Post on 08-Oct-2020
transcript
rocha@indiana.eduhttp://informatics.indiana.edu/rocha/i-bic
biologicallyInspired
computing
INDIANAUNIVERSITY
Informatics luis rocha 2015
biologically-inspired computinglecture 18
biologicallyInspired
computing
rocha@indiana.eduhttp://informatics.indiana.edu/rocha/i-bic
INDIANAUNIVERSITY
Informatics luis rocha 2015
course outlook
Assignments: 35% Students will complete 4/5 assignments based on algorithms
presented in class Lab meets in I1 (West) 109 on Lab Wednesdays
Lab 0 : January 14th (completed) Introduction to Python (No Assignment)
Lab 1 : January 28th
Measuring Information (Assignment 1) Graded
Lab 2 : February 11th
L-Systems (Assignment 2) Graded
Lab 3: March 25th
Cellular Automata & Boolean Networks (Assignment 3) Graded
Lab 4: April 8th
Genetic Algorithms (Assignment 4) Due: April 22nd
Lab 5: April 22nd
Ant Clustering Algorithm (Assignment 5) Due May 4th
Sections I485/H400
biologicallyInspired
computing
rocha@indiana.eduhttp://informatics.indiana.edu/rocha/i-bic
INDIANAUNIVERSITY
Informatics luis rocha 2015
Readings until now
Class Book Nunes de Castro, Leandro [2006]. Fundamentals of Natural
Computing: Basic Concepts, Algorithms, and Applications. Chapman & Hall. Chapters 1, 2, 3, 7.1-7.5, 8.1-8.2, 8.3.10
Lecture notes Chapter 1: “What is Life?” Chapter 2: “The Logical Mechanisms of Life” Chapter 3: “Formalizing and Modeling the World” Chapter 4: “Self-Organization and Emergent Complex
Behavior” Chapter 5: “Reality is Stranger than Fiction”
posted online @ http://informatics.indiana.edu/rocha/i-bic Other materials
Flake’s [1998], The Computational Beauty of Nature. MIT Press Chapters 20
biologicallyInspired
computing
rocha@indiana.eduhttp://informatics.indiana.edu/rocha/i-bic
INDIANAUNIVERSITY
Informatics luis rocha 2015
final project schedule
Projects Due by May 4th in Oncourse
ALIFE 15 (14) Actual conference due date: 2016 http://blogs.cornell.edu/alife14nyc/
8 pages (LNCS proceedings format) http://www.springer.com/computer/lncs?SGWI
D=0-164-6-793341-0 Preliminary ideas overdue!
Individual or group With very definite tasks assigned per
member of group
ALIFE 15
rocha@indiana.eduhttp://informatics.indiana.edu/rocha/i-bic
biologicallyInspired
computing
INDIANAUNIVERSITY
Informatics luis rocha 2015
Kyle Nealyhighlights
Lab 3: CA & BN
Matthew Remmel
Darlan Farias
rocha@indiana.eduhttp://informatics.indiana.edu/rocha/i-bic
biologicallyInspired
computing
INDIANAUNIVERSITY
Informatics luis rocha 2015
highlightsLab 3: CA & BN
Jonathan Stout Rafael Paiva
Lorander Saggu
biologicallyInspired
computing
rocha@indiana.eduhttp://informatics.indiana.edu/rocha/i-bic
INDIANAUNIVERSITY
Informatics luis rocha 2015
genetic algorithms
1) Generate Random population of bit-strings
2) Evaluate Fitness Function for each decoded solution
3) Reproduce next generation Selection by fitness Variation
crossover and mutation Fill new population
4) Go back to 2) until stop criteria is met Desired fitness Specified number of
generations Convergence
Lack of variability in population and/or fitness Tends to a peak
The workings
f(x1)
f(x2)
f(x3)
0 1 1 1 0 1 0 0 1 0
0 1 1 1 1 0 0 1 1 1
1 0 0 1 0 1 0 1 1 0
…
f(xi)
Parents0 1 1 1 0 1 0 0 1 0
0 1 0 1 1 1 0 1 0 1
0 1 1 1 0 1 0 0 1 0
0 1 0 1 1 1 0 1 0 1
0 1 0 1 0 1 0 0 1 0
0 1 1 1 1 1 0 1 0 1
0 1 0 0 0 1 0 0 1 1
0 1 1 1 1 0 0 1 0 1
0 0 1 1 0 1 0 0 1 0
0 1 0 1 0 1 0 1 0 1
rocha@indiana.eduhttp://informatics.indiana.edu/rocha/i-bic
biologicallyInspired
computing
INDIANAUNIVERSITY
Informatics luis rocha 2015
computational evolutionartificial genotype/phenotype mapping
x x x1 2 np
φCode:
!!!
