I would like to explore feature selection using genetic algorithm, particularly areas of image processing, i.e. image recognition, fingerprint matching, edge detection, OCR etc. My questions are: 1.Is there any test image with which I can test my algorithm? 2.Any suggestions about the design of fitness function? Also any good webpage recommended? Thanks a lot for help!
1
votes
Wouldn't a proper test image depend entirely on what it is you're trying to test? How could we possibly know what would be best suited for your needs? For what concerns fitness, a good fitness function can quickly assess whether or not a solution is good or bad and reduce that assessment to a simple score.
– Neil
The Q #1 is hard to answer without more information. Q#2: what have you tried so far?
– nouney
2 Answers
0
votes
0
votes
for dataset visit https://archive.ics.uci.edu/ml/datasets.html
for fitness function, firstly describe your problem: classification or regression? if classification, how many class? in general for classification problems maximizing (true detection/(false detection + 1)) is a good fitness function
also i think for feature extraction(selection) tasks, genetic programming led to better answers.
