Usually the features dimensions reduction step is performed before classification, which allows removing the redundant and irrelevant noisy data from the large features vector representing an image.
Hi Mohammed,
The best usage can be achieved through applying Genetic algorithm in conjunction with, e.g., Correlation-based Feature Subset Selection method. After that, classification algorithm will rely on the selected feature subset in order to build the ultimate prediction model.
NB: Previous process has to be evaluated under cross-validation method, for example.
HTH.
Samer
A good introductory paper that covers this subject:
http://neuro.bstu.by/our/Data-mining/fereitas-ga.pdf
The below one may also be helpful:
https://www.cs.kent.ac.uk/people/staff/aaf/pub_papers.dir/RASC-2002-Gisele.pdf
- Badges
- Science topic
- Similar topics
- Vectorization