S S np21 S
Selection
Variation
Genotype
Phenotype
Traditional Genetic Algorithm
011001
code
GenotypeDNA
RNAtranscription
translation(code)
amino acidchains
development
phenotypeorganismenvironmental
ramifications
Inhe
rited
va
riatio
n
Search algorithms based on the mechanics of Natural Selection
Based on distinction between a machine and a description of a machineSolution alternatives for optimization problems
biologicallyInspired
computing
rocha@indiana.eduhttp://informatics.indiana.edu/rocha/i-bic
INDIANAUNIVERSITY
Informatics luis rocha 2015
Types of encoding
Binary encodings Typically fixed-length
Many-letter encoding Larger alphabet (e.g. graph-generation grammars)
Real-valued encodings Genes take real values
Tree Encodings Genetic programming
Indirect Encodings Modeling Phenotype development or post-
transcription processes L-Systems, Dynamical systems, evolutionary robotics
rocha@indiana.eduhttp://informatics.indiana.edu/rocha/i-bic
biologicallyInspired
computing
INDIANAUNIVERSITY
Informatics luis rocha 2015
Cellular automatahomogenous lattice of state-determined systems
xx-1 x+1
Cellular Automata
xt
Density Task (a.k.a majority classification problem)
#Lattices of 149 Binary Cells (599, 999)#Rules of Radius 3 (7 Cells in
Neighborhood)#Task: Organize to< All 1's if Initial Configuration (IC) has more
1 Cells< All 0's if IC has more 0 Cells
xt
x
12827 NK
Possible neighborhood states
38128 104.32 NKK
Possible CA transition functions
rocha@indiana.eduhttp://informatics.indiana.edu/rocha/i-bic
biologicallyInspired
computing
INDIANAUNIVERSITY
Informatics luis rocha 2015
Cellular automataencoding in GA with binary encoding
xt
x12827 NKPossible neighborhood states
0 0 0 0 0 0 0 0
0 0 0 0 0 0 1 0
0 0 0 0 0 1 0 1
0 0 0 0 0 1 1 0
0 0 0 0 1 0 0 1
0 0 0 0 1 0 1 0
0 0 0 0 1 1 0 0
0 0 0 0 1 1 1 0
0 0 0 1 0 0 0 1
0 0 0 1 0 0 1 0
0 0 0 1 0 1 0 1
0 0 0 1 0 1 1 1
0 0 0 1 1 0 0 0
0 0 0 1 1 0 1 0………
0
0
1
0
1
0
0
0
1
0
1
1
0
0
010010101100100
010010101100100
010010101100100
010010101100100
010010101100100
Used in the evolutionary search by GA (elite selection)
x x x1 2 np
φCode:
!!!
S S np21 S
Selection
Variation
Genotype
Phenotype
Traditional Genetic Algorithm
011001
code
Pop of rules
rocha@indiana.eduhttp://informatics.indiana.edu/rocha/i-bic
biologicallyInspired
computing
INDIANAUNIVERSITY
Informatics luis rocha 2015
Evolving CA rulesWith genetic algorithms
#Das, Mitchell and Crutchfield< Used Genetic Algorithm to
evolve rules for this task
Das,R., Mitchell,M., Crutchfield,J.P., [1994]. "A genetic algorithmdiscovers particle-based computation in cellular automata". In:Parallel Problem Solving from Nature - PPSN III. Davidor,Y.,
Schwefel,H.-P., Manner,R. (Eds.), Springer-Verlag, pp. 344-353.
Typical Result:Block Expansion
Regulardomains
{1+}
{10+}
{0+}
Particles
rocha@indiana.eduhttp://informatics.indiana.edu/rocha/i-bic
biologicallyInspired
computing
INDIANAUNIVERSITY
Informatics luis rocha 2015
to evolve photos with numerical encodingsevolutionary algorithms
biologicallyInspired
computing
rocha@indiana.eduhttp://informatics.indiana.edu/rocha/i-bic
INDIANAUNIVERSITY
Informatics luis rocha 2015
real and integer encoding
1) Genotypes contain real or integer values1) Crossover is performed in the same way2) Mutation assigns a random number in a given interval
2) More computationally demanding for Reals3) Attention to crossover points
1) Conversion to binary avoids crossover issues, but longer genotypes
In genetic algorithms
0.3 1.7 3.8 1.7 6 1.2 3.2 6.4 2.8 0
0.3 1.7 3.8 1.7 2.9 1.2 3.2 6.4 2.8 0
n Genes
54 108 25 201 11 192
x y r R G B254 18 100 1 141 12 254 18 100 1 141 12…
Agent Chromosome/Genotype (Population of p agents)
n circles
biologicallyInspired
computing
rocha@indiana.eduhttp://informatics.indiana.edu/rocha/i-bic
INDIANAUNIVERSITY
Informatics luis rocha 2015
evolving computer programs
Fogel, Owens and Walsh (1966) Artificial Intelligence through simulated evolution. Wiley.
