HARAM: A Hierarchical ARAM Neural Network for Large-Scale Text Classification
Authors:
- Fernando BenitesFBFernando Benites
- Elena SapozhnikovaESElena Sapozhnikova
- 1 November 2015
- conference paper
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 847-854
Abstract
With the rapid development of the Web, the need for text classification of large data volumes is permanently growing. Texts represented as bags-of-words possess usually very high dimensionality in the input space and often also in the output space if labeled with many categories. As a result, neural classifiers should be adapted to such large-scale data. We present here a well scalable extension to the fuzzy Adaptive Resonance Associative Map (ARAM) neural network which was specially developed for high-dimensional and large data. This extension aims at increasing the classification speed by adding an extra ART layer for clustering learned prototypes into large clusters. In this case the activation of all prototypes can be replaced by the activation of a small fraction of them, leading to a significant reduction of the classification time. This extension can be especially useful for multi-label classification tasks.Keywords
This publication has 15 references indexed in Scilit:
- Large-Scale Multi-label Text Classification — Revisiting Neural Networks2014
- Factors of overtraining with fuzzy ARTMAP neural networks2006
- AFC: ART-Based Fuzzy Classifier2003
- Overtraining in fuzzy ARTMAP: Myth or reality?2002
- MicroARTMAP: use of mutual information for category reduction in fuzzy ARTMAP2000
- Cascade fuzzy ART: a new extensible database for model-based object recognition1996
- NewsWeeder: Learning to Filter Netnews1995
- Fuzzy ARTMAP: A neural network architecture for incremental supervised learning of analog multidimensional mapsIEEE Transactions on Neural Networks, 1992
- R-treesACM SIGMOD Record, 1984
- Multidimensional binary search trees used for associative searchingCommunications of the ACM, 1975