Computer vision, machine/deep learning
Aix-Marseille Université – Laboratoire d’Informatique et Systèmes (LIS)
Campus de Saint Jérôme, 52 Av. Escadrille Normandie Niemen, 13013 Marseille
This thesis aims at modelling behaviors in five autism spectrum disorder (ASD) mouse models, caused by genetic mutations or by excessive gestational weight gain (GWG, detected today in half of the women), or a combination of both. We will join the complementary expertise of three teams from three laboratories to investigate lifelong the « spontaneous behavior » of mice freely moving into an open space, by using a recently developed system based on novel and evolutive tools (Live Mouse Tracker – LMT https://livemousetracker.org, DeepLabCut – DLC http://www.mackenziemathislab.org/deeplabcut).
Behaviors of all living beings consists of patterns in time: identifying this set of patterns available to an animal is key to making quantitative descriptions of behavior. We thus propose a 3D behavioral analysis pipeline dedicated to the study of temporal series of single actions (including individual, dyadic or group actions). This framework will rely on computer vision (tracking, pose recognition, ….) and recent deep learning techniques (transformers, auto-encoders, …), and will consider these action time series at multiple time-scales.
The main goal of this thesis is then to develop algorithms to Identify novel behavioral motifs and sequences in mouse models of ASD compared with controls. The thesis will we split into two complementary parts:
Part 1: modeling complex locomotor activity applied to five mouse models of ASD. We will develop an algorithm to find sequences of behaviors extracted from the analysis of locomotor activity (trajectories points and behaviors extracted by LMT), consisting in:
Part 2: Identify novel behavioral motifs and sequences in mouse models of ASD compared with controls.
Salary: 3 years contract, 39.000 euros/year
QUALIFICATIONS/SKILLS/EDUCATION & RESEARCH REQUIREMENTS
We are looking for a highly motivated candidate with a strong background in both mathematics and computer science. He/She should have recently completed a master’s degree or should be about to complete it. The candidate should demonstrate good skills in computer vison (tracking, video sequence analysis) and in machine/deep learning (practice in transformers, auto-encoder is needed). We also expect the candidate to be open-minded, curious and autonomous
01/12/2022 – 0h00 (Paris Time)
REQUESTED DOCUMENTS OF APPLICATION
- Curriculum Vitæ
- Cover letter
- Grades and transcripts after High school, including rankings