search

Balletie / asml-seminar-report

Bachelor Seminar report on active state machine learning
0 stars 0 forks source link

[RDY] Integrating variants, mealy and w-method (#6, #8, #9) #16

Closed gfokkema closed 8 years ago

gfokkema commented 8 years ago

As the title says, make it one story instead of 3 stories by different people.

Links for everyone's convenience:

6 Variants

9 Mealy

8 W-method

Please note that this PR is based on #13, and therefore also includes (parts of) #7, #11 and #12. Some more recent commits are not included in #13, notably the latest commits by @alangerak.

main.pdf

gfokkema commented 8 years ago

Now that everything's a whole, the section about W-method seems out of place. Shouldn't this be moved to variants, possibly replacing / merging with the section 'No EQUIV(M) queries?'

@StefanBoodt I know this is almost exactly what you've commented on the variants PR, but I'd like to remind you that the variants PR was opened a few days prior to the W-method PR ;p

alangerak commented 8 years ago

@gfokkema I actually agree. I also made a comment that is was strange that only the W-method was discussed. The "No EQUIV(M)" section is actually perfect for the W-Method (or other methods that are mentionend in other papers)

StefanBoodt commented 8 years ago

I think it is best to merge the W-method and NO EQUIV pieces as well. I think the short description is necessary as well as the lower bound of the method. Of course the fact that this method provides a guarantee to a certain degree must also be mentioned. The NO EQUIV is indeed a proper place to discuss the W-method.

@gfokkema I'd like to remind you that I opened the PR when the piece was finished, not when I was started writing, as was done in the Variants PR.

@alangerak I want to remind you that the W-method is the method to use to validate the FSM. Generating a long path is a different variant (No resets) and random sampling does not prove anything.

StefanBoodt commented 8 years ago

After checking we now have 6908 words in the pdf linked in the description of this PR. Since we have a 7000 words limit, shouldn't we be cautious about adding any additional words? I simply wanted to state the number of words.

Balletie commented 8 years ago

Yes we should watch out, thanks for checking @StefanBoodt. There's probably some duplicate explanations here and there. I will try to remove some in my PR #15.

gfokkema commented 8 years ago

Latest version of the report: main.pdf