Postdoctoral position on Optimal Transport and Machine Learning at University of Rouen, Rouen Normandy

The Machine Learning Team of LITIS (Rouen) is seeking a postdoctoral researcher for OATMIL a French
Research Agency funded project. OATMIL is a 4-year funded project which duration
goes up to 2021 aiming at brigding the gap between Optimal Transport and Machine Learning.
Scientific context
---
Wasserstein distance is a powerful tool based on the theory of optimal transport to compare
data distributions with wide applications in image processing, computer vision and machine learning [1]. In a context of machine learning, it has recently found numerous applications, e.g. domain adaptation [2,10], or word embedding [3]. In the context of deep learning, the Wasserstein distance appeared recently to be a powerful loss in generative models [4] and in multi-label classification [5]. Its power comes from two major reasons: i) it allows to operate on empirical data distributions in a non-parametric way ii) the geometry of the underlying space can be leveraged to compare
the distributions in a geometrically sound way. Yet, the deployment of Wasserstein distances in a wider class of applications is somehow limited, especially because of an heavy computational burden. Recent strategies implying entropic regularization [6] also fail to consider large scale datasets. Remarkably, the problem is amenable to stochastic programming thanks to its dual (and potentially regularized) formulation [7, 8]. Those recent advances pave the way for a large number of applications in learning with with deep networks, as soon as the Wasserstein distance serves as loss function. Very recently, the Wasserstein distance has also been considered for
learning unsupervised alignment of data points in high-dimensional space [9].
Objective
---
The objectives for the postdoc is to develop novel methods for learning
geometric transformation for alignment problems using OT.
- learning parametric transformation
- optimal alignment for dataset living in different spaces
- learning transformation for multimodal data (image+text, image+audio, ...)
Both theoretical and practical aspects of these problems will be developed
Other possible topics depending on candidate profiles are
- efficient/stochastic algorithms for regularized wasserstein problems
- matrix factorization problems using Wasserstein distances
Desired profile
---
strong background in mathematics, machine learning, statistics and algorithms,
Python.
Informations
---
Application deadline: November 30, 2018
Starting date: February 1, 2019 or later
Duration: 1 year (potentially renewable 1 year on other budget)
Location: Université Rouen Normandie, Saint Etienne du Rouvray, Rouen France
Salary range : the post-doc funded by the ANR (French National Agency for Research).
Salary is according to the French national scale ( 2200 - 2,700 gross per month depending on experience).
contact: Alain Rakotomamonjy
How-to
---
Please send applications via email, including:
* a complete CV including a list of publications
* the name and email address of three references
Applications should be sent to Alain Rakotomamonjy alain.rakotomamonjy-AT-univ-rouen.fr . Applicants can be asked to do a short assignment in order to demonstrate their research abilities.
References
---
[1] G. Peyré and M. Cuturi, Computational Optimal Transport.
2018. [Online]. Available: https://optimaltransport.github.io
To be published in Foundations and Trends in Computer Science,
[2] N. Courty, R. Flamary, D. Tuia, and A. Rakotomamonjy, “Optimal transport for domain adaptation,” IEEE Transactions on
Pattern Analysis and Machine Intelligence, 2017.
[3] G. Huang, C. Guo, M. Kusner, Y. Sun, F. Sha, and K. Weinberger, “Supervised word mover’s distance,” in Advances in Neural
Information Processing Systems, 2016, pp. 4862–4870.
[4] M. Arjovsky, S. Chintala, and L. Bottou, “Wasserstein generative adversarial networks,” in Proceedings of the 34th International
Conference on Machine Learning, vol. 70, Sydney, Australia, 06–11 Aug 2017, pp. 214–223.
[5] C. Frogner, C. Zhang, H. Mobahi, M. Araya, and T. Poggio, “Learning with a Wasserstein loss,” in NIPS, 2015.
[6] M. Cuturi, “Sinkhorn distances: Lightspeed computation of optimal transportation,” in Advances on Neural Information Processing
Systems (NIPS), 2013, pp. 2292–2300.
[7] A. Genevay, M. Cuturi, G. Peyré, and F. Bach, “Stochastic optimization for large-scale optimal transport,” in Advances in Neural
Information Processing Systems, 2016, pp. 3432–3440.
[8] V. Seguy, B. Bhushan Damodaran, R. Flamary, N. Courty, A. Rolet, and M. Blondel, “Large-scale optimal transport and mapping
estimation,” in International Conference on Learning Representations (ICLR), 2018.
[9] E. Grave, A. Joulin, P. Berthet, Unsupervised Alignment of Embeddings with
Wasserstein Procrustes, https://arxiv.org/pdf/1805.11222.pdf
[10] N. Courty, R. Flamary, A. Habrard, A. Rakotomamonjy, "Joint distribution optimal transportation for domain adaptation", NIPS 2017