Sparse Compositional Metric Learning

Abstract : We propose a new approach for metric learning by framing it as learning a sparse combination of locally discriminative metrics that are inexpensive to generate from the training data. This flexible framework allows us to naturally derive formulations for global, multi-task and local metric learning. The resulting algorithms have several advantages over existing methods in the literature: a much smaller number of parameters to be estimated and a principled way to generalize learned met-rics to new testing data points. To analyze the approach theoretically, we derive a generalization bound that justifies the sparse combination. Empirically, we evaluate our algorithms on several datasets against state-of-the-art metric learning methods. The results are consistent with our theoretical findings and demonstrate the superiority of our approach in terms of classification performance and scalability.
Complete list of metadatas

Cited literature [14 references]  Display  Hide  Download

https://hal.inria.fr/hal-01430847
Contributor : Aurélien Bellet <>
Submitted on : Thursday, January 12, 2017 - 2:04:03 PM
Last modification on : Sunday, December 17, 2017 - 7:04:03 AM
Long-term archiving on : Friday, April 14, 2017 - 11:46:02 AM

File

aaai14.pdf
Files produced by the author(s)

Identifiers

  • HAL Id : hal-01430847, version 1

Citation

Yuan Shi, Aurélien Bellet, Fei Sha. Sparse Compositional Metric Learning. AAAI Conference on Artificial Intelligence (AAAI 2014), Jul 2014, Quebec City, Canada. ⟨hal-01430847⟩

Share

Metrics

Record views

61

Files downloads

30