Learning compact class codes for fast inference in large multi class classification

Abstract : We describe a new approach for classification with a very large number of classes where we assume some class similarity information is available, e.g. through a hierarchical organization. The proposed method learns a compact binary code using such an existing similarity information defined on classes. Binary classifiers are then trained using this code and decoding is performed using a simple nearest neighbor rule. This strategy, related to Error Correcting Output Codes methods, is shown to perform similarly or better than the standard and efficient one-vs-all approach, with much lower inference complexity.
Complete list of metadatas

https://hal.archives-ouvertes.fr/hal-01273309
Contributor : Lip6 Publications <>
Submitted on : Friday, February 12, 2016 - 11:38:43 AM
Last modification on : Thursday, March 21, 2019 - 1:13:43 PM

Links full text

Identifiers

Citation

Moustapha Cissé, Thierry Artières, Patrick Gallinari. Learning compact class codes for fast inference in large multi class classification. European Conference on Machine Learning, Sep 2012, Bristol, United Kingdom. pp.506-520, ⟨10.1007/978-3-642-33460-3_38⟩. ⟨hal-01273309⟩

Share

Metrics

Record views

92