Brainscape's Knowledge GenomeTM

Browse over 1 million classes created by top students, professors, publishers, and experts.


See full index

ACIT4610 - Evolutionary AI > Lecture 05 - Genetic Algorithms > Flashcards

Lecture 05 - Genetic Algorithms Flashcards

1
Q

Genetic Algorithms

Who were the main 3 communities developing genetic algorithms?

A

Biologists, computer scientists/engineers, artificial-life researchers

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
2
Q

Genetic Algorithms

Why is there terminology overlap within evolutionary computing?

A

Competing communities of research hasn’t properly converged on terminology.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
3
Q

Genetic Algorithms

What is an individual in GA?

A

One data point with an attached fitness, which can evolve into new offspring.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
4
Q

Genetic Algorithms

What is a population in GA?

A

A collection of individuals to be evaluated.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
5
Q

Genetic Algorithms

How are individuals represented internally?

A

Pretty much anything that can be evaluated with a fitness function.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
6
Q

Genetic Algorithms

What is direct coding?

A

Some specific representation, like a final image. Genotype == phenotype in this case.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
7
Q

Genetic Algorithms

What is indirect coding?

A

Some abstract, recipe-y way of making the individuals.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
8
Q

Genetic Algorithms

What is a genotype?

A

Genotypes are the instruction for creating an individual.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
9
Q

Genetic Algorithms

What is a phenotype?

A

A phenotype is a manifested version of a genotype. Humans are phenotypes; genes are genotypes.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
10
Q

Genetic Algorithms

Why do we use the genotype/phenotype distinction in GAs? (3)

A
  • To resemble nature
  • To ease manipulation
  • To allow reuse, hence enabling actual usage of EC
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
11
Q

Genetic Algorithms

What is the concept of “limited resources” in GA?

A

The population size is capped; not all individuals can survive.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
12
Q

Genetic Algorithms

What’s the difference between an overlapping and non-overlapping generational model?

A
  • Overlapping model: Both parents and offspring can survive.
  • Non-overlapping model: Only offspring survive.
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
13
Q

Genetic Algorithms

What are some problems with the overlapping model?

A

Parents can survive, so what is a “new generation?”

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
14
Q

Genetic Algorithms

When using an overlapping model, how do you measure time flowing?

A

Track the number of births. A new generation happens at “#m births”.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
15
Q

Genetic Algorithms

Name the most common fitness selection methods. (FRUTT)

A
  • Fitness-proportional selection
  • Rank-proportional selection
  • Uniform (neutral) selection
  • Truncation selection
  • Tournament selection
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
16
Q

Genetic Algorithms

Describe Fitness-proportional selection.

A

1) Given the numerical fitness of each individual
2) Randomly pick one individual with probability proportional to the fitness (the better, the larger probability)

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
17
Q

Genetic Algorithms

Describe Rank-proportional selection.

A

1) Given the rank of each individual in a fitness-based ranking
2) Randomly pick one individual with probability proportional to the rank (the better, the larger probability)

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
18
Q

Genetic Algorithms

Describe uniform selection.

A

Pick randomly an individual (with uniform probability)

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
19
Q

Genetic Algorithms

Describe Truncation selection

A

Pick the best individual(s?) (elitism)

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
20
Q

Genetic Algorithms

Describe Tournament selection

A

Given a parameter n_size (size of the tournament):
1) Randomly (with uniform probability) pick n_size individuals
2) From them, choose the one with the best fitness

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
21
Q

Genetic Algorithms

How do you determine if criterion A better than criterion B?

A

Hypothesis testing

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
22
Q

Genetic Algorithms

Does uniform selection prefer fit or unfit individuals?

A

No preferences for either.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
23
Q

Genetic Algorithms

Does truncation selection prefer fit or unfit individuals?

A

Strong preference for fit individuals

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
24
Q

Genetic Algorithms

Does tournament selection prefer fit or unfit individuals?

A

Depends on n_size.
If 1: No preference.
As n_size grows -> stronger preference for fit individuals.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
25
Q

Genetic Algorithms

What does a strong preference for fit individuals lead to?

A
  • Converges to fit individuals.
  • Evolutions concentrates on improving promising solutions (exploitation).
  • Risk of falling into local minima.
26
Q

Genetic Algorithms

What does a weak preference for fit individuals lead to?

A
  • Population includes unfit individuals.
  • Many different, possibly not promising solutions, are explored (exploration).
  • Risk of not finding good solutions.
27
Q

Genetic Algorithms

What are the two most common genetic operators

A

Crossover and mutation.

28
Q

Genetic Algorithms

What is n-point crossover?

A

See the image. Select n points and flip which parent the genes come from.

