A constructive algorithm for training cooperative neural network ensembles

By M. Islam, Xin Yao, Kazuyuki Murase. 2003

329 citations recorded by the source corpus.

Subfields: Artificial Intelligence, Computer Vision and Pattern Recognition

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

1. Lior Rokach
2. Mehryar Mohri
3. Corinna Cortes
4. Vitaly Kuznetsov
5. Q. Henry Wu
6. Shang He
7. Tianyu Pang
8. Jun Zhu

In Computer Vision and Pattern Recognition

1. Lior Rokach
2. Q. Henry Wu
3. Jon R. Saunders
4. Shang He
5. Tianyu Pang
6. Xin Yao
7. Jun Zhu
8. Kazuyuki Murase

Citations in the corpus

• AdaNet: Adaptive Structural Learning of Artificial Neural Networks (2016, background): “L G ] 5 J u l 2016 3. adapts the structure and complexity of the network to the difficulty of the particular problem at hand, with no pre-defined architecture.”
• Metaheuristic design of feedforward neural networks: A review of two decades of research (2017, background): “Additionally, a constructive-cooperative-neural-network-ensemble was proposed in [320] that determined ensemble size by focusing on accuracy and diversity during a constructive, cooperative procedure [321].”
• Group Search Optimizer: An Optimization Algorithm Inspired by Animal Searching Behavior (2009, methodology): “Algorithm GSOANN GANet-best [53] COOP [58] CNNE [59] EPNet [51]”
• Comprehensive Review of Neural Network-Based Prediction Intervals and New Advances (2011, background): “Traditionally, it is proven that model aggregation is effective in improving the generalization ability [41], [42].”
• Ensemble-based classifiers (2010, result): “Nevertheless the CNNE algorithm (Islam et al. 2003) which simultaneously determines the ensemble size along with the number of hidden nodes in individual NNs, has produced improved results.”
• Hybrid and ensemble-based soft computing techniques in bankruptcy prediction: a survey (2010, background): “To achieve the tradeoff when designing an ensemble of MLPs, negative correlation learning has been proposed (Liu and Yao 1999; Liu et al. 2000; Islam et al. 2003).”
• Diversity creation methods: a survey and categorisation (2004, methodology): “[62] could easily be applied to an ensemble containing networks with differing numbers of hidden nodes, having first used an algorithm like CNNE [59] to ensure the network architectures are established appropriately.”
• Improving Adversarial Robustness via Promoting Ensemble Diversity (2019, background): “The previous analysis states that there is no strict definition of what is intuitively perceived as ensemble diversity (Kuncheva & Whitaker, 2003) and most of the previous work defines the ensemble diversity concerning the prediction errors of individual classifiers in the ensemble (Liu & Yao, 1999a”
• Layered Ensemble Architecture for Time Series Forecasting (2016, background): “It has been known that training networks using different training sets is the most effective way for producing diverse trained networks [34], [52].”