Large-Scale Wasserstein Gradient Flows
Dec 6, 2021
Speakers
Petr Mokrov
Speaker · 0 followers
Alexander Korotin
Speaker · 0 followers
Lingxiao Li
Speaker · 0 followers
Aude Genevay
Speaker · 0 followers
Justin M. Solomon
Speaker · 0 followers
Evgeny Burnaev
Speaker · 0 followers
About
Wasserstein gradient flows provide a powerful means of understanding and solving many diffusion equations. Specifically, Fokker-Planck equations, which model the diffusion of probability measures, can be understood as gradient descent over entropy functionals in Wasserstein space. This equivalence, introduced by Jordan, Kinderlehrer and Otto, inspired the so-called JKO scheme to approximate these diffusion processes via an implicit discretization of the gradient flow in Wasserstein space. Solving the optimization problem associated with each JKO step, however, presents serious computational challenges. We introduce a scalable method to approximate Wasserstein gradient flows, targeted to machine learning applications. Our approach relies on input-convex neural networks (ICNNs) to discretize the JKO steps, which can be optimized by stochastic gradient descent. Contrarily to previous work, our method does not require domain discretization or particle simulation. As a result, we can sample from the measure at each time step of the diffusion and compute its probability density. We demonstrate the performance of our algorithm by computing diffusions following the Fokker-Planck equation and apply it to unnormalized density sampling as well as nonlinear filtering.Wasserstein gradient flows provide a powerful means of understanding and solving many diffusion equations. Specifically, Fokker-Planck equations, which model the diffusion of probability measures, can be understood as gradient descent over entropy functionals in Wasserstein space. This equivalence, introduced by Jordan, Kinderlehrer and Otto, inspired the so-called JKO scheme to approximate these diffusion processes via an implicit discretization of the gradient flow in Wasserstein space. Solvin…
Organizer
NeurIPS 2021
Account · 1.9k followers
About NeurIPS 2021
Neural Information Processing Systems (NeurIPS) is a multi-track machine learning and computational neuroscience conference that includes invited talks, demonstrations, symposia and oral and poster presentations of refereed papers. Following the conference, there are workshops which provide a less formal setting.
Like the format? Trust SlidesLive to capture your next event!
Professional recording and live streaming, delivered globally.
Sharing
Recommended Videos
Presentations on similar topic, category or speaker
Sample Complexity of Tree Search Configuration: Cutting Planes and Beyond
Watch later
Nina Balcan, …
Total of 0 viewers voted for saving the presentation to eternal vault which is 0.0%
Coarse-to-fine Animal Pose and Shape Estimation
Watch later
Chen Li, …
Total of 0 viewers voted for saving the presentation to eternal vault which is 0.0%
Damage Estimation and Localization from Sparse Aerial Imagery
Watch later
Total of 0 viewers voted for saving the presentation to eternal vault which is 0.0%
Physics-Aware Downsampling with Deep Learning for Scalable Flood Modeling
Watch later
Niv Giladi, …
Total of 0 viewers voted for saving the presentation to eternal vault which is 0.0%
Zero Time Waste: Recycling Predictions in Early Exit Neural Networks
Watch later
Total of 0 viewers voted for saving the presentation to eternal vault which is 0.0%
Self-Supervised Learning with Kernel Dependence Maximization
Watch later
Yazhe Li, …
Total of 0 viewers voted for saving the presentation to eternal vault which is 0.0%