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

Authors:
M. Cissé;T. Artières;Patrick Gallinari
Affiliations:
Laboratoire d'Informatique de Paris 6 (LIP6), Université Pierre et Marie Curie, Paris, France;Laboratoire d'Informatique de Paris 6 (LIP6), Université Pierre et Marie Curie, Paris, France;Laboratoire d'Informatique de Paris 6 (LIP6), Université Pierre et Marie Curie, Paris, France
Venue:
ECML PKDD'12 Proceedings of the 2012 European conference on Machine Learning and Knowledge Discovery in Databases - Volume Part I
Year:
2012

Citing 10
Cited 0

Reducing multiclass to binary: a unifying approach for margin classifiers

The Journal of Machine Learning Research
In Defense of One-Vs-All Classification

The Journal of Machine Learning Research
Hierarchical document categorization with support vector machines

Proceedings of the thirteenth ACM international conference on Information and knowledge management
An incremental node embedding technique for error correcting output codes

Pattern Recognition
Extracting and composing robust features with denoising autoencoders

Proceedings of the 25th international conference on Machine learning
Refined experts: improving classification in large taxonomies

Proceedings of the 32nd international ACM SIGIR conference on Research and development in information retrieval
Zero-data learning of new tasks

AAAI'08 Proceedings of the 23rd national conference on Artificial intelligence - Volume 2
Solving multiclass learning problems via error-correcting output codes

Journal of Artificial Intelligence Research
Conditional probability tree estimation analysis and algorithms

UAI '09 Proceedings of the Twenty-Fifth Conference on Uncertainty in Artificial Intelligence
The ECIR 2010 large scale hierarchical classification workshop

ACM SIGIR Forum

Quantified Score

Hi-index	0.00

Visualization

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.