Identifying defects on plywood using a minimum distance classifier and a neural network
This paper describes the application of a minimum distance classifier and a neural network in identifying defects on plywood. The performance achieved by these two classifiers in this study has been used to compare the two methods for classification tasks. While the neural network misclassification fell to 13.5% that of the minimum distance classifier remained at 37% showing the superiority of the neural network. This is due to its inherent ability to deal with nonlinearity and create soft decision boundaries to separate pattern classes, thus making it a more efficient and intelligent classifier.
Hi Ebubekir,
Thank you for your interest in my work. I used neural network classifier and minimum distance classifier for the problem of identifying defects on plywood. The minimum distance classifier accuracy (using 17 input features) was 63%. The neural network accuracy (using 17 input features) was 86.5% as reported in this paper. The neural network accuracy was further improved to 88% by applying a novel feature selection technique using 12 input features. Finally, a decision tree based on modular neural networks achieved an accuracy of 95.7%. The reason for using neural network based classifier is apparent from the results obtained and is due to its ability to learn non linear decision boundaries from training data.
Thank you,
Michael.
Hi, thank you for this interesting study and paper.. I am curious that did you try any other methods for your trouble with plywoods and Have this technique that you are using in here, compared with them? If so, what is the result of this comparison, and why did you chose this technique...
Many Thanks
Ebubekir