Genetic Algorithms Read Chapter Exercises - - PDF document

genetic algorithms read chapter exercises
SMART_READER_LITE
LIVE PREVIEW

Genetic Algorithms Read Chapter Exercises - - PDF document

Genetic Algorithms Read Chapter Exercises Ev olutionary computation Protot ypical GA An example GABIL Genetic Programming Individual


slide-1
SLIDE 1 Genetic Algorithms Read Chapter
  • Exercises
  • Ev
  • lutionary
computation
  • Protot
ypical GA
  • An
example GABIL
  • Genetic
Programming
  • Individual
learning and p
  • pulation
ev
  • lution
  • lecture
slides for textb
  • k
Machine L e arning T Mitc hell McGra w Hill
slide-2
SLIDE 2 Ev
  • luationary
Computation
  • Computational
pro cedures patterned after biological ev
  • lution
  • Searc
h pro cedure that probabilisti cal l y applies searc h
  • p
erators to set
  • f
p
  • in
ts in the searc h space
  • lecture
slides for textb
  • k
Machine L e arning T Mitc hell McGra w Hill
slide-3
SLIDE 3 Biological Ev
  • lution
Lamarc k and
  • thers
  • Sp
ecies transm ute
  • v
er time Darwin and W allace
  • Consisten
t heritable v ariation among individuals in p
  • pulation
  • Natural
selection
  • f
the ttest Mendel and genetics
  • A
mec hanism for inheriting traits
  • genot
yp e
  • phenot
yp e mapping
  • lecture
slides for textb
  • k
Machine L e arning T Mitc hell McGra w Hill
slide-4
SLIDE 4 GAF itness F itness thr eshol d p r
  • m
  • Initialize
P
  • p
random h yp
  • theses
  • Evaluate
for eac h h in P
  • compute
F itnessh
  • While
max h F itnessh
  • F
itness thr eshol d
  • Sele
ct Probabilistic al ly select
  • r
p mem b ers
  • f
P to add to P S
  • Pr
h i
  • F
itnessh i
  • P
p j
  • F
itnessh j
  • Cr
  • ssover
Probabilistic al l y select r p
  • pairs
  • f
h yp
  • theses
from P
  • F
  • r
eac h pair hh
  • h
  • i
pro duce t w
  • spring
b y applying the Crosso v er
  • p
erator Add all
  • spring
to P s
  • Mutate
In v ert a randomly selected bit in m
  • p
random mem b ers
  • f
P s
  • Up
date P
  • P
s
  • Evaluate
for eac h h in P
  • compute
F itnessh
  • Return
the h yp
  • thesis
from P that has the highest tness
  • lecture
slides for textb
  • k
Machine L e arning T Mitc hell McGra w Hill
slide-5
SLIDE 5 Represen ting Hyp
  • theses
Represen t O utl
  • ok
  • O
v er cast
  • R
ain
  • W
ind
  • S
tr
  • ng
  • b
y O utl
  • ok
W ind
  • Represen
t IF W ind
  • S
tr
  • ng
THEN P l ay T ennis
  • y
es b y O utl
  • ok
W ind P l ay T ennis
  • lecture
slides for textb
  • k
Machine L e arning T Mitc hell McGra w Hill
slide-6
SLIDE 6 Op erators for Genetic Algorithms

Single-point crossover:

11101001000 00001010101 11111000000 11101010101

Initial strings Crossover Mask Offspring Two-point crossover:

11101001000 00001010101 00111110000 11001011000 10011010011

Uniform crossover: Point mutation:

11101001000 00001010101 10001000100 11101001000 11101011000 00101000101 00001001000 01101011001

