An Evaluation of Multi-Expert Configurations for the Recognition of Handwritten Numerals

Rahman, Fuad; Fairhurst, Michael

doi:10.1016/s0031-3203(97)00161-1

Cited by 48 publications

(25 citation statements)

References 0 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…Rather what we have is a bag of tools and a bag of problems." (Kanal 1974) This observation has engendered considerable recent interest in Multiple Classifier Systems [1,2,3,4,5,6], which seek to make use of the divergence in design methodologies to limit such a priori impositions and obtain a correspondingly better estimate of the decision boundary in order to boost classification performance.…”

Section: Construction Of a Generalized Theory Of Classifier Fusionmentioning

confidence: 99%

Morphologically Debiased Classifier Fusion: A Tomography-Theoretic Approach

Windridge

2005

Advances in Imaging and Electron Physics

View full text Add to dashboard Cite

Section: Construction Of a Generalized Theory Of Classifier Fusionmentioning

confidence: 99%

Morphologically Debiased Classifier Fusion: A Tomography-Theoretic Approach

Windridge

2005

Advances in Imaging and Electron Physics

View full text Add to dashboard Cite

“…So ultimately, the decision combination is based on evaluating the confidences in recognising individual patterns, but this still does not include information about the strengths of individual experts in classifying various classes or any other special considerations. It is entirely possible to evaluate other confidence indices extracted from the participating experts beside the confidence index associated with each decision and incorporate them in the decision combination process [19,20]. Traditional approaches to decision combination fail to include these diverse sources of discriminating information while trying to arrive at a consensus.…”

Section: Traditional Approaches To Parallel Decision Fusionmentioning

confidence: 99%

A Novel Confidence-Based Framework for Multiple Expert Decision Fusion

Rahman

Fairhurst²

1998

Procedings of the British Machine Vision Conference 1998

View full text Add to dashboard Cite

A novel confidence-based parallel multiple expert decision combination framework is introduced. The traditional approaches to parallel multiple source decision fusion either take no account of the confidence of each decision by each participating expert in the combined framework or only exploit the confidences associated with each decision. But it is entirely possible to incorporate more additional a priori information in the form of various confidence indices that can be estimated from the performances of various participating experts. The confidence-based parallel multiple expert decision combination framework proposed here addresses this shortcoming. Very encouraging results have been obtained by implementing this proposed framework in combining decisions of multiple experts applied to the problem of handwritten and machine printed character recognition.

show abstract

“…In the case of the sum rule combination scheme, the mathematical analogy with back-projection is exact (the remaining combination schemes (eg [16][17][18][19][20][21]) having been demonstrated to act as specifically constrained forms of back-projection), and the proposed 'tomographically-filtered' combination scheme thus acts in relation to this particular fusion methodology. Precise algorithmic details of the methodology arising from this approach are set out in [4], the main results of which are reprinted (with permission) in appendix 1, which will thus serve in the current scenario as the test-bed tomographic combination methodology through which we shall address the paper's objectives.…”

Section: Introductionmentioning

confidence: 99%

Performance Measures of the Tomographic Classifier Fusion Methodology

Windridge

Kittler

2005

Int. J. Patt. Recogn. Artif. Intell.

View full text Add to dashboard Cite

We seek to quantify both the classification performance and estimation error robustness of the authors' tomographic classifier fusion methodology by contrasting it in field tests and model scenarios with the sum and product classifier fusion methodologies.In particular, we seek to confirm that the tomographic methodology represents a generally optimal strategy across the entire range of problem dimensionalities, and at a sufficient margin to justify the general advocation of its use. Final results indicate, in particular, a near 25% improvement on the next nearest performing combination scheme at the extremity of the tested dimensional range.

show abstract

An Evaluation of Multi-Expert Configurations for the Recognition of Handwritten Numerals

Cited by 48 publications

References 0 publications

Morphologically Debiased Classifier Fusion: A Tomography-Theoretic Approach

Morphologically Debiased Classifier Fusion: A Tomography-Theoretic Approach

A Novel Confidence-Based Framework for Multiple Expert Decision Fusion

Performance Measures of the Tomographic Classifier Fusion Methodology

Contact Info

Product

Resources

About