Hi Samuli,
thanks for your questions.
– Do you have knowledge of the usage of JFuzzylite? eg. is it used in production applications and if so how wide is the usage?
Yes, I have customers using it for commercial purposes. I do not know how wide is the usage, but I am sure it is the best option out there today.
– Are Java & C++ versions of the lib similar in features?
Yes, the Java and C++ have the exact same features. There are minor differences in implementation, mostly to take advantage of the languages, but the results, approach, and everything else is the same.
– What is the current development status of the libraries?
The latest version was released March 2017, and a very few updates have been performed focused on minor details. The two libraries are currently maintained, mainly by me, but occassionally I get help from different members of the community. I am currently working towards the Python version of the libraries.
– As part of our evaluation we are building a small prototype which implements a process of learning new rules. The learning process is based on my co-workers [1] previous work & research but I was wondering if you had any pointers on the subject and maybe some tips on how to best do it using jfuzzylite?
For learning rules, maybe you want to check some work on Genetic Fuzzy Systems. Basically, you have an optimisation problem when learning the rules. You could optimise fuzzy systems using other libraries for optimisation.
Cheers.