joshua/nominal-lstar
1
Fork 0
mirror of https://github.com/Jaxan/nominal-lstar.git synced 2025-04-27 14:47:45 +02:00
Code Issues Releases Activity
28 commits 1 branch 0 tags 214 KiB
Commit graph

28 commits

This branch
This branch
All branches
Author SHA1 Message Date
Joshua Moerman
6337defa0e Adds a counter to the teacher (bery hacky) 2016-06-30 10:58:32 +02:00
Joshua Moerman
28145815c6 Improves performance by using more simplify 2016-06-28 12:08:10 +02:00
Joshua Moerman
8946bc941e Adds the examples from the paper 2016-06-28 10:35:12 +02:00
Joshua Moerman
a10c1cb84a Simplifies my own naive take on NL* 2016-06-27 14:51:36 +02:00
Joshua Moerman
8998042874 Adds a teacher for NFAs (bounded to a depth). 
Also includes a variation of the up-to technique found in the 'hacking
non-determinism' paper by Bonchi and Pous.
 2016-06-27 14:50:57 +02:00
Michał Szynwelski
a9b3738cd3 Optimization of consistencyTestDirect.candidates function. 2016-06-26 21:06:18 +02:00
Joshua Moerman
ec5fe35c79 Simplified consistency in Angluin 2016-06-24 13:57:17 +02:00
Joshua Moerman
7109eb1ec6 Improved the Bollig implementation 2016-06-24 10:41:00 +02:00
Joshua Moerman
e65a099948 Fixes a bug in my Bollig implementation 2016-06-23 17:16:02 +02:00
Joshua Moerman
9ac25c4d9b Refactors code to be more modular. 
Now all three variations: Angluin, Bollig and homebrew NL* are using the
same framework. I did not extensively test the refactor.
 2016-06-23 16:20:33 +02:00
Joshua Moerman
f24ed31ac8 Implemented NL* by Bollig et al (verbatim) 2016-06-23 14:50:30 +02:00
Joshua Moerman
25d47a3550 Refactors the IO monad out. Made functions pure again. 2016-06-23 11:16:47 +02:00
Joshua Moerman
f5f88a2ef7 Adds the NFA example by Bollig et al 2016-06-23 11:16:01 +02:00
Joshua Moerman
43c85612bb Refactors the code to use simpler types. 2016-06-22 17:30:26 +02:00
Joshua Moerman
9ee755117e Adds ways to control strictness. 
Increases efficiency by 1.5x (if used in observation table)
 2016-06-22 17:27:35 +02:00
Joshua Moerman
600f2fe1e8 Simplification in observation table 2016-06-20 16:40:18 +02:00
Joshua Moerman
24e2c1ff88 Hacked NL* together. 
Not complete yet, but it already works for the prototypical NFA: all
words with duplicate atoms.
 2016-06-20 16:40:08 +02:00
Joshua Moerman
2e901070d9 Code now compiles on windows as well. (Doesn't run well yet) 2016-06-20 09:54:32 +02:00
Michał Szynwelski
fbcba8c580 Optimization of incompleteness function. 2016-06-17 14:12:09 +02:00
Joshua Moerman
c3c6645039 Merge branch 'master' of github.com:UCL-PPLV/nominal-learning 2016-06-17 08:58:10 +02:00
Joshua Moerman
e403d7dbd6 Adds definition for inconsistency by Bartek 2016-06-17 08:57:15 +02:00
Michał Szynwelski
e683a4e384 Optimization of row function. 2016-06-14 00:09:16 +02:00
Joshua Moerman
cb09b2d306 Code: Fixes some warnings and made sure everything compiles out of the box (ghc 7.10) 2016-05-10 11:57:27 +02:00
Joshua Moerman
afd27369a1 Code: Teacher now uses bisimulation to find counter examples 2016-04-29 11:15:42 +01:00
Joshua Moerman
7c0cdcf7c3 Code: Implemented adding CEs as columns (Rivest Schapire variant?) 2016-04-28 17:47:51 +01:00
Joshua Moerman
004e71ccd9 Code: some cleanup 2016-04-28 17:26:16 +01:00
Michał Szynwelski
440e5ef854 Formula simplifications + constants. 2016-04-24 23:15:07 +02:00
Joshua Moerman
d52f991a74 Code: Angluin's algorithm with NLambda 2016-04-14 17:36:03 +01:00