W3C home > Mailing lists > Public > semantic-web@w3.org > July 2019

Re: neural networks being purported as KR?

From: Stephen D. Williams <sdw@lig.net>
Date: Fri, 26 Jul 2019 13:50:31 -0700
To: semantic-web@w3.org
Message-ID: <c4eb5734-056a-969b-92da-3c3268de4798@lig.net>
I'm assuming that "decision tree" refers to something like graph reasoning.  When I studied Bayesian graph reasoning & Hidden 
Markov, it seemed clear, logical, and a pure form of what we do for reasoning.  And then you get to the sticky problem of actually 
training a network that implements a lot of interesting logic, reasoning, knowledge, etc.  My understanding of the situation was 
that training those networks for any interesting or reasonably complex problem was still a mostly unsolved problem.

I view ML as generally solving the same problems that those methods solve (but also more), but as having a completely practical and 
pragmatic training mechanism for some problems and promising avenues for most other problems.  ML is about equal to graph reasoning, 
but with far better training techniques.  I think it is possible to prove that in some sense.

Stephen

On 7/26/19 9:24 AM, Pascal Hitzler wrote:
> Yes. Both are mathematical and computational models for "cognition" (for lack of a better term). They just have rather different 
> features.
>
> The deep divide between both paradigms (including in terms of community) is very unfortunate, but due to many reasons, some of 
> them historic.
>
> Some resources to look into a more unified perspective:
>
> https://arxiv.org/abs/1711.03902 - this is from a cognitive science perspective.
>
> http://www.semantic-web-journal.net/content/neural-symbolic-integration-and-semantic-web - with more of a focus on semantic web 
> issues
>
> http://neural-symbolic.org/ - a community which has kept an integrated view alive during the AI winter. Right now, it's bubbling 
> up larger-scale.
>
> P.
>
>
> On 7/26/2019 10:59 AM, Krzysztof Janowicz wrote:
>> On 7/26/19 5:00 PM, Agnieszka Ławrynowicz wrote:
>>> Hi All,
>>>
>>> Of course deep neural networks may be seen as form of knowledge representation in my opinion, more precisely they are 
>>> sub-symbolic or connectionist representations versus symbolic representations which are the standard in Semantic Web.
>>
>>
>> Same here (and leaving aside that the 'knowledge' in knowledge representation is a problematic term anyway).
>>
>>
>>> Though it is not „latest KR”, but have been there for a long time under exactly the above name (sub-symbolic representations).
>>>
>>> Best Regards and cheers,
>>> Agnieszka
>>>
>>>
>>>
>>>> Wiadomość napisana przez Diogo FC Patrao <djogopatrao@gmail.com <mailto:djogopatrao@gmail.com>> w dniu 26.07.2019, o godz. 16:40:
>>>>
>>>> Hi Paola
>>>>
>>>> I'd say a NN is not as "knowledgy" as a decision tree. I would argue that NN is a mathematical model that compiles previous 
>>>> data representing cause/consequences, so it's the same type of knowledge as, say, a logarithm table, versus the type of 
>>>> knowledge the infinte sum formula for evaluating logarithms would represent.
>>>>
>>>> They certainly don't look the same thing to me.
>>>>
>>>> Cheers,
>>>>
>>>> dfcp
>>>>
>>>> -- 
>>>> diogo patrão
>>>>
>>>>
>>>>
>>>>
>>>> On Thu, Jul 25, 2019 at 11:58 PM Paola Di Maio <paola.dimaio@gmail.com <mailto:paola.dimaio@gmail.com>> wrote:
>>>>
>>>>     Sorry to bang on this topic, but its the task at hand at the moment
>>>>
>>>>     I just found an article, which is good scientific survey then     purports NN as a type of KR
>>>>     (casually sneaks in NN as the latest KR)
>>>>
>>>>     This is published in a Springer peer reviewed publication and my
>>>>     makes all of my hairs stand up on my head
>>>>
>>>>     This is the kind of rubbish that without further qualification is
>>>>     being passed down
>>>>     as the latest research, and  which the future generations of AI
>>>>     scientists are being fed-
>>>>
>>>>     wonder if anyone else has a problem with this proposition
>>>>     (sign of the times?)
>>>>     I am doing my best within my means to identify and contain this peril
>>>>
>>>>     Article
>>>> https://link-springer-com.nls.idm.oclc.org/article/10.1007/s00170-018-2433-8
>>>> <https://urldefense.proofpoint.com/v2/url?u=https-3A__link-2Dspringer-2Dcom.nls.idm.oclc.org_article_10.1007_s00170-2D018-2D2433-2D8&d=DwMDaQ&c=3buyMx9JlH1z22L_G5pM28wz_Ru6WjhVHwo-vpeS0Gk&r=TpLLn6m0QS9xFWETRsVn6EgCZn90oD7nTZw4u7dKTkE&m=-OdcX_2BApaWb3B2Sm-OQqhYaSB9woE4DPmc5xD1Xjg&s=AwOZ0y96WXrVfkqtN4AIzI3du90yzPxz1lpf3hn2HUo&e=>
>>>>
>>>>
>>>>     A survey of knowledge representation methods and applications in
>>>>     machining process planning
>>>>
>>>>     The machining process is the act of preparing the detailed
>>>>     operating instructions for changing an engineering design into an
>>>>     end product, which involves the removal of material from the
>>>>     part. Today, machining ...
>>>>
>>>>     Xiuling Li, Shusheng Zhang, Rui Huang… in The International
>>>>     Journal of Advanced Manu… (2018)
>>>>
>>>>
>>>>
>>>
>>
>> -- 
>> Krzysztof Janowicz
>>
>> Geography Department, University of California, Santa Barbara
>> 4830 Ellison Hall, Santa Barbara, CA 93106-4060
>>
>> Email:jano@geog.ucsb.edu
>> Webpage:http://geog.ucsb.edu/~jano/
>> Semantic Web Journal:http://www.semantic-web-journal.net
>>
>

-- 
Stephen D. Williams sdw@lig.net stephendwilliams@gmail.com
V:650-450-UNIX (8649) V:866.SDW.UNIX V:703.371.9362 F:703.995.0407
Hangouts:stephendwilliams@gmail.com AIM:sdw Skype:StephenDWilliams
Personal: http://sdw.st facebook.com/sdwlig twitter.com/scienteer
LinkedIn: http://sdw.st/in Resume: http://sdw.st/resume
Received on Friday, 26 July 2019 20:50:59 UTC

This archive was generated by hypermail 2.4.0 : Thursday, 24 March 2022 20:42:07 UTC