Talk rehearsal and internship presentation

Abstract

FairPipes: Data Mutation Pipelines for Machine Learning Fairness, Paul (preprint)

Machine Learning (ML) models are ubiquitous in decision making applications impacting citizens’ lives: credit attribution, crime recidivism, etc. In addition to seeking high performance and generalization abilities, insuring that ML models do not discriminate against citizens regarding their age, gender, or race is essential. To this end, researchers developed various fairness assessment techniques, comprising fairness metrics and mitigation approaches, notably at the model level. However, the sensitivity of ML models to fairness data perturbations has been less explored. This paper presents mutation-based pipelines to emulate fairness variations in the data once the model is deployed. FairPipes propose a first set of mutation operators that can be further extended. We evaluated FairPipes on seven ML models over three datasets. Our results highlight different fairness sensitivity behaviors across models, from the most sensitive perceptrons to the insensitive support vector machines.

From Moldova to Inria in 5 Easy Steps, Sergiu [slides]

My name is Sergiu MOCANU. I am a second year student at Istic’s Software Engineering Master. Last year I did an internship at Inria (team: LACODAM, mentor: Alexandre TERMIER) on the subjects of “Program Synthesis Using ChatGPT” – a rather short experience (that spanned over only 2 and a half months) that gave me a taste of what is research in the field of IT. As a natural continuation, for this year’s internship I chose the subject “Program Synthesis Using Open LLMs”, having as supervisors Olivier BARAIS, Mathieu ACHER and Herman KOUADIO. My interests in the field of programming lean more towards the abstract world (e.g., Logic & Calculability). Outside programming, my interests span over numerous and various subjects – ranging from ‘Music’ and ‘Human Psychology’ to ‘Astronomy’ and ‘Quantum Physics’.