Il giorno 18/apr/2011, alle ore 21.43, c. ha scritto:
>
>>>>>>>>> On 15 Apr 2011, at 11:32, L. Markowsky wrote:
>>>>>>>>>
>>>>>>>>> I'd like to register as a developer. My SourceForge username is
>>>>>>>>> lmarkov.
>>>>>>>>>
>>>>>>>>> I've started writing a package called fuzzy-logic-toolkit that
>>>>>>>>> implements many of the functions in the MATLAB Fuzzy Logic Toolbox.
>>>>>>>>> The fuzzy-logic-toolkit is currently usable from the commandline, and
>>>>>>>>> my next task is to add the GUIs. It is currently available on
>>>>>>>>> SourceForge at octave-fuzzy.sf.net and sf.net/projects/octave-fuzzy.
>>>>>>>>> The tarball is attached.
>>>>>>>>>
>>>>>>>>> Thank you.
>>>>>>>>>
>>>>>>>>>
>>>>>>>> Hi,
>>>>>>>>
>>>>>>>> Thanks for your contribution to Octave Forge!
>>>>>>>>
>>>>>>>> Octave-forge already has a package for that topic, called <fl-core
>>>>>>>> can you have a look to see if there is much overlap between the two
>>>>>>>> packages and whether they can made into
>>>>>>>> one single package?
>>>>>>>>
>>>>>>>>
>>>>>>>> I am not an expert in fuzzy logic, so I CC the developers of fl-core,
>>>>>>>> Gianvito Pio and Piero Molino
>>>>>>>> who might give better comments on this.
>>>>>>>>
>>>>>>>> c.
>>>>>>>
>>>>>>> Hi,
>>>>>>>
>>>>>>> Yes, I've seen fl-core, and I think that the two packages do not
>>>>>>> overlap. My package is meant to be an open-source equivalent of the
>>>>>>> MATLAB Fuzzy Logic Toolbox, and fl-core provides fuzzzy set operations
>>>>>>> and function composition. Also, combining them into one package would
>>>>>>> make my package incompatible with MATLAB's toolbox -- and this is the
>>>>>>> primary goal.
>>>>>>>
>>>>>>> Thanks, L.
>>>>>>
>>>>>> Hello,
>>>>>>
>>>>>> this mail is just to let the contributor know what out package is about
>>>>>> so he can decide what to do about it: out package contains the basic
>>>>>> operators useful for any fuzzy logic application. In particular we
>>>>>> developed a c++ native implementation of the relation composition
>>>>>> operator that's takes advantage multithread (if more than one core/cpu
>>>>>> is detected) and sparseness of data (if your input matrices of the
>>>>>> relations are sparse). It has a basic st of t-norms and s-norms to work
>>>>>> with (min, max, product, probabilistic sum) in an extremely efficient
>>>>>> way, but you can use a custom t-norm and a custom s-norm (but in this
>>>>>> case the performances drop). It's a really fast implementation, and we
>>>>>> built abouve it the basic fuzzy set theory operations: union,
>>>>>> intersection, complement and cartesian product.
>>>>>>
>>>>>> If you are building as i suppose a fuzzy logic toolbax similar to the
>>>>>> one in matlab, you probably implemented the above operators to deal with
>>>>>> linguistic variables and fuzzy rules for fuzzy inference systems. if so
>>>>>> you for sure had the need to compute compositions of relations and also
>>>>>> the other fuzzy set theory operations so you may consider using our
>>>>>> package for that and put it as dependency. Otherwise we may consider
>>>>>> merging both our and your package, but i would prefer not, as in our
>>>>>> intention the proposed operatos are the basic operators for fuzzy logic
>>>>>> and being used to construct other levels of fuzzy systems above it is
>>>>>> how is was meant to be used.
