Learning Good Edit Similarities with Generalization Guarantees

Abstract : Similarity and distance functions are essential to many learning algorithms, thus training them has attracted a lot of interest. When it comes to dealing with structured data (e.g., strings or trees), edit similarities are widely used, and there exists a few methods for learning them. However, these methods offer no theoretical guarantee as to the generalization performance and discriminative power of the resulting similarities. Recently, a theory of learning with good similarity functions was proposed. This new theory bridges the gap between the properties of a similarity function and its performance in classification. In this paper, we propose a novel edit similarity learning approach (GESL) driven by the idea of goodness, which allows us to derive generalization guarantees using the notion of uniform stability. We experimentally show that edit similarities learned with our method induce classification models that are both more accurate and sparser than those induced by the edit distance or edit similarities learned with a state-of-the-art method.
Document type :
Conference papers
Complete list of metadatas

Cited literature [18 references]  Display  Hide  Download

Contributor : Marc Sebban <>
Submitted on : Wednesday, January 7, 2015 - 3:26:40 PM
Last modification on : Tuesday, September 10, 2019 - 11:32:08 AM
Long-term archiving on : Friday, September 11, 2015 - 1:30:43 AM


Files produced by the author(s)


  • HAL Id : hal-00608631, version 1


Aurélien Bellet, Amaury Habrard, Marc Sebban. Learning Good Edit Similarities with Generalization Guarantees. European Conference on Machine Learning, Sep 2011, Athens, Greece. pp.188-203. ⟨hal-00608631⟩



Record views


Files downloads