Photo de profil

Henri Hoyez

Generative AI Engineer - PhD Student

đź“§ Contact me!
📍 Yutz, France
đź”— LinkedIn

About me

Do you want an effective implementation of innovative AI solutions that truly make an impact?

I am a passionate Machine Learning Engineer and PhD Researcher specializing in Generative Deep Learning, uniquely positioned at the intersection of industry and academia. This dual expertise enables me to transform cutting-edge research into real-world applications, delivering both innovation and measurable results.


Currently pursuing my PhD in collaboration with Paul Wurth Luxembourg and RPTU Kaiserslautern, I focus on designing novel deep learning architectures for domain mapping in multivariate time series. My work directly fuels industrial process optimization, with a particular emphasis on metallurgical operations.


By combining rigorous scientific methodology with hands-on engineering experience, I develop AI solutions that are not only innovative, but also practical, scalable, and impactful. My mission is simple: bridge the gap between advanced AI research and industrial success.

Vous souhaitez une mise en œuvre efficace de solutions IA innovantes qui ont un réel impact ?

Je suis ingénieur Machine Learning et doctorant spécialisé en Deep Learning génératif, à l'interface de l'industrie et de la recherche académique. Cette double expertise me permet de transformer la recherche de pointe en applications concrètes.


En collaboration avec Paul Wurth Luxembourg et la RPTU Kaiserslautern, je conçois de nouvelles architectures de deep learning pour le domain mapping de séries temporelles multivariées, avec un focus sur l'optimisation des procédés métallurgiques.


En combinant rigueur scientifique et expérience terrain, je développe des solutions IA innovantes, pratiques et évolutives, au service de la réussite industrielle.

Education

PhD in Generative Deep Learning

2021 — Present

Paul Wurth Luxembourg & RPTU Kaiserslautern

Thesis topic: Deep Domain Mapping for Multivariate Time Series.

Thesis directors: Dr. ir. Cédric Schockaert & Prof. Dr. Didier Stricker.

Engineering Degree in Artificial Intelligence

2015 — 2020

Higher Institute of Electronics and Digital Science (ISEN Lille)

Bachelor of Science in Computer Science

2015 — 2018

Higher Institute of Electronics and Digital Science (ISEN Lille)

Doctorat en Deep Learning génératif

2021 — Aujourd'hui

Paul Wurth Luxembourg & RPTU Kaiserslautern

Sujet de thèse : Deep Domain Mapping pour séries temporelles multivariées.

Directeurs de thèse : Dr. ir. Cédric Schockaert & Prof. Dr. Didier Stricker.

Diplôme d'ingénieur en Intelligence Artificielle

2015 — 2020

Institut Supérieur de l'Électronique et du Numérique (ISEN Lille)

Licence en informatique

2015 — 2018

Institut Supérieur de l'Électronique et du Numérique (ISEN Lille)

Experiences

Skills

Techniques

Python LaTeX Git TensorFlow MLflow Linux Docker System administration

Research

Methodology Scientific writing Presentation Supervised learning Unsupervised learning Semi-supervised learning Representation learning

Languages

French (native) English (B2)

Techniques

Python LaTeX Git TensorFlow MLflow Linux Docker Administration système

Recherche

Méthodologie Rédaction scientifique Présentation Apprentissage supervisé Apprentissage non supervisé Apprentissage semi-supervisé Apprentissage de représentations

Langues

Français (langue maternelle) Anglais (B2)

Papers

  • "MISTI: Multi-Style Transfer for Multivariate Time Series" (Accepted, EUSIPCO 2025)
  • "Which Time Series Domain Shifts can Neural Networks Adapt to?" (Accepted, EUSIPCO 2024)
  • "Unsupervised Image-to-Image Translation: A Review" (Accepted, MDPI)
  • "JuCify: a step towards Android code unification for enhanced static analysis" (Accepted, ICSE 2022)
  • "Mts-cyclegan: An adversarial-based deep mapping learning network for multivariate time series domain adaptation applied to the ironmaking industry" (ArXiv)
  • « MISTI: Multi-Style Transfer for Multivariate Time Series » (AcceptĂ©, EUSIPCO 2025)
  • « Which Time Series Domain Shifts can Neural Networks Adapt to? » (AcceptĂ©, EUSIPCO 2024)
  • « Unsupervised Image-to-Image Translation: A Review » (AcceptĂ©, MDPI)
  • « JuCify: a step towards Android code unification for enhanced static analysis » (AcceptĂ©, ICSE 2022)
  • « Mts-cyclegan: An adversarial-based deep mapping learning network for multivariate time series domain adaptation applied to the ironmaking industry » (ArXiv)