<div dir="ltr"><h2 class="gmail-part" id="gmail-deep-statistical-manifolds">Deep Statistical Manifolds</h2><h3 class="gmail-part" id="gmail-project"><a class="gmail-anchor gmail-hidden-xs" href="https://notes.inria.fr/t0bx4VaNSHiyMPPa9P2OZw#project" title="project"><i class="gmail-fa gmail-fa-link"></i></a>Project</h3><p class="gmail-part">The
internship aims to explore the usefulness of the Fisher-Ráo [1] metric
combined with deep probabilistic models [2]. The main question is
whether or not this metric has some relationship with the training of
deep generative models. In plain, we would like to understand if the
training and/or fine-tuning of such probabilistic models follow optimal
paths on the manifold of probability distributions [3].</p><h3 class="gmail-part" id="gmail-task"><a class="gmail-anchor gmail-hidden-xs" href="https://notes.inria.fr/t0bx4VaNSHiyMPPa9P2OZw#task" title="task"><i class="gmail-fa gmail-fa-link"></i></a>Task</h3><p class="gmail-part">Your
task will be to design and implement an experimental framework allowing
to measure what kind of paths are followed on the manifold of
probability distributions when such deep probabilistic models are
trained. To that aim, one must first be able to measure distances in
this manifold, and here is where the Fisher-Ráo metric comes in the
game. The candidate does not need to be familiar with the specific
concepts of Fisher-Ráo metric, but needs to be open to learning new
mathematical concepts. The implementation of these experiments will
require knowledge in Python and in PyTorch.</p><h3 class="gmail-part" id="gmail-team"><a class="gmail-anchor gmail-hidden-xs" href="https://notes.inria.fr/t0bx4VaNSHiyMPPa9P2OZw#team" title="team"><i class="gmail-fa gmail-fa-link"></i></a>Team</h3><p class="gmail-part">You
will join the RobotLearn team [4], an international team of
researchers, students, and engineers at Inria Grenoble [5]. The team has
a strong background in machine learning for audio-visual computation
and its application to robotics, and in particular with deep generative
models. The team is headed by Xavier Alameda-Pineda [6], who will be
your supervisor, together with Xiaoyu Lin (PhD student) [7].</p><h3 class="gmail-part" id="gmail-requirements"><a class="gmail-anchor gmail-hidden-xs" href="https://notes.inria.fr/t0bx4VaNSHiyMPPa9P2OZw#requirements" title="requirements"><i class="gmail-fa gmail-fa-link"></i></a>Requirements</h3><p class="gmail-part">Our
main requirements are 1) motivation, 2) general knowledge of Machine
Learning and Mathematics, and 3) knowledge of Python programming.
Knowledge of Riemannian geometry or differential geometry in general is a
plus but it is NOT mandatory.</p><h3 class="gmail-part" id="gmail-conditions"><a class="gmail-anchor gmail-hidden-xs" href="https://notes.inria.fr/t0bx4VaNSHiyMPPa9P2OZw#conditions" title="conditions"><i class="gmail-fa gmail-fa-link"></i></a>Conditions</h3><p class="gmail-part">The
internship should start in the second half of 2023. It has a duration
of 5 to 6 months. There will be a compensation of 500 - 600 Euro per
month. Additionally, you will receive subsidized lunch meals (one lunch
costs 2 - 4 Euro). You will have a dedicated working space at Inria with
a workstation that has a GPU. Moreover, you will have access to one CPU
and two GPU clusters to run experiments.</p><h3 class="gmail-part" id="gmail-application"><a class="gmail-anchor gmail-hidden-xs" href="https://notes.inria.fr/t0bx4VaNSHiyMPPa9P2OZw#application" title="application"><i class="gmail-fa gmail-fa-link"></i></a>Application</h3><p class="gmail-part">Please send an e-mail to <a href="mailto:xavier.alameda-pineda@inria.fr" target="_blank" rel="noopener">xavier.alameda-pineda@inria.fr</a> including a paragraph about your motivation, your CV, and a recent transcript of your grades.</p><h3 class="gmail-part" id="gmail-references"><a class="gmail-anchor gmail-hidden-xs" href="https://notes.inria.fr/t0bx4VaNSHiyMPPa9P2OZw#references" title="references"><i class="gmail-fa gmail-fa-link"></i></a>References</h3><p class="gmail-part">[1] Fisher-Ráo metric in Wikipedia: <a href="https://en.wikipedia.org/wiki/Fisher_information_metric" target="_blank" rel="noopener">https://en.wikipedia.org/wiki/Fisher_information_metric</a><br>
[2] Girin, Laurent, et al. “Dynamical variational autoencoders: A
comprehensive review.” Foundations and Trends in Machine Learning, 2021.<br>
[3] Statistical manifolds in Wikipedia: <a href="https://en.wikipedia.org/wiki/Statistical_manifold" target="_blank" rel="noopener">https://en.wikipedia.org/wiki/Statistical_manifold</a><br>
[4] <a href="https://team.inria.fr/robotlearn/" target="_blank" rel="noopener">https://team.inria.fr/robotlearn/</a><br>
[5] <a href="https://www.inria.fr/fr/centre-inria-grenoble-rhone-alpes" target="_blank" rel="noopener">https://www.inria.fr/fr/centre-inria-grenoble-rhone-alpes</a><br>
[6] <a href="http://xavirema.eu/" target="_blank" rel="noopener">http://xavirema.eu/</a><br>
[7] <a href="https://team.inria.fr/robotlearn/team-members/xiaoyu-lin/" target="_blank" rel="noopener">https://team.inria.fr/robotlearn/team-members/xiaoyu-lin/</a></p></div>