Assessing the predictive performance of artifIcial neural network‐based classifiers based on different data preprocessing methods, distributions and training mechanisms
Abstract
We analyse the implications of three different factors (preprocessing method, data distribution and training mechanism) on the classification performance of artificial neural networks (ANNs). We use three preprocessing approaches: no preprocessing, division by the maximum absolute values and normalization. We study the implications of input data distributions by using five datasets with different distributions: the real data, uniform, normal, logistic and Laplace distributions. We test two training mechanisms: one belonging to the gradient‐descent techniques, improved by a retraining procedure, and the other is a genetic algorithm (GA), which is based on the principles of natural evolution. The results show statistically significant influences of all individual and combined factors on both training and testing performances. A major difference with other related studies is the fact that for both training mechanisms we train the network using as starting solution the one obtained when constructing the network architecture. In other words we use a hybrid approach by refining a previously obtained solution. We found that when the starting solution has relatively low accuracy rates (80–90%) the GA clearly outperformed the retraining procedure, whereas the difference was smaller to non‐existent when the starting solution had relatively high accuracy rates (95–98%). As reported in other studies, we found little to no evidence of crossover operator influence on the GA performance. Co…
Integrity signals
Registry
Cites
Statistics
Claims the paper makes
Claims other papers make about this one
Each is checked against this paper’s own text: does it say what the citing paper says it does?
Backwards propagation is the most popular artificial neural network method.
“They also note backwards propagation (BP) is the most popular ANN method.”
From A Neural Network Approach to Estimating the Allowance for Bad Debt (2011) · cites this paper as Costea and Nastac, 2005
If backwards propagation is applied to a fairly simple problem, the training process may take longer.
“Second, if backwards propagation is applied to a fairly simple problem, the training process may take longer.”
From A Neural Network Approach to Estimating the Allowance for Bad Debt (2011) · cites this paper as Costea and Nastac, 2005
Predictive data mining has two goals: discovering hidden relationships and patterns in data, and building usable prediction models.
“Costea and Nastac (2005) believe predictive data mining has two aims.The first aim is to uncover hidden relationships and patterns in the data.The second aim is to construct usable prediction models.”
From A Neural Network Approach to Estimating the Allowance for Bad Debt (2011) · cites this paper as Costea and Nastac, 2005
As the complexity of the event being analyzed increases, training time for backwards propagation increases non-linearly.
“First, as the complexity of the event being analyzed increases, the training time increases non-linearly.”
From A Neural Network Approach to Estimating the Allowance for Bad Debt (2011) · cites this paper as Costea and Nastac, 2005
The main advantages of neural network classification over traditional approaches are freedom from distributional assumptions, universal approximation, elimination of intercorrelated data problems, and mapping from inputs to outputs without prior knowledge of the function form.
“Costea and Nastac (2005, p. 230) stated that “the main advantages of neural network approaches for classification over the traditional approaches are that ANNs are free of any distributional assumptions, are universal approximations, eliminate problems with intercorrelated data, and provide a mapping function from the input to the outputs without any a priori knowledge about the function form.””
From A Neural Network Approach to Estimating the Allowance for Bad Debt (2011) · cites this paper as Costea and Nastac, 2005
Artificial neural networks that use backwards propagation often rely heavily on initial starting values.
“Often, ANNs which use BP, rely heavily on initial starting values.”
From A Neural Network Approach to Estimating the Allowance for Bad Debt (2011) · cites this paper as Costea and Nastac, 2005
Artificial neural networks are more robust and reliable than conventional analytical procedures.
“Costea and Nastac (2005) find ANNs to be more robust and reliable than conventional analytical procedures.”
From A Neural Network Approach to Estimating the Allowance for Bad Debt (2011) · cites this paper as Costea and Nastac (2005)
Preprocessing the input data can reduce problems associated with backwards propagation.
“As a way to reduce these problems Costea and Nastac (2005, p. 231) suggest preprocessing the input data.”
From A Neural Network Approach to Estimating the Allowance for Bad Debt (2011) · cites this paper as Costea and Nastac, 2005
Cited by
Sorted by citations
A Neural Network Approach to Estimating the Allowance for Bad Debt
Joyner, D. (2011) · VCU Libraries · doi:10.25772/h720-xy24 · Cited by 2 · see citing works
- Authors
- Costea, A., & Nastac, I.
- Affiliations
- Åbo Akademi University (FI)
- Corresponding author
- Costea, Adrian
- ORCID
- Costea, Adrian
- Venue
- Intelligent Systems in Accounting, Finance and Management
- Publisher
- Wiley
- Published
- 2005-12
- Volume
- 13
- Issue
- 4
- Pages
- 217-250
- ISSN
- 1055-615X (ISSN-L)
- 1550-1949
- 2160-0074
- OpenAlex
- W4236234215
- Cited by
- 4
- References
- 0
- Citations by year
- 2020: 1 · 2017: 2 · 2014: 1
- FWCI
- 0.00
- Type
- journal article
- Language
- en
- Open access
- closed
- Locations
- Intelligent systems in accounting, finance and management/Intelligent systems in accounting, finance & management (publishedVersion)
- Topics
- Physical Sciences › Computer Science › Artificial Intelligence › Neural Networks and Applications
- Physical Sciences › Computer Science › Artificial Intelligence › Metaheuristic Optimization Algorithms Research
- Physical Sciences › Computer Science › Information Systems › Data Mining and Machine Learning Applications
- Keywords
- artificial neural networks, classification performance, data preprocessing, normalization, genetic algorithm training, hybrid neural network training, crossover operator
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
