Re: Tha('c ma('c: Fuzzy logic

[Thanh Tung Truong <tung.truong@e-technik.tu-chemnitz.de>]

Hi Anh AiViet,

In the discussion about fuzzy hardware I am a novice, too. Still, I dare to 
say that fuzzy hardware exists. However, the undergoing physical principle is 
not neccessary different from that of non-fuzzy hardware. It could be just 
the architecture that gives fuzzy hardware ist name. There is again a problem 
of definition here. To me the term fuzzy hardware is used in the sense of 
dedicated fuzzy hardware which is characterized by:
1) electronic circuits carrying out fuzzy functional primitives such as 
fuzzyfication, implication, composition and defuzzyfication,
2) specific architecture that is adapted to fuzzy algorithms and allows 
parallel signal processing.

As to fuzzy functional primitives, I think you are not quite on the wrong 
path when trying to build a „pseudo fuzzy hardware“ from conventional 
circuits. But you may need more than just conventional binary logic gates. I 
can think of two alternatives for fuzzy functional primitives:
A) analog circuit, i.e. integration of conventional analog circuits for the 
undergoing operations Max, Min, Division, Multiplication, Addition ...
B) discrete circuit, i.e. an extension of combinatory and probably also 
sequential Boolean logic circuit  that allow to handle more than two 
voltages. As far as I can remember there are people working on fuzzy flip 
flop that must be a fuzzy extension of Boolean logic flip flop. 

The discrete circuit is obviously close to the conventional digital concept 
that has several advantages such as accuracy and programability.

As to architecture, I am not informed well enough at this time to give you an 
idea, sorry.

I wish I had some time to make some recherche about this issue. Maybe I can 
do that one day.

Cheers,
Thanh-Tung Truong

You wrote:
> 
> Anh Tung,
>   Your article was very interesting. However, I still have the unanswer
> question: Does fuzzy hardware exist at all as a novice ?
>   I thought that there is only a fuzzy interface built on the non-fuzzy
> one that simulates the fuzzy logics as I did. As far as I understand from
> your article, you have asserted that the fuzzy hardware does exist.
> If that is true, what is the undergoing physical principle. ( Like
> digital
> devices are based on high and low voltages and gates (registers)...).
> And how do the fuzzy processors work?
>   Sorry about my ignorance. I am clear about other things.
> Thanks
> Aiviet