VideoFlow: A Flow-Based Generative Model for Video

Abstract: Generative models that can model and predict sequences of future events can, in principle, learn to capture complex real-world phenomena, such as physical interactions. In particular, learning predictive models of videos offers an especially appealing mechanism to enable a rich understanding of the physical world: videos of real-world interactions are plentiful and readily available, and a model that can predict future video frames can not only capture useful representations of the world, but can be useful in its own right, for problems such as model-based robotic control. However, a central challenge in video prediction is that the future is highly uncertain: a sequence of past observations of events can imply many possible futures. Although a number of recent works have studied probabilistic models that can represent uncertain futures, such models are either extremely expensive computationally (as in the case of pixel-level autoregressive models), or do not directly optimize the likelihood of the data. In this work, we propose a model for video prediction based on normalizing flows, which allows for direct optimization of the data likelihood, and produces high-quality stochastic predictions. To our knowledge, our work is the first to propose multi-frame video prediction with normalizing flows. We describe an approach for modeling the latent space dynamics, and demonstrate that flow-based generative models offer a viable and competitive approach to generative modeling of video.
From: Manoj Kumar [view email]
[v1] Mon, 4 Mar 2019 18:55:45 UTC (929 KB)

arXiv® is a registered trademark of Cornell University.

If you have a disability and are having trouble accessing information on this website or need materials in an alternate format, contact for assistance.