We are seeking a highly motivated postdoctoral researcher to join our team working at the intersection of particle physics and machine learning. The ideal candidate will have a strong background in physics, particularly in particle physics, and expertise in machine learning techniques. Project Description: The successful candidate will focus on analyzing and advancing uncertainty estimation methods for neural networks, with a particular emphasis on generative models. The research will aim to develop and experiment with novel algorithms that can provide reliable uncertainty estimates for machine learning models used in high-energy physics simulations. The application context for this research will be the study of neutrinos in the KM3NeT telescopes, specifically focusing on atmospheric neutrino oscillations and the potential existence of sterile neutrinos. Currently, Monte Carlo (MC) simulations are widely used for modeling neutrino observations, providing results with guarantees on data quality but at a high computational cost. The project will explore the potential of replacing these MC simulations in some applications with generative models such as diffusion models or other generative approaches, while addressing the critical question of reliability compared to traditional MC simulations.
Offer Description
We are seeking a highly motivated postdoctoral researcher to join our team working at the intersection of particle physics and machine learning. The ideal candidate will have a strong background in physics, particularly in particle physics, and expertise in machine learning techniques.
Project Description: The successful candidate will focus on analyzing and advancing uncertainty estimation methods for neural networks, with a particular emphasis on generative models. The research will aim to develop and experiment with novel algorithms that can provide reliable uncertainty estimates for machine learning models used in high-energy physics simulations.
The application context for this research will be the study of neutrinos in the KM3NeT telescopes, specifically focusing on atmospheric neutrino oscillations and the potential existence of sterile neutrinos. Currently, Monte Carlo (MC) simulations are widely used for modeling neutrino observations, providing results with guarantees on data quality but at a high computational cost. The project will explore the potential of replacing these MC simulations in some applications with generative models such as diffusion models or other generative approaches, while addressing the critical question of reliability compared to traditional MC simulations.
Key Responsibilities:
- Conduct a comprehensive literature review on uncertainty estimation in neural networks, focusing on generative models.
- Propose and experiment with novel algorithms for uncertainty quantification in generative models.
- Apply developed methods to neutrino physics simulations and compare results with traditional Monte Carlo approaches.
- Collaborate with physicists and machine learning experts to ensure the relevance and applicability of the developed methods.
- Present research findings at conferences and prepare manuscripts for publication in peer-reviewed journals.
Qualifications:
- Ph.D. in Physics, Computer Science, or a related field
- Background in Monte Carlo simulation, particle physics and machine learning
- Experience with generative models and uncertainty quantification techniques
- Proficiency in programming languages commonly used in scientific computing and machine learning (e.g., Python, PyTorch, TensorFlow)
- Excellent communication and collaborative skills
- Publication record in relevant fields is highly desirable
This position offers an exciting opportunity to contribute to cutting-edge research at the intersection of fundamental physics and advanced machine learning techniques. The successful candidate will integrate a dynamic interdisciplinary group composed of researchers, AI experts and PhD students from the KM3NeT experimental group at LPC and from GREYC. Access to state-of-the-art computing resources is available locally and at national computing centers.
To apply, please submit your CV, a brief statement of research interests, and contact information for three references to [« Frederic Jurie » <frederic.jurie@unicaen.fr>, « Antonin Vacheret » <vacheret@lpccaen.in2p3.fr>].
References:
[1] Barbetti, M., 2023. Lamarr: LHCb ultra-fast simulation based on machine learning models deployed within Gauss.
[2] Sjostrand T, Mrenna S and Skands P Z 2008 Comput. Phys. Commun. 178 852–867 (Preprint 0710.3820)
[3] Lange D J 2001 Nucl. Instrum. Meth. A 462 152–155
[4] Bellagente M, Haußmann M, Luchmann M and Plehn T 2021 Understanding event-generation networks via uncertainties (Preprint 2104.04543)
Where to apply
frederic.jurie@unicaen.fr
Requirements
Research Field: Computer science » Other