  • lecture
slides for textb
  • k
Machine L e arning T Mitc hell McGra w Hill
slide-7
SLIDE 7 Selecting Most Fit Hyp
  • theses
Fitness prop
  • rtionate
selection Prh i
  • F
itnessh i
  • P
p j
  • F
itnessh j
  • can
lead to cr
  • wding
T
  • urnamen
t selection
  • Pic
k h
  • h
  • at
random with uniform prob
  • With
probabilit y p select the more t Rank selection
  • Sort
all h yp
  • theses
b y tness
  • Prob
  • f
selection is prop
  • rtional
to rank
  • lecture
slides for textb
  • k
Machine L e arning T Mitc hell McGra w Hill
slide-8
SLIDE 8 GABIL DeJong et al
  • Learn
disjunctiv e set
  • f
prop
  • sitional
rules comp etitiv e with C Fitness F itnessh
  • cor
r ecth
  • Represen
tatio n IF a
  • T
a
  • F
THEN c
  • T
  • IF
a
  • T
THEN c
  • F
represen ted b y a
  • a
  • c
a
  • a
  • c
  • Genetic
  • p
erators
  • w
an t v ariable length rule sets
  • w
an t
  • nly
w ellformed bitstring h yp
  • theses
  • lecture
slides for textb
  • k
Machine L e arning T Mitc hell McGra w Hill
slide-9
SLIDE 9 Crosso v er with V ariableLength Bit strings Start with a
  • a
  • c
a
  • a
  • c
h
  • h
  • c
ho
  • se
crosso v er p
  • in
ts for h
  • eg
after bits
  • no
w restrict p
  • in
ts in h
  • to
those that pro duce bitstrings with w elldened seman tics eg h i h i h i if w e c ho
  • se
h i result is a
  • a
  • c
h
  • a
  • a
  • c
a
  • a
  • c
a
  • a
  • c
h
  • lecture
slides for textb
  • k
Machine L e arning T Mitc hell McGra w Hill
slide-10
SLIDE 10 GABIL Extensions Add new genetic
  • p
erators also applied probabilistic al l y
  • A
ddA lternative generalize constrain t
  • n
a i b y c hanging a
  • to
  • Dr
  • pCondition
generalize constrain t
  • n
a i b y c hanging ev ery
  • to
  • And
add new eld to bitstring to determine whether to allo w these a
  • a
  • c
a
  • a
  • c
AA D C
  • So
no w the learning strategy also ev
  • lv
es
  • lecture
slides for textb
  • k
Machine L e arning T Mitc hell McGra w Hill
slide-11
SLIDE 11 GABIL Results P erformance
  • f
GABIL comparable to sym b
  • lic
ruletree learning metho ds C IDR A Q Av erage p erformance
  • n
a set
  • f
  • syn
thetic problems
  • GABIL
without AA and D C
  • p
erators
  • accuracy
  • GABIL
with AA and D C
  • p
erators
  • accuracy
  • sym
b
  • lic
learning metho ds ranged from
  • to
  • lecture
slides for textb
  • k
Machine L e arning T Mitc hell McGra w Hill
slide-12
SLIDE 12 Sc hemas Ho w to c haracterize ev
  • lution
  • f
p
  • pulation
in GA Sc hema
  • string
con taining
  • dont
care
  • T
ypical sc hema
  • Instances
  • f
ab
  • v
e sc hema
  • Characterize
p
  • pulation
b y n um b er
  • f
instances represen ting eac h p
  • ssible
sc hema
  • ms
t
  • n
um b er
  • f
instances
  • f
sc hema s in p
  • p
at time t
  • lecture
slides for textb
  • k
Machine L e arning T Mitc hell McGra w Hill
slide-13
SLIDE 13 Consider Just Selection
  • f
t
  • a
v erage tness
  • f
p
  • p
at time t
  • ms
t
  • instances
  • f
sc hema s in p
  • p
at time t
  • us
t
  • a
v e tness
  • f
instances
  • f
s at time t Probabilit y
  • f
selecting h in
  • ne
selection step Pr h
  • f
h P n i f h i
  • f
h n
  • f
t Probabilt y
  • f
selecting an instance
  • f
s in
  • ne
step Pr h
  • s
  • X
hsp t f h n
  • f
t
  • us
t n
  • f
t ms t Exp ected n um b er
  • f
instances
  • f
s after n selections E ms t
  • us
t
  • f
t ms t
  • lecture
slides for textb
  • k
Machine L e arning T Mitc hell McGra w Hill
slide-14
SLIDE 14 Sc hema Theorem E ms t
  • u
s t
  • f
t ms t
  • B
B
  • p
c ds l
  • C
C A p m
  • s
  • ms
t
  • instances
  • f
sc hema s in p
  • p
at time t
  • f
t
  • a
v erage tness
  • f
p
  • p
at time t
  • us
t
  • a
v e tness
  • f
instances
  • f
s at time t
  • p
c
  • probabilit
y
  • f
single p
  • in
t crosso v er
  • p
erator
  • p
m
  • probabilit
y
  • f
m utation
  • p
erator
  • l
  • length
  • f
single bit strings
  • s
n um b er
  • f
dened non
  • bits
in s
  • ds
  • distance
b et w een leftmost righ tmost dened bits in s
  • lecture
slides for textb
  • k
Machine L e arning T Mitc hell McGra w Hill
slide-15
SLIDE 15 Genetic Programming P
  • pulation
  • f
programs represen ted b y trees sin x
  • r
x
  • y

^ sin x y 2 + x +

  • lecture
slides for textb
  • k
Machine L e arning T Mitc hell McGra w Hill
slide-16
SLIDE 16 Crosso v er

^ sin x y 2 + x + ^ sin x y 2 + x + sin x y + x + ^ sin x y 2 + x + ^ 2

  • lecture
slides for textb
  • k
Machine L e arning T Mitc hell McGra w Hill
slide-17
SLIDE 17 Blo c k Problem

