Algorithms for metric learning via contrastive embeddings

@inproceedings{Centurion2019AlgorithmsFM,
  title={Algorithms for metric learning via contrastive embeddings},
  author={Diego Ihara Centurion and Neshat Mohammadi and Anastasios Sidiropoulos},
  booktitle={SoCG},
  year={2019}
}
We study the problem of supervised learning a metric space under discriminative constraints. Given a universe $X$ and sets ${\cal S}, {\cal D}\subset {X \choose 2}$ of similar and dissimilar pairs, we seek to find a mapping $f:X\to Y$, into some target metric space $M=(Y,\rho)$, such that similar objects are mapped to points at distance at most $u$, and dissimilar objects are mapped to points at distance at least $\ell$. More generally, the goal is to find a mapping of maximum accuracy (that is… 
1 Citations

Figures from this paper

Learning Lines with Ordinal Constraints
TLDR
This work studies the problem of finding a mapping from a set of points into the real line, under ordinal triple constraints, and presents an approximation algorithm for the dense case of this problem.

References

SHOWING 1-10 OF 34 REFERENCES
Metric clustering via consistent labeling
TLDR
This work is the first to emphasize relative guarantees, that compare the produced solution to the optimal one for the input at hand, and provides a family of linear programming relaxations and simple randomized rounding procedures that achieve provably good approximation guarantees.
Distance Metric Learning for Large Margin Nearest Neighbor Classification
TLDR
This paper shows how to learn a Mahalanobis distance metric for kNN classification from labeled examples in a globally integrated manner and finds that metrics trained in this way lead to significant improvements in kNN Classification.
Ordinal Embedding: Approximation Algorithms and Dimensionality Reduction
TLDR
Two polynomial-time constant-factor approximation algorithms for minimizing the relaxation in an embedding of an unweighted graph into a line metric and into a tree metric are developed.
Information-theoretic metric learning
TLDR
An information-theoretic approach to learning a Mahalanobis distance function that can handle a wide variety of constraints and can optionally incorporate a prior on the distance function and derive regret bounds for the resulting algorithm.
Reality Distortion: Exact and Approximate Algorithms for Embedding into the Line
TLDR
Algorithms for the problem of minimum distortion embeddings of finite metric spaces into the real line (or a finite subset of the line) are described, which yields a quasipolynomial running time for constant δ, and polynomial D.
Metric Learning: A Survey
  • B. Kulis
  • Computer Science
    Found. Trends Mach. Learn.
  • 2013
TLDR
Metric Learning: A Review presents an overview of existing research in this topic, including recent progress on scaling to high-dimensional feature spaces and to data sets with an extremely large number of data points.
INAPPROXIMABILITY FOR METRIC EMBEDDINGS INTO R
TLDR
This work considers the problem of computing the smallest possible distortion for embedding of a given n-point metric space into Rd, where d is fixed (and small), and derives inapproximability with a factor roughly n1/(22d−10) for every fixed d ≥ 2 by a conceptually very simple reduction.
Inapproximability for metric embeddings into Rd
TLDR
This work considers the problem of computing the smallest possible distortion for embedding of a given n-point metric space into ℝ d, where d is fixed (and small), and derives inapproximability with a factor roughly n 1/(22d-10) for every fixed d ≥ 2 by a conceptually very simple reduction.
Probabilistic approximation of metric spaces and its algorithmic applications
  • Y. Bartal
  • Computer Science, Mathematics
    Proceedings of 37th Conference on Foundations of Computer Science
  • 1996
TLDR
It is proved that any metric space can be probabilistically-approximated by hierarchically well-separated trees (HST) with a polylogarithmic distortion.
Learning task-specific similarity
TLDR
An algorithmic approach to learning similarity from examples of what objects are deemed similar according to the task-specific notion of similarity at hand, as well as optional negative examples, which allows to predict when two previously unseen examples are similar and to efficiently search a very large database for examples similar to a query.
...
1
2
3
4
...