Search Results for author:
Found 3 papers, 0 papers with code
Investigation of Frame Differences as Motion Cues for Video Object Segmentation
Our results suggest the usefulness of employing frame differences as motion cues in cases with limited computational resources.
From Coupled Oscillators to Graph Neural Networks: Reducing Over-smoothing via a Kuramoto Model-based Approach
We propose the Kuramoto Graph Neural Network (KuramotoGNN), a novel class of continuous-depth graph neural networks (GNNs) that employs the Kuramoto model to mitigate the over-smoothing phenomenon, in which node features in GNNs become indistinguishable as the number of layers increases.
An Analytic Solution to the Inverse Ising Problem in the Tree-reweighted Approximation
By solving the moment-matching and self-consistency conditions analytically, we can derive the interaction matrix as a function of the given data statistics.
Cannot find the paper you are looking for? You can
Submit a new open access paper.
