Choosing a Case Base Maintenance Algorithm using a Meta-Case Base

Cummins, Lisa; Bridge, Derek

doi:10.1007/978-1-4471-2318-7_12

Cited by 7 publications

(2 citation statements)

References 20 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Even though some of the algorithm performances may be described by the decision tree rules, self-organizing map visualisations showed that there are clusters of datasets whose performance was not easily explained by the chosen characteristics. One of the key difficulties in meta-learning is still creating features that accurately describe a dataset and reflect the variety of issue complexity [4]. This study suggests a meta-case-based classifier for choosing the optimum Case-Based Reasoning maintenance method (CBR).…”

Section: Literature Surveymentioning

confidence: 99%

Designing a Set of Complexity Measures for Meta-Learning Based Instance Selection

2023

View full text Add to dashboard Cite

A pre-processing method for data mining and machine learning is instance selection. The fundamental issue is that earlier methods still have trouble creating quality samples for classifier training. In this work, the system updates the specification of a few measures that were created for instance selection based on meta-learning. Moreover, the system Three sets of measurements, 59 databases, 16 instance selection techniques, two classifiers, and eight regression learners acting as meta-learners were all employed in an experimental research to compare the two. The findings imply that our metrics outperform those often employed by academics who used a meta-learning-based viewpoint perspective to approach instance selection. Compared to other papers used as benchmarks, this one offers learners more pertinent material. Based on the data that c-measures offer, the algorithm split them into three groups: overlap in property values, class separation, Manifold density, topology, and geometry. New criteria for dealing with meta-learning in IS are established by the approach employed in this work to construct the meta-data together with performance assessment. The system suggests an alternative strategy to the conventional one, in which the top candidate technique is chosen using a trained meta-classifier. Regression meta-learners that had been trained to forecast each potential method's performance were employed in this instance. Such an approach offers additional flexibility for handling variable weight allocations to the objectives in multi-objective problems, such as IS.

show abstract

Section: Literature Surveymentioning

confidence: 99%

Designing a Set of Complexity Measures for Meta-Learning Based Instance Selection

2023

View full text Add to dashboard Cite

show abstract

“…Although Ho and Basu's metrics were designed for two-class problems, some researchers generalize it to more than two classes by averaging measures obtained between all possible pair of classes [37,41]. Also, although recent generalizations of the Ho-Basu c-measures have been proposed [2,12,41], none addresses explicitly the problem of classifying large image datasets.…”

Section: Previous Workmentioning

confidence: 99%

Spectral Metric for Dataset Complexity Assessment

Branchaud-Charron

Achkar²,

Jodoin

2019

2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)

View full text Add to dashboard Cite

In this paper, we propose a new measure to gauge the complexity of image classification problems. Given an annotated image dataset, our method computes a complexity measure called the cumulative spectral gradient (CSG) which strongly correlates with the test accuracy of convolutional neural networks (CNN). The CSG measure is derived from the probabilistic divergence between classes in a spectral clustering framework. We show that this metric correlates with the overall separability of the dataset and thus its inherent complexity. As will be shown, our metric can be used for dataset reduction, to assess which classes are more difficult to disentangle, and approximate the accuracy one could expect to get with a CNN. Results obtained on 11 datasets and three CNN models reveal that our method is more accurate and faster than previous complexity measures.

show abstract

An Optimal Case-Base Maintenance Method for Compositional Adaptation Applications

Mathew

Chakraborti

2019

Case-Based Reasoning Research and Development

View full text Add to dashboard Cite

Choosing a Case Base Maintenance Algorithm using a Meta-Case Base

Cited by 7 publications

References 20 publications

Designing a Set of Complexity Measures for Meta-Learning Based Instance Selection

Designing a Set of Complexity Measures for Meta-Learning Based Instance Selection

Spectral Metric for Dataset Complexity Assessment

An Optimal Case-Base Maintenance Method for Compositional Adaptation Applications

Contact Info

Product

Resources

About