Data Science : Non-linear and fuzzy clustering concerning medical data

This internship is done in collaboration with the laboratory Analgesia.
Analgesia develops a mobile e-health application (e-DOL) in order to improve the evaluation of pain and its repercussions, and to support patients with chronic pain on a daily basis. The data collected by the application are in the form of a survey repeated every week. They should allow to know if there are groups of patients who have the same lengthwise pattern and if there is an explanation for these similarities.
The goal of the internship is therefore to apply unsupervised classification methods to the data and to analyze the results obtained using explanatory variables.
LIMOS - UMR CNRS 6158 - Aubiere

Light recommendation system for autonomous factories

Project done in collaboration with Braincube.
In the era of information development, the internet and connected devices, factories must also follow the trend and use their data. They must now be able to retrieve data from their assembly lines and use it to their advantage. Thus, these smart factories use technologies such as artificial intelligence and the Internet of Things to operate without interruption and with minimal human intervention. The potential benefits of these factories include increased productivity and reduced labor costs.
The goal of this recommendation system would be to propose a simple use of data from a production chain in order to create a machine learning model. This model would propose settings values as well as the most accurate evaluation of the finished product for a given context. This historical data processing pipeline would be a valuable tool for autonomous factories. It would reduce production costs, defect rates, energy consumption,... while optimizing production quality.
At the end of this project, we can propose several models of recommendation systems that take into account the historical data and use different calculation methods.
The whole project is written in Python with the help of Data Science libraries such as Scikit-Learn.

June 2022 - Video game based on Reinforcement Learning

The goal of this 1st year project was to create a game based on reinforcement learning in C with the SDL2 library.
As a team we have decided to do several mazes where an agent has to learn how to escape them with different traps on its path. The main algorithm used is Q-learning.
More information about this project here (only in french).

Other stuff that I made