29
Q

Genetic Algorithms

What is variable-length crossover?

A

Offspring can have variable length output which are concatenations of various parts of the parents.

30
Q

Genetic Algorithms

How does mutation work on trees?

A

Replace a random subtree with another randomly generated subtree.

31
Q

Genetic Algorithms

How does crossover work for trees?

A

Select two parents, then swap a random subtree.

32
Q

Genetic Algorithms

How are genetic operators constrained for trees?

A

Usually by depth.

33
Q

Genetic Algorithms

What is gaussian mutation?

A

Add parameterized gaussian noise.

34
Q

Genetic Algorithms

What is geometric crossover/crossover for real-valued vectors?

A

c = p1 + C*(p1 - p2), for C ~ U(0, 1)

35
Q

Genetic Algorithms

What’s the drawback of geometric crossover?

A

It lacks the ability to explore outside of the hyperrectangle enclosing the population

36
Q

Genetic Algorithms

What’s the hypothesized role of mutation and crossover?

A
  • Mutation = exploitation
  • Crossover = exploration
    (Debated)
37
Q

Genetic Algorithms

What techniques can you use to solve multiobjective problems?

A
  • Linearization
  • Lexicographical order
  • Pareto dominance
38
Q

Genetic Algorithms

What is linearization for multiobjective problems?

A

Weight and sum each objective value.

39
Q

Genetic Algorithms

What is lexicographical order for multiobjective problems?

A

Check each pair of objectives in turn.
1) Check f1(i1) > f1(i2). Select winner or advance to (2).
2) Check f2(i1) > f2(i2). Select winner or advance to (3).
…

40
Q

Genetic Algorithms

What is the drawback of too little diversity?

A

Too much exploitation → local minimum

41
Q

Genetic Algorithms

What is the drawback of too much diversity?

A

No exploitation, just coarse exploration (random walk)

42
Q

Genetic Algorithms

What is Expressiveness?

A

Whether good solutions can be expressed as phenotypes, e.g. the number of dimensions.

43
Q

Genetic Algorithms

What’s the problem with Low expressiveness?

A

Good/optimal solution might not be representable, or might not be reachable

44
Q

Genetic Algorithms

What’s the problem with Large expressiveness?

A

Large search space → very long or infiniti convergence time

45
Q

Genetic Algorithms

What is Grammatical Evolution?

A

Evolutions of strings given a grammar. See the image for an example.

46
Q

Genetic Algorithms

What does the Φ operator do?

A

Maps a genotype to a phenotype.

47
Q

Genetic Algorithms

What is ⊥?

A

An invalid solution.

48
Q

Genetic Algorithms

What is an invalid solution?

A

An illegal phenotype.

49
Q

Genetic Algorithms

What is f(i_j)?

A

The fitness function applied to the j-th individual.

50
Q

Genetic Algorithms

What is ≺?

A

A partial ordering of fitness values, i.e. fitness values can be compared and sorted.

51
Q

Genetic Algorithms

What is the distance dG?

A

A distance measure between two genotypes. (See image)

52
Q

Genetic Algorithms

What is the distance dP?

A

The distance measure between two phenotypes. (See image)

53
Q

Genetic Algorithms

What is invalidity?

A

The proportion of genotypes that are mapped to invalid phenotypes.

54
Q

Genetic Algorithms

What is Degeneracy?

A

The ratio of genotypes that are mapped to the same phenotype.

55
Q

Genetic Algorithms

What’s a synonym for degeneracy?

A

Redundancy

56
Q

Genetic Algorithms

What is Uniformity of degeneracy?

A

The degree to which the sizes of different sets of genotypes mapping to the same phenotype differ:

57
Q

Genetic Algorithms

What is Redundancy?

A

Parts of the genotype might not “agree” when being mapped to phenotypes, e.g. redundant genes might map to the same phenotype.

58
Q

Genetic Algorithms

What is locality?

A

The degree to which close genotypes are mapped to close phenotype

59
Q

Genetic Algorithms

What is evolvability?

A

The likelihood of obtaining a better individual after the application of a genetic operator

60
Q

Genetic Algorithms

Describe pareto dominance

A

1) Start with a graph with points.
2) While the graph has unselected points, repeat 3-6.
3) Select one of the outermost points on the graph that doesn’t belong to a pareto frontier.
4) Sweep a (hyper)plane around that point.
5) If you hit a point, select that as the next point.
6) If you hit an axis, assign a new pareto frontier to the selected points and move to step 2.

ACIT4610 - Evolutionary AI (11 decks)

Brainscape helps you reach your goals faster, through stronger study habits.
© 2024 Bold Learning Solutions. Terms and Conditions