Evolving hybrid ensembles of learning machines for better generalisation

Chandra, Arjun; Yao, Xin

doi:10.1016/j.neucom.2005.12.014

Cited by 132 publications

(52 citation statements)

References 21 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…This includes the inference procedures and excellent predictive capabilities deployed by the techniques. A single overall technique that comes out of this approach of combining two or more existing techniques is called a hybrid system (Chandra and Yao 2006;Khashei et al 2011). It is an approach that combines different theoretical backgrounds and algorithms such as data mining and soft computing methodologies.…”

Section: Hybrid Intelligent Systemsmentioning

confidence: 99%

Hybrid intelligent systems in petroleum reservoir characterization and modeling: the journey so far and the challenges ahead

Anifowose

Labadin

Abdulraheem

2016

J Petrol Explor Prod Technol

View full text Add to dashboard Cite

Computational intelligence (CI) techniques have positively impacted the petroleum reservoir characterization and modeling landscape. However, studies have showed that each CI technique has its strengths and weaknesses. Some of the techniques have the ability to handle datasets of high dimensionality and fast in execution, while others are limited in their ability to handle uncertainties, difficult to learn, and could not deal with datasets of high or low dimensionality. The ''no free lunch'' theorem also gives credence to this problem as it postulates that no technique or method can be applicable to all problems in all situations. A technique that worked well on a problem may not perform well in another problem domain just as a technique that was written off on one problem may be promising with another. There was the need for robust techniques that will make the best use of the strengths to overcome the weaknesses while producing the best results. The machine learning concepts of hybrid intelligent system (HIS) have been proposed to partly overcome this problem. In this review paper, the impact of HIS on the petroleum reservoir characterization process is enumerated, analyzed, and extensively discussed. It was concluded that HIS has huge potentials in the improvement of petroleum reservoir property predictions resulting in improved exploration, more efficient exploitation, increased production, and more effective management of energy resources. Lastly, a number of yet-to-be-explored hybrid possibilities were recommended.

show abstract

Section: Hybrid Intelligent Systemsmentioning

confidence: 99%

Hybrid intelligent systems in petroleum reservoir characterization and modeling: the journey so far and the challenges ahead

Anifowose

Labadin

Abdulraheem

2016

J Petrol Explor Prod Technol

View full text Add to dashboard Cite

show abstract

“…Nevertheless, ensemble learning methods which directly encourage diversity by considering the interaction among all ensemble members during the learning, such as NCL, have been showing to outperform other ensemble learning approaches in off-line mode [10], [11], [21]. So, ensemble learning methods in which the ensemble members are not trained independently have a potential advantage over methods which perform independent training in on-line mode.…”

Section: A On-line Ensemble Learningmentioning

confidence: 99%

“…In this method, the training of an ensemble member is influenced by the others, directly encouraging diversity through the use of a penalty correlation term in the error function of the base model learning algorithm. In off-line mode, ensembles which directly encourage diversity considering the interaction among all the ensemble members during the learning [10], [11], [21] have been showing to outperform other ensemble learning methods such as bagging and boosting. So, this is a potentially powerful advantage of NCL over the other on-line ensemble methods existent in the literature.…”

Section: A On-line Ensemble Learningmentioning

confidence: 99%

On-line bagging Negative Correlation Learning

Minku¹,

Yao²

2008

2008 IEEE International Joint Conference on Neural Networks (IEEE World Congress on Computational Intelligence)

View full text Add to dashboard Cite

Abstract-Negative Correlation Learning (NCL) has beenshowing to outperform other ensemble learning approaches in off-line mode. A key point to the success of NCL is that the learning of an ensemble member is influenced by the learning of the others, directly encouraging diversity. However, when applied to on-line learning, NCL presents the problem that part of the diversity has to be built a priori, as the same sequence of training data is sent to all the ensemble members. In this way, the choice of the base models to be used is limited and the use of more adequate neural network models for the problem to be solved may be not possible. This paper proposes a new

show abstract

“…Classification approaches based on ensemble learning is a way to combine weak learners to produce a strong learner (Chandra and Yao, 2006). In such cases, handling of small inaccuracies in the training data is not crucial for ensemble classifiers (when comprised of advanced non-parametric classifiers such as neural networks and Support Vector Machines) (Colditz et al, 2008).…”

Section: Introductionmentioning

confidence: 99%

“…In a study proposed by (Brodley and Friedl, 2011), a consensus voting scheme is employed to filter results from an ensemble of classifiers to eliminate mislabeled samples. Other methodologies to remove mislabels using predicted membership in other fields, like computer vision and machine learning, can be seen in (Chandola et al, 2009;Escalante, 2005;Hodge and Austin, 2004). …”

mentioning

confidence: 99%

Automatic Training Sample Selection for a Multi-Evidence Based Crop Classification Approach

Menaka

Ferré

Greve

2014

Int. Arch. Photogramm. Remote Sens. Spatial Inf. Sci.

View full text Add to dashboard Cite

ABSTRACT:An approach to use the available agricultural parcel information to automatically select training samples for crop classification is investigated. Previous research addressed the multi-evidence crop classification approach using an ensemble classifier. This first produced confidence measures using three Multi-Layer Perceptron (MLP) neural networks trained separately with spectral, texture and vegetation indices; classification labels were then assigned based on Endorsement Theory. The present study proposes an approach to feed this ensemble classifier with automatically selected training samples. The available vector data representing crop boundaries with corresponding crop codes are used as a source for training samples. These vector data are created by farmers to support subsidy claims and are, therefore, prone to errors such as mislabeling of crop codes and boundary digitization errors. The proposed approach is named as ECRA (Ensemble based Cluster Refinement Approach). ECRA first automatically removes mislabeled samples and then selects the refined training samples in an iterative training-reclassification scheme. Mislabel removal is based on the expectation that mislabels in each class will be far from cluster centroid. However, this must be a soft constraint, especially when working with a hypothesis space that does not contain a good approximation of the targets classes. Difficulty in finding a good approximation often exists either due to less informative data or a large hypothesis space. Thus this approach uses the spectral, texture and indices domains in an ensemble framework to iteratively remove the mislabeled pixels from the crop clusters declared by the farmers. Once the clusters are refined, the selected border samples are used for final learning and the unknown samples are classified using the multi-evidence approach. The study is implemented with WorldView-2 multispectral imagery acquired for a study area containing 10 crop classes. The proposed approach is compared with the multi-evidence approach based on training samples selected randomly and border samples based on initial cluster centroids within agricultural parcels without any refinement. The results clarify the improvement in overall classification accuracy to 82.3% based on the proposed approach from 74.9 % based on random selection and 71.4% on non-refined border samples.* Corresponding author. This is useful to know for communication with the appropriate person in cases with more than one author.

show abstract

Evolving hybrid ensembles of learning machines for better generalisation

Cited by 132 publications

References 21 publications

Hybrid intelligent systems in petroleum reservoir characterization and modeling: the journey so far and the challenges ahead

Hybrid intelligent systems in petroleum reservoir characterization and modeling: the journey so far and the challenges ahead

On-line bagging Negative Correlation Learning

Automatic Training Sample Selection for a Multi-Evidence Based Crop Classification Approach

Contact Info

Product

Resources

About