DAG Learning on the PermutahedronDownload PDF

Published: 25 Mar 2022, Last Modified: 25 Nov 2024ICLR2022 OSC PosterReaders: Everyone
Keywords: structure, learning, directed, acyclic, graphs
TL;DR: We propose a DAG learning strategy for learning a total ordering of the variables in the simplex of permutation vectors (the permutahedron).
Abstract: We introduce Daguerro, a strategy for learning directed acyclic graphs (DAGs). In contrast to previous methods, our problem formulation (i) guarantees to learn a DAG, (ii) does not propagate errors over multiple stages, and (iii) can be trained end-to-end without pre-processing steps. Our algorithm leverages advances in differentiable sparse structured inference for learning a total ordering of the variables in the simplex of permutation vectors (the permutahedron), allowing for a substantial reduction in memory and time complexities w.r.t. existing permutation-based continuous optimization methods.
Community Implementations: [![CatalyzeX](/images/catalyzex_icon.svg) 1 code implementation](https://www.catalyzex.com/paper/dag-learning-on-the-permutahedron/code)
3 Replies

Loading