Supervision : Patrick GALLINARI

Co-supervision : LAMPRIER Sylvain

Representation Learning and Deep Generative Modeling in Dynamical Systems

The recent rise of deep learning has been motivated by numerous scientific breakthroughs, particularly regarding representation learning and generative modeling. However, most of these achievements have been obtained on image or text data, whose evolution through time remains challenging for existing methods. Given their importance for autonomous systems to adapt in a constantly evolving environment, these challenges have been actively investigated in a growing body of work. In this thesis, we follow this line of work and study several aspects of temporality and dynamical systems in deep unsupervised representation learning and generative modeling. Firstly, we present a general-purpose deep unsupervised representation learning method for time series tackling scalability and adaptivity issues arising in practical applications. We then further study in a second part representation learning for sequences by focusing on structured and stochastic spatiotemporal data: videos and physical phenomena. We show in this context that performant temporal generative prediction models help to uncover meaningful and disentangled representations, and conversely. We highlight to this end the crucial role of differential equations in the modeling and embedding of these natural sequences within sequential generative models. Finally, we more broadly analyze in a third part a popular class of generative models, generative adversarial networks, under the scope of dynamical systems. We study the evolution of the involved neural networks with respect to their training time by describing it with a differential equation, allowing us to gain a novel understanding of this generative model.

2019-2022 Publications

• 2022
  • J.‑Y. Franceschi : “Apprentissage de représentations et modèles génératifs profonds dans les systèmes dynamiques”, thesis, phd defence 02/14/2022, supervision Gallinari, Patrick, co-supervision : Lamprier, Sylvain (2022)
• 2021
  • J. Donà, J.‑Y. Franceschi, S. Lamprier, P. Gallinari : “PDE-Driven Spatiotemporal Disentanglement”, The Ninth International Conference on Learning Representations, Vienne, Austria (2021)
  • J. Donà, J.‑Y. Franceschi, S. Lamprier, P. Gallinari : “PDE-Driven Spatiotemporal Disentanglement”, The Ninth International Conference on Learning Representations, Vienne (virtual), Austria (2021)
• 2020
  • J.‑Y. Franceschi, E. Delasalles, M. Chen, S. Lamprier, P. Gallinari : “Stochastic Latent Residual Video Prediction”, Proceedings of the 37^th International Conference on Machine Learning, vol. 119, Proceedings of Machine Learning Research, Vienne, Austria, pp. 3233-3246, (PMLR) (2020)
• 2019
  • J.‑Y. Franceschi, A. Dieuleveut, M. Jaggi : “Unsupervised Scalable Representation Learning for Multivariate Time Series”, Thirty-third Conference on Neural Information Processing Systems, vol. 32, Advances in Neural Information Processing Systems, Vancouver, Canada, (Curran Associates, Inc.) (2019)
  • J.‑Y. Franceschi, A. Dieuleveut, M. Jaggi : “Unsupervised Scalable Representation Learning for Multivariate Time Series”, Thirty-third Conference on Neural Information Processing Systems, vol. 32, Advances in Neural Information Processing Systems, Vancouver, Canada, pp. 4650-4661, (Curran Associates, Inc.) (2019)