CIFRE PhD - Uncertainty Quantification in Neural Networks for Critical Applications F/H

Framatome

Oct 2, 2024

Applications are closed

Internship
Full-time
Off-cycle Internship
Data
Research & Development

Requirements

Strong background in applied maths, data science, computer science, machine learning and deep learning, and skills in embedded systems and nuclear physics would be appreciated;
Proficiency in programming languages such as Python or C/C++.;
Good analytical and problem-solving skills, with a strong passion for discovery and cutting-edge research;
Effective communication skills in English, and the ability to work both independently and as part of a multidisciplinary team;
Experience with machine learning applied to physics and the PyTorch framework would be ideal.

Responsibilities

This CIFRE PhD deals with uncertainty quantification in neural network for critical applications.
Neural networks are statistical learning models capable of processing complex data and learning non-linear relationships between inputs and outputs of interest.
However, neural networks, like all learning algorithms, can be subject to uncertainty, particularly in the presence of noisy data, incomplete data or data different from that used to train the algorithm.
Uncertainty can manifest itself in a number of ways, including misclassification, erroneous predictions or low confidence scores.
Uncertainty quantification (UQ) is the process of estimating the uncertainty associated with a measurement or estimate.
It is often used in the fields of statistical learning and data science, where it is important to take into account the uncertainty of data and models.
UQ can therefore be applied to estimate the uncertainty of a neural network.
This uncertainty can then be used to improve the robustness of the network to noisy and incomplete data, or to make more informed decisions.
Available techniques offer different ways of estimating uncertainty in neural networks, and the choice of method depends on the specific application and available resources.
It is important to note that there are ongoing research efforts in this field, and new techniques may emerge in the future.
Multiple approaches exist for estimating the uncertainty of neural network predictions.
However, many questions remain unanswered if these methods are to be used for critical applications such as aeronautics, nuclear, medical, autonomous driving, etc., i.e., all applications that may have a consequent impact on human life regarding their robustness in particular.
Indeed, it is important to assess both the quality of a model's predictions and to ensure that the model's prediction and uncertainty estimate are of high quality, i.e., it is necessary to know whether the prediction and uncertainty can be trusted.
This question is all the more complicated when the algorithm is used with data far removed from the training base.
We speak of distribution shift when the distribution of the new data is (partially) different from that of the training data, or Out Of Distribution when the new data has nothing to do with the training data.
These cases arise when the algorithm is deployed for "real-world" scenarios which may differ from the framework built during the training phase.
Available scientific literature mentions a few avenues of research into uncertainty estimation in the context of non-distributed data (the conformal approach, for example).
However, further research is still needed in the context of out-of-distribution data that is not known in advance, and for which current UQ methods are not very robust.
This PhD is funded by CIFRE, with a joint supervision between Framatome and CentraleSupélec (Centre for Visual Computing laboratory).

FAQs

What is the title of the position for this PhD opportunity?

The title of the position is CIFRE PhD - Uncertainty Quantification in Neural Networks for Critical Applications F/H.

What kind of contract is being offered for this PhD?

The contract being offered is a CIFRE contract.

What is the salary range for this position?

The salary range for this position is between 35,000 to 40,000 euros per year.

Where is the job location?

The job location is in France, specifically in Ile-de-France, Hauts-de-Seine (92), at La Defense.

What educational qualifications are required for this position?

A minimum educational qualification required for this position is a Bac+5 (equivalent to a Master’s degree).

What level of experience is expected from candidates?

The position is suitable for young graduates, hence minimal experience is required.

What kind of research will the PhD focus on?

The PhD will focus on uncertainty quantification in neural networks for critical applications, particularly regarding their robustness and reliability in various domains such as aeronautics, nuclear energy, and medical fields.

What skills are preferred for applicants?

Preferred skills for applicants include a strong background in applied mathematics, data science, computer science, machine learning, deep learning, proficiency in programming languages like Python or C/C++, and experience with machine learning applications in physics, particularly using the PyTorch framework.

Is experience in embedded systems and nuclear physics appreciated?

Yes, skills in embedded systems and nuclear physics would be appreciated for this position.

Will there be collaboration with any educational institutions?

Yes, this PhD is funded by CIFRE, with joint supervision between Framatome and CentraleSupélec (Centre for Visual Computing laboratory).

Are there any language requirements for the job?

Effective communication skills in English are required for this position.

Is the position subject to an administrative investigation?

Yes, the position is subject to an administrative investigation.

Is prior knowledge of uncertainty quantification methods necessary?

While prior knowledge of uncertainty quantification methods is not explicitly stated, a strong background in related areas such as data science and machine learning is essential.

Will the position involve teamwork?

Yes, the role will require the ability to work both independently and as part of a multidisciplinary team.

Framatome

An international leader in nuclear energy recognized for its innovative solutions and value added technologies

Energy

Industry

10,001+

Employees

1958

Founded Year

Mission & Purpose

Framatome is an international leader in nuclear energy recognized for its innovative, digital and value added solutions for the global nuclear fleet. With worldwide expertise and a proven track record for reliability and performance, the company designs, services and installs components, fuel, and instrumentation and control systems for nuclear power-plants. Its more than 16,000 employees work every day to help Framatome’s customers supply ever cleaner, safer and more economical low-carbon energy. plants.