This paper will be referenced for this week's meeting:
Spherical CNNs on unstructured grids
Authors: Chiyu “Max” Jiang, Prabhat, Jingwei Huang, Philip Marcus, Karthik Kashinath, Matthias Nießner
Abstract: We present an efficient convolution kernel for Convolutional Neural Networks (CNNs) on unstructured grids using parameterized differential operators while focusing on spherical signals such as panorama images or planetary signals. To this end, we replace conventional convolution kernels with linear combinations of differential operators that are weighted by learnable parameters. Differential operators can be efficiently estimated on unstructured grids using one-ring neighbors, and learnable parameters can be optimized through standard back-propagation. As a result, we obtain extremely efficient neural networks that match or outperform state-of-the-art network architectures in terms of performance but with a significantly smaller number of network parameters. We evaluate our algorithm in an extensive series of experiments on a variety of computer vision and climate science tasks, including shape classification, climate pattern segmentation, and omnidirectional image semantic segmentation. Overall, we (1) present a novel CNN approach on unstructured grids using parameterized differential operators for spherical signals, and (2) show that our unique kernel parameterization allows our model to achieve the same or higher accuracy with significantly fewer network parameters.
Full details about this meeting will be posted here: https://www.benty-fields.com/manage_jc?groupid=191.
The Berkeley Statistics and Machine Learning Forum meets weekly to discuss current applications across a wide variety of research domains and software methodologies. Register here to view, propose and vote for this group's upcoming discussion topics. All interested members of the UC Berkeley and LBL communities are welcome and encouraged to attend. Questions may be directed to François Lanusse.