u i v a n e r s l

Goal sp ell UNIVERSAL T erminals
  • CS
curren t stac k
  • name
  • f
the top blo c k
  • n
stac k
  • r
F
  • TB
top correct blo c k
  • name
  • f
topmost correct blo c k
  • n
stac k
  • NN
next necessary
  • name
  • f
the next blo c k needed ab
  • v
e TB in the stac k
  • lecture
slides for textb
  • k
Machine L e arning T Mitc hell McGra w Hill
slide-18
SLIDE 18 Primitiv e functions
  • MS
x mo v e to stac k if blo c k x is
  • n
the table mo v es x to the top
  • f
the stac k and returns the v alue T
  • Otherwise
do es nothing and returns the v alue F
  • MT
x mo v e to table if blo c k x is somewhere in the stac k mo v es the blo c k at the top
  • f
the stac k to the table and returns the v alue T
  • Otherwise
returns F
  • EQ
x y
  • equal
returns T if x equals y
  • and
returns F
  • therwise
  • NOT
x returns T if x
  • F
  • else
returns F
  • DU
x y
  • do
un til executes the expression x rep eatedly un til expression y returns the v alue T
  • lecture
slides for textb
  • k
Machine L e arning T Mitc hell McGra w Hill
slide-19
SLIDE 19 Learned Program T rained to t
  • test
problems Using p
  • pulation
  • f
  • programs
found this after
  • generations
EQ DU MT CSNOT CS DU MS NNNOT NN
  • lecture
slides for textb
  • k
Machine L e arning T Mitc hell McGra w Hill
slide-20
SLIDE 20 Genetic Programming More in teresting example design electronic lter circuits
  • Individuals
are programs that transform b egining circuit to nal circuit b y addingsubtracting comp
  • nen
ts and connections
  • Use
p
  • pulation
  • f
  • run
  • n
  • no
de parallel pro cessor
  • Disco
v ers circuits comp etitiv e with b est h uman designs
  • lecture
slides for textb
  • k
Machine L e arning T Mitc hell McGra w Hill
slide-21
SLIDE 21 GP for Classifying Images T eller and V eloso Fitness based
  • n
co v erage and accuracy Represen tatio n
  • Primitiv
es include Add Sub Mult Div Not Max Min Read W rite IfThenElse Either Pixel Least Most Av e V ariance Dierence Mini Library
  • Mini
refers to a lo cal subroutine that is separately coev
  • lv
ed
  • Library
refers to a global library subroutine ev
  • lv
ed b y selecting the most useful minis Genetic
  • p
erators
  • Crosso
v er m utation
  • Create
mating p
  • ls
and use rank prop
  • rtionate
repro duction
  • lecture
slides for textb
  • k
Machine L e arning T Mitc hell McGra w Hill
slide-22
SLIDE 22 Biolog i cal Ev
  • lution
Lamark th cen tury
  • Believ
ed individual genetic mak eup w as altered b y lifeti me exp erience
  • But
curren t evidence con tradicts this view What is the impact
  • f
individual learning
  • n
p
  • pulation
ev
  • lution
  • lecture
slides for textb
  • k
Machine L e arning T Mitc hell McGra w Hill
slide-23
SLIDE 23 Baldwin Eect Assume
  • Individual
learning has no direct in!uence
  • n
individual DNA
  • But
abilit y to learn reduces need to hard wire traits in DNA Then
  • Abilit
y
  • f
individual s to learn will supp
  • rt
more div erse gene p
  • l
  • Because
learning allo ws individual s with v arious hard wired traits to b e successful
  • More
div erse gene p
  • l
will supp
  • rt
faster ev
  • lution
  • f
gene p
  • l
  • individual
learning indirectl y increases rate
  • f
ev
  • lution
  • lecture
slides for textb
  • k
Machine L e arning T Mitc hell McGra w Hill
slide-24
SLIDE 24 Baldwin Eect Plausible example
  • New
predator app ears in en vironmen t
  • Individuals
who can learn to a v
  • id
it will b e selected
  • Increase
in learning individuals will supp
  • rt
more div erse gene p
  • l
  • resulting
in faster ev
  • lution
  • p
  • ssibly
resulting in new nonlearned traits suc h as instin tiv e fear
  • f
predator
  • lecture
slides for textb
  • k
Machine L e arning T Mitc hell McGra w Hill
slide-25
SLIDE 25 Computer Exp erimen ts
  • n
Baldwin Eect Hin ton and No wlan Ev
  • lv
e simple neural net w
  • rks
  • Some
net w
  • rk
w eigh ts xed during lifeti me
  • thers
trainable
  • Genetic
mak eup determines whic h are xed and their w eigh t v alues Results
  • With
no individual learning p
  • pulation
failed to impro v e
  • v
er time
  • When
individual learning allo w ed
  • Early
generations p
  • pulation
con tained man y individual s with man y trainable w eigh ts
  • Later
generations higher tness while n um b er
  • f
trainable w eigh ts decreased
  • lecture
slides for textb
  • k
Machine L e arning T Mitc hell McGra w Hill
slide-26
SLIDE 26 Summary Ev
  • lutionary
Program ming
  • Conduct
randomized parallel hillcl i m bing searc h through H
  • Approac
h learning as
  • ptimization
problem
  • ptimize
tness
  • Nice
feature ev aluation
  • f
Fitness can b e v ery indirect
  • consider
learning rule set for m ultistep decision making
  • no
issue
  • f
assigning creditblame to indiv steps
  • lecture
slides for textb
  • k
Machine L e arning T Mitc hell McGra w Hill