Benchmark of supervised machine learning model interpretation methods - Application to catalytic hydrotreatment

IFP Energies nouvelles (IFPEN) est un acteur majeur de la recherche et de la formation dans les domaines de l’énergie, du transport et de l’environnement. De la recherche à l’industrie, l’innovation technologique est au cœur de son action, articulée autour de quatre priorités stratégiques : Mobilité Durable, Energies Nouvelles, Climat / Environnement / Economie circulaire et Hydrocarbures Responsables.

Dans le cadre de la mission d’intérêt général confiée par les pouvoirs publics, IFPEN concentre ses efforts sur :

• l’apport de solutions aux défis sociétaux de l’énergie et du climat, en favorisant la transition vers une mobilité durable et l’émergence d’un mix énergétique plus diversifié ;
• la création de richesse et d’emplois, en soutenant l’activité économique française et européenne et la compétitivité des filières industrielles associées.

Partie intégrante d’IFPEN, l’école d’ingénieurs IFP School prépare les générations futures à relever ces défis.

Benchmark of supervised machine learning model interpretation methods - Application to catalytic hydrotreatment

The use of Machine Learning models (neural networks, SVM, random forest, …) is growing in the field of chemical engineering. One of the main drawbacks of these data-driven models in their lack of insight and interpretability (“black box” models). To answer this problem, a lot of methods exist or are emerging (Gini and permutation feature importance, partial dependance plot, SHAPE, LIME,…).

The aim of this internship is, using controlled and artificial data, to assess the efficiency and the robustness of the most common methods for interpretation of black-box models.

Description

The internship will focus on modeling the hydrotreating process, which involves purifying a complex feedstock derived from biomass, refining or recycling plastic waste. This process generally targets the removal of sulfur, nitrogen or oxygen compounds.

The internship will include the following tasks:

• Creation of an artificial dataset with a physicochemical model. Allow to know exactly the causal links inside the data and then to assess the accuracy of the interpretation methods for Machine learning models.
• Review of the main methods for interpretation of supervised Machine Learning (SML) models
• Training of several SML models on the artificial dataset and assessment of the selected model interpretation methods.
• Study of the sensitivity/robustness of the various methods: with the number and nature of data, the type of SML model, etc.

Required profile

• A student in the field of Machine learning, data science, applied mathematics with a strong interest in applied sciences.
  Or
• Chemical or process Engineering student with a strong interest (and notions) in the fields of Machine Learning and data science.

Language: French or English

Additional information

Duration of the internship: 6 months
Workplace: IFPEN Lyon, Rond-point de l'échangeur de Solaize, 69360 Solaize
Transport: public transportation / personal vehicle
Paid internship 1080€/month