2
$\begingroup$

Soft computing is defined in [1] by the capability of "operating with uncertain, imprecise and incomplete information in a manner that reflects human thinking". So, based on my limited understanding, SVM is not a soft computing method as it is not intrinsically capable of dealing with imprecise information like it is the case with fuzzy logic, for example.

I assume that I am wrong. If so, what are the properties and capabilities of SVM that makes it a "soft computing" method?

[1]: Negnevitsky, Michael. Artificial intelligence: a guide to intelligent systems. Pearson Education, 2005.

$\endgroup$
1
  • $\begingroup$ Note that there are several other meanings "soft" can take in this context. e.g. Soft models as opposed to hard models which are based on first principles (SVM is soft in that respect, too). And AFAIK, fuzzy (as in fuzzy sets; meaning belongign to a degree) is not the same as uncertain. $\endgroup$ Feb 14 '16 at 15:16
1
$\begingroup$

I have received the following answer through a private communication:

====

Answer to your question is simple - if you calculate 386* 79703 you'll get the number which is extremely precise. This is not a soft computing.

If you put 1000 dollars into bank with interest of 3%, you can exactly know how much will you have after 4 years. This is not a soft computing.

If you are given equation of parabola y = 3x^2 + 2x - 3, and given x, you will exactly now what is the value of y. This is not a soft computing. And so on, and so on.

But whenever your results are not exact, because the data are not exact, assumptions are not exact, uncertainty is involved, i.e., whenever there is some stochastic part in your calculation and, despite that, you want to get some 'crisp', 'precise' answer, obtaining it, will be, and is, the soft computing.

So Negnevitsky's definition falls under the definition of soft computing, but so does the design of SVM too.

SVM model is obtained from some (experimental, empirical, etc, ...) datasets which is always "... uncertain, imprecise and incomplete ...", and such a stochastic, random i.e., uncertain, environment makes SVM a soft computing tool too.

VK

====

$\endgroup$
1
  • $\begingroup$ Incomplete, really? How does the SVM handle that kind of data? An MI-SVM should jump in at this point or you require further pre-processing but is it still softcomputing then? $\endgroup$
    – pAt84
    Feb 14 '16 at 14:45

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.