Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Added files via upload #19

Open
wants to merge 1 commit into
base: master
Choose a base branch
from
Open

Added files via upload #19

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
58 changes: 47 additions & 11 deletions evolve_text.py
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -1,4 +1,8 @@
"""
Anna Buchele

Evolutionary Algorithm Toolbox

Evolutionary algorithm, attempts to evolve a given message string.

Uses the DEAP (Distributed Evolutionary Algorithms in Python) framework,
Expand Down Expand Up @@ -92,8 +96,35 @@ def get_text(self):
# Genetic operators
#-----------------------------------------------------------------------------

# TODO: Implement levenshtein_distance function (see Day 9 in-class exercises)
# HINT: Now would be a great time to implement memoization if you haven't
def levenshtein_distance(s1,s2):
#calculates the edit distance between two strings

# Controls whether edit distance will take into account how "far away" a letter
# is from another letter:
# If True: A -> Z is a distance of 3, while A -> B is 1.
# If False: A -> Z is a distance of 1, as is A -> B.
ORDON = True

if not s1:
return len(s2)
if not s2:
return len(s1)
s1s = s1[1:]
s2s = s2[1:]
s10 = s1[0]
s20 = s2[0]
if s10 == s20:
return levenshtein_distance(s1s,s2s)
diffchar = (s10 != s20)
if ORDON:
orddist = abs((ord(s10)-65)-(ord(s20)-65))
ordlevdist = (orddist / 12.5) + diffchar
else:
ordlevdist = diffchar
lev1 = levenshtein_distance(s1s,s2) + 1
lev2 = levenshtein_distance(s1,s2s) + 1
lev3 = levenshtein_distance(s1s,s2s) + ordlevdist
return min(lev1,lev2,lev3)

def evaluate_text(message, goal_text, verbose=VERBOSE):
"""
Expand All @@ -119,15 +150,13 @@ def mutate_text(message, prob_ins=0.05, prob_del=0.05, prob_sub=0.05):
Substitution: Replace one character of the Message with a random
(legal) character
"""

if random.random() < prob_ins:
# TODO: Implement insertion-type mutation
pass

# TODO: Also implement deletion and substitution mutations
# HINT: Message objects inherit from list, so they also inherit
# useful list methods
# HINT: You probably want to use the VALID_CHARS global variable
message.insert(random.randint(0,len(message)), random.choice(VALID_CHARS))
if random.random() < prob_del:
del message[random.randint(0,len(message)-1)]
if random.random() < prob_sub:
message[random.randint(0,len(message)-1)] = random.choice(VALID_CHARS)

return (message, ) # Length 1 tuple, required by DEAP

Expand Down Expand Up @@ -197,9 +226,12 @@ def evolve_string(text):

if __name__ == "__main__":

# Open Save file
saveresults = open('resultstrue2.txt', 'w')

# Get goal message from command line (optional)
import sys
if len(sys.argv) == 1:
if len(sys.argv) == 0:
# Default goal of the evolutionary algorithm if not specified.
# Pretty much the opposite of http://xkcd.com/534
goal = "SKYNET IS NOW ONLINE"
Expand All @@ -216,3 +248,7 @@ def evolve_string(text):

# Run evolutionary algorithm
pop, log = evolve_string(goal)

# Save data to file, then close
saveresults.write(str(log))
saveresults.close()
100 changes: 100 additions & 0 deletions results.txt
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -0,0 +1,100 @@
Anna Buchele
Toolbox: Evolutionary Algorithms


For my experiment, I added to the levenshtein_distance function to make it account for distances between letters. I set it on a scale from 0-3, where 0 is the same letter and 3 is A -> Z. Each letter is 2/25 distance from the next or previous. I previously wrote it so that it simply added the letter distance to the levenshtein distance, but that didn't work well. I think a normalized 0-3 scale is better.

Regardless of whether the distance between letters was being applied, the first instance of the correct goal word (I used "HELLO") occurred in the 12th generation. When the distance between letters was not being calculated, the average distance from the correct phrase started lower and existed below 0.5 earlier. This is to be expected, as taking into account the distance between letters was expected to increase the average levenshtein distance in each instance. However, other than slight differences (+/- 1 generation), they both seemed to give about the same results.