>>>>>>
>>>>>> I hope this email will help you,
>>>>>> Piero
>>>>>
>>>>> Hello,
>>>>>
>>>>> Thanks for your response. You're correct -- my package is a usable subset
>>>>> of the functions defined by Matlab's toolbox. Fuzzy inference systems can
>>>>> be built, modified, and evaluated from the commandline and read/written
>>>>> to/from .fis files. It wasn't necessary to implement any of the functions
>>>>> you mentioned (computing compositions of relations and set theory ops),
>>>>> however, so there's no need to make my package dependent on yours. Like
>>>>> you, I also prefer not to merge the packages since their principle
>>>>> purposes are distinct.
>>>>>
>>>>> Thanks,
>>>>> L.
>>>>
>>>> Hello,
>>>>
>>>> you're welcome. I think that there could be a problem from he octave
>>>> package system: this way there would be duplicated functions. I suggest
>>>> you to try the functions in our package, see if you have significant
>>>> decrease in performance (i don't think so because we did a deep
>>>> optimization work) and then decide what to do. If the performance are
>>>> better or equal you can rely on our package as dependency, otherwise
>>>> ignore it.
>>>>
>>>> Piero
>>>>
>>> Il giorno 17/apr/2011, alle ore 20.50, L. Markowsky ha scritto:
>>>
>>> Hello,
>>>
>>> Which functions are you referring to? As far as I can tell, there isn't any
>>> duplication.
>>>
>>> Thanks,
>>> L.
>>>
>> On 17 Apr 2011, at 20:59, w4nderlust wrote:
>> Hello,
>>
>> sorry i misread, you told you had no need to implement fuzzy set theory
>> operators, i understood the opposite :). So i don't know, let's see what
>> Carlo thinks about it. I think that maybe we can rename our package as
>> "fuzzy set theory" and your as "fuzzy inference system" or whatever. Anyway
>> it sounds strange to me that you didnt implement the fuzzy set theory
>> operators to build something that conceptually is on top of them. How have
>> you implemented for example the selection of rules to apply if not by
>> min-maxing the conditions? I'm just curious about it :)
>>
>> Piero
>
>
> Hi,
>
> First of all, please avoid top-posting, and remember to keep the mailing list
> address in CC,
> as other list members might want to join the discussion.
>
> As I said, I have no knowledge whatsoever about fuzzy logic so I cannot
> comment about the functionality of the contributed code,
> therefore I would be very grateful if anyone on the list with the required
> knowledge would jump in and write a brief comment.
>
> If this is not the case in, say, a couple of days, I will just look at the
> usual coding style issues and then add lmarkov to the OF developers, so that
> he can commit the code.
>
> As for the code duplication issue, it would be great if some of the code in
> fl-core could be reused in fuzzy-logic, so I would encourage
> Piero and Gianvito to check more in depth if this is possible, if it is not I
> have no objection to keeping the two distinct packages on OF.
>
> c.
>
Hi,
I'll check as soon as possible, but i don't think it would be possible as L.
said that he didn't use fuzzy set theory operators in his work, and our package
is mainly about fuzzy set theory operators that should be used for building
fuzzy logic operators that are on top of them, so i don't know how he
implemented it. To make it clearer i make an analogy: our package is to fuzzy
logic what crisp union, crisp complement and crisp intersection are to boolean
logic (and also any composition of them), so if you are making a prolog like
system for building inference systems, probably you have to build on top of
these operators. Anyway I'll check and let you know. If there's nothing to be
reused, then probably i think renaming fl-core to fuzzy-set-theory (or
something like that) would be better to avoid confusion for the users.
Piero
------------------------------------------------------------------------------
Benefiting from Server Virtualization: Beyond Initial Workload
Consolidation -- Increasing the use of server virtualization is a top
priority.Virtualization can reduce costs, simplify management, and improve
application availability and disaster protection. Learn more about boosting
the value of server virtualization. http://p.sf.net/sfu/vmware-sfdev2dev
_______________________________________________
Octave-dev mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/octave-dev
