Adaptive neuro-fuzzy inference system (ANFIS) is a soft computing method, in which a given input–output data set is expressed in a fuzzy inference system (FIS).
What are the suitable applications for neural networks/ ANFIS?
Why do we use the adaptive neuro-fuzzy inference system?
When is it appropriate to use fuzzy logic instead of neural networks?
neural networks can be used for all those applications where there seems some relationship between input features and output that generally can not be described by a mathematical relationship. Also, they are used where you are ready to accept a slight percent of error and the application results are not critical.
Whereas Fuzzy logic is used for a subset of some specific applications that can be solved by ANN and thus are less useful. But this is my opinion and someone may disagree
if neural network is better than fuzzy logic, then why they mix them in ANFIS. I think that there are some applications where is better to use anfis instead ANN. I want to know what is the type of these applications.
I think my question is not clear to you. I want to know if there is any rules to select ANN or ANFIS for example: some people has build control system using ANN while others use ANFIS for the same control system. The results differ sometimes ANN is better than ANFIS or the opposite
@rufaida, there is no clear merit to use only ANN or ANFIS as both have advantages and disadvantages. People try out different variant and amalgamation/cross overs of techniques to become little different so that they can publish and then declare their own work as the best based on biased supporting arguments while leaving nonsupporting arguments.
You need to determine or choose a particular method based on your intended research based on literature survey as both or either technique might favor your work.
In my view, the only rule to select a method is a very good LITERATURE SURVEY of your area. Preferably, one should not be over ambitious to do something which has never been done before, e.g, the best papers is many high impact conferences in signal processing are still on a simple FILTER DESIGN which is a quite an old and heavily worked out
topic
Historically, ANFIS was introduced in the mid-nineties. At that time a clear advantage of ANFIS with respect to classical NNs was that priori knowledge given as fuzzy rules were able to be inserted within the network (rules such as: if temperature is hot then ...), thus providing high probability to converge toward good generalization. Note also that GG Towell in 1990-91 introduced a model to insert crisp rules into a multi layer perceptron. There have been a lot of papers on rule extraction from NNs in the last 15 years, however ANFIS has the merit to be one of the rare softwares allowing users to insert priori knowledge into a NN as rules.
- Badges
- Science topic