Below is a sample of my data from both sets. Only the first 40 generations are included because after that, there isn't much change.

Without accounting for distance between letters:

gen nevals avg std min max
0 300 15.6867 7.03007 2 29
1 194 9.98 4.76581 3 27
2 191 6.46 2.43209 3 15
3 176 4.85333 0.889844 2 10
4 186 4.33 0.712577 2 7
5 178 3.84 0.643739 2 5
6 176 3.38333 0.704549 1 5
7 193 2.88333 0.665624 1 4
8 171 2.43 0.715379 1 4
9 199 1.92667 0.633474 1 4
10 190 1.52667 0.537856 1 3
11 180 1.14 0.383493 1 4
12 186 1.02333 0.171626 0 2
13 172 1.01667 0.172401 0 2
14 186 0.996667 0.191456 0 2
15 193 0.973333 0.243493 0 2
16 182 0.95 0.327872 0 2
17 178 0.836667 0.435877 0 2
18 178 0.563333 0.515741 0 2
19 186 0.193333 0.411447 0 2
20 198 0.0566667 0.245198 0 2
21 179 0.0333333 0.197203 0 2
22 166 0.0233333 0.15096 0 1
23 190 0.03 0.170587 0 1
24 173 0.02 0.14 0 1
25 201 0.0366667 0.204912 0 2
26 171 0.0166667 0.151841 0 2
27 185 0.05 0.246644 0 2
28 175 0.0433333 0.203606 0 1
29 182 0.02 0.16207 0 2
30 180 0.0333333 0.179505 0 1
31 162 0.0333333 0.179505 0 1
32 178 0.02 0.14 0 1
33 173 0.03 0.189121 0 2
34 194 0.03 0.170587 0 1
35 190 0.04 0.22745 0 2
36 184 0.0233333 0.15096 0 1
37 166 0.0233333 0.15096 0 1
38 181 0.0366667 0.204912 0 2
39 178 0.0233333 0.171626 0 2
40 175 0.0166667 0.128019 0 1

With accounting for distance between letters:

gen nevals avg std min max
0 300 16.6852 6.70496 3.44 29.8
1 194 11.2684 4.44317 3.08 27.56
2 191 8.06787 2.3173 3.8 15.88
3 176 6.40707 1.12643 3.24 11.88
4 186 5.62627 1.03029 2 8.44
5 178 4.94053 0.911634 1 7.12
6 176 4.28627 0.964488 1 6.8
7 193 3.60013 0.957399 1 6.8
8 171 2.85867 0.939931 1 6.08
9 199 2.12107 0.89776 1 5.12
10 190 1.45533 0.582642 1 3.24
11 180 1.12307 0.36973 1 3.8
12 186 1.03267 0.247127 0 3
13 172 1.0252 0.217892 0 3
14 186 1.00667 0.227493 0 3
15 193 1.002 0.266356 0 3
16 182 1.0052 0.293593 0 2.72
17 178 0.959067 0.340171 0 2.72
18 178 0.886 0.405023 0 2.24
19 186 0.707067 0.514254 0 3
20 198 0.4184 0.569241 0 2.96
21 179 0.102 0.356728 0 2.4
22 166 0.0272 0.183699 0 2
23 190 0.0349333 0.206755 0 2
24 173 0.0249333 0.18261 0 2
25 201 0.0398667 0.22678 0 2.16
26 171 0.0190667 0.180648 0 2.56
27 185 0.0578667 0.296523 0 2.88
28 175 0.0476 0.228294 0 1.72
29 182 0.0254667 0.214282 0 2.64
30 180 0.0390667 0.21849 0 2
31 162 0.0413333 0.231973 0 1.96
32 178 0.0248 0.178448 0 1.72
33 173 0.0326667 0.205042 0 2
34 194 0.0388 0.231179 0 2
35 190 0.0496 0.289816 0 2.56
36 184 0.0249333 0.163339 0 1.48
37 166 0.0254667 0.168276 0 1.64
38 181 0.0421333 0.2447 0 2.48
39 178 0.0298667 0.220409 0 2
40 175 0.0233333 0.190058 0 2