STAGE: Self-explainable model for audio identification of bird species H/F

Job title:

STAGE: Self-explainable model for audio identification of bird species H/F

Company:

Commissariat à l’Énergie Atomique

Job description

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 FranceReference 2024-32911Description de l’unitéAmong other activities, CEA LIST Software Safety and Security Laboratory (LSL) research teams design and implement automated analysis in order to make software systems more trustworthy, to exhaustively detect their vulnerabilities, to guarantee conformity to their specifications, and to accelerate their certification. The lab recently extended its activities on the topic of AI trustworthiness and gave birth to a new research group: AISER (Artificial Intelligence Safety, Explainability and Robustness).Position descriptionCategoryMathematics, information, scientific, softwareContractInternshipJob titleSTAGE: Self-explainable model for audio identification of bird species H/FSubjectSelf-explainable model for audio identification of bird species H/FContract duration (months)5-6 monthsJob descriptionThrough the recent developments of AI, the use of models produced by machine learning has become widespread, even in industrial settings. However, studies are flourishing showing the dangers that such models can bring, in terms of safety, privacy or even fairness. To mitigate these dangers and improve trust in AI, one possible avenue of research consists in designing methods for generating *explanations* of the model behaviour. Such methods, regrouped under the umbrella term “eXplainable AI” (XAI), empower the user by providing them with relevant information to make an informed choice to trust the model (or not).In particular, rather than attempting to explain the model behaviour after it has been
trained (post-hoc explanations), some XAI methods propose to enforce explainability
constraints directly during the design phase of the machine learning process, resulting
in so-called self-explainable models. In this regard, case-based reasoning is currently
considered a viable alternative to more opaque “black box” convolutional neural networks (CNN), with works such as (Chen et al. 2019) or (Nauta, Bree, and Seifert 2021). In case-based reasoning, new instances of a problem are solved by drawing comparisons with examples encountered before, such that the decision is taken and motivated by the fact that the new instance resembles some known cases (also called prototypes).
CEA-LIST has already developed an open-source library for case-based reasoning net-
works, called CaBRNet(Xu-Darme et al. 2024), and wish to extend its application to new domains and modalities.This internship focuses on the use of machine learning models for the recognition and
classification of bird songs. Audio clips are often encoded in the form of spectrograms, i.e. 2D representations of the intensity of the signal at various frequencies, across a given period of time. Since spectograms can be interpreted as images, a common practice consists in processing them using deep CNNs originally designed for computer vision (Kahl et al. 2021). Hence, the goal of the internship is to extend the case-based reasoning approach to this new task, by adapting existing computer vision methods to learn audio prototypes. In particular, the new approach will take into account the temporal specificities of audio samples. Indeed, contrary to computer vision models which are spacially invariant (the nature of an object remains identical regardless of its position inside the image), spatial location is crucial in spectograms as it corresponds to different frequency ranges and different periods of times.
In practice, the internship will be split in several subtasks as follows:

• Establish a baseline using the reference BirdNET model.
• Identify a body of existing works on self-explainable models for audio classification
• Design and train a case-based reasoning model for audio classifiation, using the CaBRNet framework.

Methods / MeansExplainable AI, Case-based Reasoning, Convolutional Neural NetworksApplicant ProfileAs it is not realistic to be expert in machine-learning, computer vision and XAI, we encourage candidates that do not meet the full qualification requirements to apply nonetheless. We strive to provide an inclusive and enjoyable workplace. We are aware of discriminations based on gender (especially prevalent on our fields), race or disability, we are doing our best to fight them.Minimal qualifications:

• Master student or equivalent (2nd/3rd engineering school year) in computer science
• knowledge of Python and the Pytorch framework
• ability to work in a team, some knowledge of version control

Preferred:

• notions of AI and neural networks
• notions of Computer Vision
• notions of explainable AI

Position locationSiteGrenobleJob locationFrance, Auvergne-Rhône-Alpes, Isère (38)LocationGrenobleCandidate criteriaLanguages

• English (Fluent)
• French (Fluent)

Prepared diplomaBac+5 – Master 2Recommended trainingMaster degree in AI, Machine-Learning or Data SciencePhD opportunityOuiRequesterPosition start date01/02/2025

Expected salary

Location

Isère

Job date

Sun, 15 Sep 2024 05:10:22 GMT

To help us track our recruitment effort, please indicate in your email/cover letter where (vacanciesineu.com) you saw this job posting.