Timezone: »
Many manifold learning algorithms aim to create embeddings with low or no distortion (i.e. isometric). If the data has intrinsic dimension d, it is often impossible to obtain an isometric embedding in d dimensions, but possible in s > d dimensions. Yet, most geometry preserving algorithms cannot do the latter. This paper proposes an embedding algorithm that overcomes this problem. The algorithm directly computes, for any data embedding Y, a distortion loss(Y), and iteratively updates Y in order to decrease it. The distortion measure we propose is based on the push-forward Riemannian metric associated with the coordinates Y. The experiments confirm the superiority of our algorithm in obtaining low distortion embeddings.
Author Information
James McQueen (University of Washington)
Marina Meila (University of Washington)
Dominique Perrault-Joncas (Google)
More from the Same Authors
-
2020 Session: Orals & Spotlights Track 27: Unsupervised/Probabilistic »
Marina Meila · Kun Zhang -
2018 : Invited Talk 1 »
Marina Meila -
2018 Poster: How to tell when a clustering is (approximately) correct using convex relaxations »
Marina Meila -
2017 : Riemannian metric estimation and the problem of isometric embedding »
Dominique Perrault-Joncas -
2017 : Topological Data Analisys with GUDHI and scalable manifold learning and clustering with megaman »
Vincent Rouvreau · Marina Meila -
2017 : Discussion: Geometric Data Analysis »
Frederic Chazal · Marina Meila -
2017 Workshop: Synergies in Geometric Data Analysis (TWO DAYS) »
Marina Meila · Frederic Chazal · Yu-Chia Chen -
2017 Poster: Improved Graph Laplacian via Geometric Self-Consistency »
Dominique Perrault-Joncas · Marina Meila · James McQueen -
2016 Poster: Graph Clustering: Block-models and model free results »
Yali Wan · Marina Meila -
2015 Poster: A class of network models recoverable by spectral clustering »
Yali Wan · Marina Meila -
2014 Poster: Recursive Inversion Models for Permutations »
Christopher Meek · Marina Meila -
2011 Poster: Directed Graph Embedding: an Algorithm based on Continuous Limits of Laplacian-type Operators »
Dominique C Perrault-Joncas · Marina Meila -
2011 Spotlight: Directed Graph Embedding: an Algorithm based on Continuous Limits of Laplacian-type Operators »
Dominique C Perrault-Joncas · Marina Meila -
2009 Workshop: Learning with Orderings »
Tiberio Caetano · Carlos Guestrin · Jonathan Huang · Risi Kondor · Guy Lebanon · Marina Meila