Period, duration
January – July 2024
Profile required
Student in the 4th or 5th year of a Master's degree or engineering school in computer science
Gratuity
4.35€/h (standard)
Location
Annecy-le-vieux ; Maison de la mécatronique ; Symme laboratory
Subject
Context:
MUKROBS is a clinical study in which we record pathological respiratory sounds at different sites in the rib cage.
During the clinical study, we collect acoustic data on different patients for which we try to identify and count mainly 3 types of sounds (crackling, sibilant, ronchis). We are in the process of collecting thousands of raw recordings of different durations, on a disease group and a control group, for a total of 60 subjects.
This internship is offered following a first two-month internship which allowed us to lay the foundations for work, in particular by using the ICBHI database to develop a first AI.
Missions:
• Complete and exploit the state of the art on machine learning models and signal processing techniques that already exist for the detection of adventitious lung sounds.
• If necessary, continue the implementation of a ResNet 50 network, with layer optimization.
• Build traceability of tests and settings carried out to allow statistical analysis of performance.
• Transfer learning: We will then have to move forward with the construction of our own database resulting from the measurements of the clinical study. In the end, we will have to apply this model(s) to our data.
Profile sought:
•Autonomy
•Curiosity
•Dynamism
• Machine Learning
• Signal processing
• Matlab / Python
• PyTorch/TensorFlow
• Knowledge of wavelets or prosody-based sound and speech processing would be a plus.
Contact
laurent.goujon
@univ-smb.fr
Project team
Marine Loubet, SYMME PhD student
Laurent Goujon, lecturer-researcher IUT Annecy / SYMME
Argheesh Bhanot, lecturer-researcher IUT Annecy / LISTIC
Christine Barthod, lecturer-researcher IUT Annecy / SYMME