How can hybridization Artifical Neural Network (ANN) and Naive Bayes (NB) Method used to classify text?
Should I do ANN first or NB first? Why?
And how to model the design architecture? What is the input and output?
Hi Gusti,
You can use, e.g., either Vote or Stacking classifier. With Vote classifier, the order (mostly) has no influence on the performance of the utilized algorithms. However, in the case of Stacking classifier, its base classifier has the highest contribution to the general classification process. Knowing that selecting the base classifier is a task-related method.
HTH.
Samer
A paper has been published entitled : A Hybrid Artificial Neural Network-Naive Bayes for solving imbalanced dataset problems in semiconductor manufacturing test process.
They use those two techniques to solve the problems. In this case they use output from ANN as additional input for Naive Bayes (it means ANN first then NB second). however, you need to see the nature of the data that u will used. Hope that help.
Please go through the paper
Adaptive multi-classifier fusion approach for gene expression dataset based on probabilistic theory
http://www.sciencedirect.com/science/article/pii/S1226319214000738
where we have used Naïve Bayes classifier to predict the secondary level classifier for the prediction of class label.
1. I have done variety of experiment for that project... And ended with such idea.
Now You can frame the An adaptive Neural Network where the hidden neurons will or will not participate in sending the signal further depending upon the naïve bayes probability... I know it sound odd but believe me it will show good result..
Only thing you have to figure out condition to make call for bayes probability...
Keep writing .. and post your update... You can mail me at [email protected]
- Badges
- Science topic