Hybrid approaches for classification under information acquisition cost constraint

Pendharkar, Parag C.

doi:10.1016/j.dss.2004.07.001

Cited by 11 publications

(5 citation statements)

References 37 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Several studies in the literature report the sensitivity of machine learning approaches to different data distributions (Koehler, 1991; Bhattacharyya & Pendharkar, 1998; Pendharkar, 2001, 2002; Pendharkar & Rodger, 2004). The use of simulated data allows a researcher to test the robustness and performance of different classification approaches across different data distributions (Pendharkar, 2005a). Using the methodology described in Vale and Maurelli (1983), we generated 40 random data samples for each of three different data distributions.…”

Section: Data Experiments and Resultsmentioning

confidence: 99%

A comparison of gradient ascent, gradient descent and genetic‐algorithm‐based artificial neural networks for the binary classification problem

Pendharkar¹

2007

Expert Systems

Self Cite

View full text Add to dashboard Cite

We compare log maximum likelihood gradient ascent, root-mean-square error minimizing gradient descent and genetic-algorithm-based artificial neural network procedures for a binary classification problem. We use simulated data and real-world data sets, and four different performance metrics of correct classification, sensitivity, specificity and reliability for our comparisons. Our experiments indicate that a genetic-algorithmbased artificial neural network that maximizes the total number of correct classifications generally fares well for the binary classification problem. However, if the training data set contains inconsistent decisions or noise then the log maximum likelihood maximizing gradient ascent may be the best classification approach to use. The root-mean-square minimizing gradient descent approach appears to overfit training data and has the lowest reliability among the approaches considered for our research. At the end of the paper, we provide a few guidelines, including computational complexity, for selection of an appropriate technique for a given binary classification problem.

show abstract

Section: Data Experiments and Resultsmentioning

confidence: 99%

A comparison of gradient ascent, gradient descent and genetic‐algorithm‐based artificial neural networks for the binary classification problem

Pendharkar¹

2007

Expert Systems

Self Cite

View full text Add to dashboard Cite

show abstract

“…Their paper surveyed 169 works on hybrids and concluded with a brief guideline to develop hybrid algorithms. It can be observed that although hybrid approaches have traditionally merged exact and/or heuristic algorithms (Gallardo et al, 2007), more and more research is devoted to hybrid methods including other approaches such as simulation (Peng et al, 2006), constraint programming (Correa et al, 2004;Hooker, 2006), neural networks (Pendharkar, 2005;Sahoo and Maity, 2007) and multi-agent approaches (Yan and Zhou, 2006).…”

Section: Decisionsmentioning

confidence: 98%

A hybrid collaborative algorithm to solve an integrated wood transportation and paper pulp production problem

Pécora

Ruiz

Soriano

2016

Journal of the Operational Research Society

View full text Add to dashboard Cite

This paper proposes a hybrid algorithm to tackle a real-world problem arising in the context of pulp and paper production. This situation is modelled as a production problem where one has to decide which wood will be used by each available processing unit (wood cooker) in order to minimize the variance of wood densities within each cooker for each period of the planning horizon. The proposed hybrid algorithm is built around two distinct phases. The first phase uses two interacting heuristic methods to identify a promising reduced search space, which is then thoroughly explored in the second phase. This hybrid algorithm produces high-quality solutions in reasonable computation times, especially for the largest test instances. Extensive computational experiments demonstrated the robustness and efficiency of the method.

show abstract

“…We compare the eight different techniques by employing two different types of data sets. The first type of data set is a simulated data set that has been used in a previous study [44]. This data set incorporates various distributional assumptions for the underlying variables.…”

Section: Experiments On Simulated and Real-world Data Setsmentioning

confidence: 99%

“…We use simulated data to evaluate the performance of different techniques because simulated data allows us to control the group data distribution, the number of examples belonging to each group, and the number of attributes. Part of the data set used in our study has been used in a previous study [44]. In this study, we use the simulator from Pendharkar [44] and generate 60 data samples, 30 samples each for uniform (kurtosis Ϫ1) and normal (kurtosis 0) group distributions.…”

Section: Simulated Datamentioning

confidence: 99%

See 1 more Smart Citation

A misclassification cost‐minimizing evolutionary–neural classification approach

Pendharkar

Nanda

2006

Naval Research Logistics

View full text Add to dashboard Cite

Machine learning algorithms that incorporate misclassification costs have recently received considerable attention. In this paper, we use the principles of evolution to develop and test an evolutionary/genetic algorithm (GA)-based neural approach that incorporates asymmetric Type I and Type II error costs. Using simulated, real-world medical and financial data sets, we compare the results of the proposed approach with other statistical, mathematical, and machine learning approaches, which include statistical linear discriminant analysis, back-propagation artificial neural network, integrated cost preference-based linear mathematical programming-based minimize squared deviations, linear integrated cost preference-based GA, decision trees (C 5.0, and CART), and inexpensive classification with expensive tests algorithm. Our results indicate that the proposed approach incorporating asymmetric error costs results in equal or lower holdout sample misclassification cost when compared with the other statistical, mathematical, and machine learning misclassification cost-minimizing approaches.

show abstract

Hybrid approaches for classification under information acquisition cost constraint

Cited by 11 publications

References 37 publications

A comparison of gradient ascent, gradient descent and genetic‐algorithm‐based artificial neural networks for the binary classification problem

A comparison of gradient ascent, gradient descent and genetic‐algorithm‐based artificial neural networks for the binary classification problem

A hybrid collaborative algorithm to solve an integrated wood transportation and paper pulp production problem

A misclassification cost‐minimizing evolutionary–neural classification approach

Contact Info

Product

Resources

About