Address : François MALGOUYRES L'Institut de Mathématiques de Toulouse (IMT, CNRS UMR 5219) Université Paul Sabatier 118 route de Narbonne F-31062 Toulouse Cedex 9 France Building: 1R1; Office :    102 Équipe : Statistique et Optimisation Équipe projet : Apprentissage, Optimisation, Complexité E-mail : FirstName.LastName@math.univ-toulouse.fr (replace FirstName.LastName by Francois.Malgouyres)

A selection of recent talks and a few news :

• Nov. 2025: The project MadLearning is accepted by the PEPR IA and will be funded by . It starts in 2026 and ends in 2030.
• March 2025: Video of a talk, by Armand Foucault, introducing our article on HadamRNN for ICLR 2025 (there is a compatibility problem with firefox)
• July 2024: Video of a talk, by Mimoun Mohamed, introducing our article on SEA for ICML 2024
• 7-11/10/2024: 5-hour lecture on the "Properties of the landscape in Neural Network Optimization", within the thematic trimester Beyond classical regimes in statistical inference and machine learning.
• 04/04/2024 : Video of a talk by Joachim Bona-Pellissier at the Math Machine learning seminar MPI MIS + UCLA: Geometry-induced implicit regularization in deep ReLU neural networks.
• 10/11/2022 : Video of a talk by El-Mehdi Achour at the Math Machine learning seminar MPI MIS + UCLA : The loss landscape of deep linear neural networks a second order analysis.
• 07/01/2021: video 1 et video 2. Journée DeepEx de la SfDS.
• 04/11/2020: video grand public. Evènement: "L'intelligence artificielle: comment ça marche", organisé par le Club Audiovisuel Numérique de Toulouse Métropole (CANtm).
• 17/07/2020: The videos of MSML videos of MSML are available (my talk was on Monday).
• 17/10/2019: Co-organizer of a GDR ISIS workshop "Theory of Deep Learning", CNAM, Paris.
• 16/07/2019: talk at the conference ICIAM 2019: "Joint landscape/expressiveness guarantee in deep learning"
• 2019-2023: I will be a participant to an ANITI chair, with Edouard Pauwels and Jérôme Bolte (PI).
• 01/2019: Course at IHP, for the students of the Master MVA. We have 80 students from the best parisian schools. See the page of the course : Fondements théoriques de l'apprentissage profond
• 30/11/2018: talk at the workshop Mathématiques du deep learning
• 04-06/09/2018: talk at the conference "Deep-learning: From theory to applications" (183 participants) Video of the talks are available
• 03/05/2018: talk at the workshop "Parcimonie et Applications" organise by the GDR MIA, at Bordeaux.
• 21-24/04/2018: talk at the conference "International workshop on image processing and inverse problems" at the Beijing Computational science Research center.
• 07/12/2017: talk at the workshop "Machine learning in for acquisition systems", within the Variational methods and effective algorithms for imaging and vision programme , at the Isaac Newton Institute, Cambridge.
• 23-27/10/2017: one week lecture Mathematical methods for image processing, at the National Institute of technology Karataka, India.