home *** CD-ROM | disk | FTP | other *** search
- #
- #
- #---- noisyxor.lf
- #
- #
- #---- This is an lf example that tries to learn the XOR
- #---- function of two boolean inputs. The training is done
- #---- with contradictory data. For each pair of values of the
- #---- input, for example A = 1, B = 0, there are four correct
- #---- training samples and one incorrect sample in the set.
- #---- The optimal decision here is to learn the XOR, but
- #---- that may not always happen. We train for 30 epochs.
- #
- #---- See example3.lf for explanations of the instructions.
-
- tree
-
- size = 15
-
- max epochs = 30
-
- save tree to "noisyxor.tre"
-
- function
-
- domain dimension = 2
-
- codomain dimension = 1
-
- coding = 1:1 1:1 1:1
-
- quantization = 2 2 2
-
- save coding to "noisyxor.cod"
-
- training set size = 20
-
- training set =
-
- # A B A xor B (right 4/5 = 80% of the time)
- 1 1 0
- 1 1 0
- 1 1 0
- 1 1 0
- # next one is noisy
- 1 1 1
-
- 1 0 1
- 1 0 1
- 1 0 1
- 1 0 1
- # next one is noisy
- 1 0 0
-
- 0 1 1
- 0 1 1
- 0 1 1
- 0 1 1
- # next one is noisy
- 0 1 0
-
- 0 0 0
- 0 0 0
- 0 0 0
- 0 0 0
- # next one is noisy
- 0 0 1
-
-
- #---- We will test on thesame 20 vectors to see if we can get close to 80% right
- test set size = 20
- test set =
-
- 1 1 0
- 1 1 0
- 1 1 0
- 1 1 0
- # next one is noisy - if ALN learns XOR, then it should get this one wrong
- 1 1 1
-
- 1 0 1
- 1 0 1
- 1 0 1
- 1 0 1
- # next one is noisy - if ALN learns XOR, then it should get this one wrong
- 1 0 0
-
- 0 1 1
- 0 1 1
- 0 1 1
- 0 1 1
- # next one is noisy - if ALN learns XOR, then it should get this one wrong
- 0 1 0
-
- 0 0 0
- 0 0 0
- 0 0 0
- 0 0 0
- # next one is noisy - if ALN learns XOR, then it should get this one wrong
- 0 0 1
-
-
- #BEST POSSIBLE ERROR HISTOGRAM
- #0 errors 16
- #1 errors 4
-