Connor's Notes

⬅️ Back to portfolio

COSC 3P71: Lecture 9 Notes

3 min read

Midterm structure

  • some short answers
  • is this true/false and why
  • no simple true/false or multiple choice
  • topics
    • agents
    • informed search
    • blind search (bit less than informed search)
    • search algorithms & cost
    • why is utility based better than x

Assignment 1

  • “not extended” to Oct 12/13

Genetic Algorithms

Note

Skipping past a lot of stuff → officially starting slide 34

Selection techniques

  • Tournament selection: select K individuals, keep best for reproduction

Important

“What is tournament selection” will be asked on a quiz/midterm/exam

  • Roulette wheel selection: probabilistic selection based on fitness

Genetic Operators

Crossover

provides a method of combining two candidates form the population to create new candidates

  • swaps pieces of genetic material between two individuals (represents mating)
    • usually two individuals (parents) combine to produce two more individuals (children)
    • can also define an asexual or single-child crossover as well

Mutation

changing gene value(s)

  • lets offspring evolve in new directions → introduces a certain amount of randomness
    • certain traits may become fixed

Replication

copy an individual to the next generation without alteration

Crossover Operations

  • 1-point, n-point crossover
  • uniform order crossover (UOX)
    • vs uniform crossover
Step 1: Setup  
Parents (P1, P2) and Mask  
P1:   6  2  1  4  5  7  3  
Mask: 0  1  1  0  1  0  1  → generate new mask for every 2 parents
P2:   4  3  7  2  1  6  5  

Step 2: Copy Genes by Mask  
- Copy genes from P1 → C1 where mask = 1  
- Copy genes from P2 → C2 where mask = 1  

C1:   -  2  1  -  5  -  3  
C2:   -  3  7  -  1  -  5  

Step 3: Fill Remaining from Opposite Parent  
- Fill blanks (-) with remaining genes from the other parent in order  

C1:   4  2  1  7  5  6  3  
C2:   6  3  7  2  1  4  5  

✅ Final Offspring  
C1:  4  2  1  7  5  6  3  
C2:  6  3  7  2  1  4  5
  • order crossover (OX)
    1. Copying a randomly selected set from the first parent.
    2. Filling the remaining positions with the order of elements from the second

Example 1

Step 1: Copy randomly selected set from first parent  
p1: 1 2 3 4 5 6 7 8 9  
p2: 9 3 7 8 2 6 5 1 4  

c1: * * * 4 5 6 7 * *  
c2: * * * 8 2 6 5 * *  

Step 2: Copy rest from second parent in order  
Remaining order from p2: 1, 9, 3, 8, 2  

C1: 3 8 2 4 5 6 7 1 9  
C2: ?

Example 2

Step 1: Copy randomly selected set from first parent  
p1: 1 2 3 4 5 6 7 8 9  
p2: 4 5 2 1 8 7 6 9 3  

c1: * * * 4 5 6 7 * *  

Step 2: Copy rest from second parent in order  
Remaining order from p2: 9, 3, 2, 1, 8  

C1: 2 1 8 4 5 6 7 9 3
  • partially mapped (PMX)
  • cycle crossover (CX)

Learning illegal structures

consider the travelling salesman problem (TSP) where an individual represents a potential solution. the standard crossover operation can produce illegal children

Parent A: Thorold, St.Catharines, Hamilton, Oakville, Toronto
Parent B: Hamilton, Oakville, Toronto, St.Catharines, Thorold

Parent AB: Thorold, St.Catharines, Hamilton, St.Catharines, Thorold
Parent BA: Hamilton, Oakville, Toronto, Oakville, Toronto

possible solution: replace duplicates with a different city

Mutation

examples:

  • Inversion: reverse it
  • Insertion: pick a random thing and move it

Note

ended slide 65 (take notes on everything that was skipped by in lecture)

Important

midterm is up to slide 65 in GA slides

Graph View