Evolution of finite-state machines
John Koza (1992) at Stanford University Genetic Programming: On the programming of
computers by means of Natural Selection. MIT Press.
Tree encodings (no clear genotype)
biologicallyInspired
computing
rocha@indiana.eduhttp://informatics.indiana.edu/rocha/i-bic
INDIANAUNIVERSITY
Informatics luis rocha 2015
genetic programming
1) Generate Random population of tree/programs
2) Evaluate Fitness Function for each program
Desired I/O, simplicity, speed3) Reproduce next generation
Selection by fitness Variation
crossover and mutation Fill new population
4) Go back to 2) until stop criteria is met Desired fitness Specified number of
generations Convergence
The workings
f(x1)
f(x2)
f(x3)
…
f(xi)
*
PI
R
R
-
* √
C
R*[(PI*C)-√R]
+
R
D
*
√
PI
R+(√PI*D)
/
PI
PI
PI
/
/ /
PI PI
+
R
D
*
√
PI
/
PI D
PIPI PI
/
/ /
PI
+
R *
√
PI
rocha@indiana.eduhttp://informatics.indiana.edu/rocha/i-bic
biologicallyInspired
computing
INDIANAUNIVERSITY
Informatics luis rocha 2015
A symbolic regression toolEureqa
Eureqa: http://ccsl.mae.cornell.edu/eureqa
biologicallyInspired
computing
rocha@indiana.eduhttp://informatics.indiana.edu/rocha/i-bic
INDIANAUNIVERSITY
Informatics luis rocha 2015
Gene expression programming
Proposed by Candida Ferreira Program trees are encoded in fixed-length linear genotypes Genotypes
Open-reading frame architecture Stop signal not necessarily at end of genotype
Non-coding genes are possible Can include genetic operators
Genes contain two types of symbols Functions (only at the head) and terminals
Multigenic solutions Assembled from non-coding operations between various open-reading frames
Including a genotype/phenotype map in GP
C. FERREIRA [2001]. Gene Expression Programming: A New Adaptive Algorithm for Solving Problems. Complex Systems, 13 (2): 87-129.
rocha@indiana.eduhttp://informatics.indiana.edu/rocha/i-bic
biologicallyInspired
computing
INDIANAUNIVERSITY
Informatics luis rocha 2015
evolving morphologies and robots with indirect encodingsKarl Sim´s simulations and The Golem Project
http://demo.cs.brandeis.edu/golem/
http://www.youtube.com/watch?v=oCXzcPNsqGA
rocha@indiana.eduhttp://informatics.indiana.edu/rocha/i-bic
biologicallyInspired
computing
INDIANAUNIVERSITY
Informatics luis rocha 2015
evolutionary designobjective function may be subjective
"Once a Darwinian process gets going in a world, it has an open-ended power to generate surprising consequences: us, for example" Richard Dawkins
Biomorphs
biologicallyInspired
computing
rocha@indiana.eduhttp://informatics.indiana.edu/rocha/i-bic
INDIANAUNIVERSITY
Informatics luis rocha 2015
Next lectures
Class Book Nunes de Castro, Leandro [2006]. Fundamentals of Natural Computing:
Basic Concepts, Algorithms, and Applications. Chapman & Hall. Chapter 3, all sections Sections 7.8 (evolving L-Systems), 8.3.2 (biomorphs) Chapter 5, all sections Section 7.7, 8.3.1,8.3.6,8.3.8-10
Lecture notes Chapter 1: “What is Life?” Chapter 2: “The logical Mechanisms of Life” Chapter 3: Formalizing and Modeling the World Chapter 4: “Self-Organization and Emergent Complex
Behavior” Chapter 5: “Reality is Stranger than Fiction”
posted online @ http://informatics.indiana.edu/rocha/i-bic Optional materials
Flake’s [1998], The Computational Beauty of Life. MIT Press Chapter 20
Scientific American: Special Issue on the evolution of Evolution, January 2009.
readings