of embedded software DESTAL Benjamin
  • 2022 Intelligent agents for offline reinforcement learning with reduced interactions GAIZI Othman
  • 2021 Configurable Polyhedral Scheduling for All-Scenario Deep Learning Compilers CONSOLARO Gianpietro
  • 2019 Nested object detection in dense scenes using Deep Learning - Application to bee and varroa mite detection KRIOUILE Yassine
  • 2018 The automation of source-to-source program optimizations using Machine Learning techniques BEREZOV Maksim
  • 2015 Performance analyses and code transformations for MATLAB applications KIEPAS Patryk