Lecture 11 - Swarm intelligence and Metaheuristics Flashcards
Swarm intelligence and metaheuristics
What is a machine?
entity capable of transfer and/or processing matter/energy
Swarm intelligence and metaheuristics
What is an automaton?
entity capable of transfer and/or processing information
Swarm intelligence and metaheuristics
What is a robot?
entity capable of transfer and/or processing both information and matter/energy
Swarm intelligence and metaheuristics
What is intelligence?
analysis and/or synthesis of ordered patterns in an ‘unpredictable’ manner
Swarm intelligence and metaheuristics
What does it mean to be unpredictable?
globally ‘intractable’ or ‘externally not-representable’
Swarm intelligence and metaheuristics
What does it mean to be Intractable?
not computable in real-time
Swarm intelligence and metaheuristics
What does it mean to be externally not-representable?
inaccessibility of internal states and representations
Swarm intelligence and metaheuristics
What is the definition of an intelligent swarm?
“A group of non-intelligent robots forming, as a group, an intelligent robot. In other words, a group of ‘machines’ capable of forming ‘ordered’ material patterns ‘unpredictably’.”
Swarm intelligence and metaheuristics
What are the key qualities of swarm intelligence in nature? (DNQSMS)
- Decentralized
- Not-synchronized
- Quasi-homogeneous
- Simple units
- Many (From hundreds to billions units, but not infinite)
- Stigmergy: modifying the environment for indirect communication
Swarm intelligence and metaheuristics
What is flocking?
Movement of a swarm of agents with specific behavior.
Swarm intelligence and metaheuristics
What are the 3 flocking rules of boids? (SAC)
- Separation: steer to avoid crowding local flockmates
- Alignment: steer towards the average heading of local flockmates
- Cohesion: steer to move towards the average position (center of mass) of local flockmates
Swarm intelligence and metaheuristics
What are metaheuristics?
Search algorithms for solving complex optimization problems that are guided by a fitness function.
Swarm intelligence and metaheuristics
Should metaheuristics find the optimal solution?
They’re mostly they’re concerned with A good solutions than THE best solution.
Swarm intelligence and metaheuristics
What is the type of search a metaheuristic performs called?
It’s a partial search - finding a good solution, not necessarily the best one.
Swarm intelligence and metaheuristics
Name the metaheuristics in this course (GAPS).
- Genetic algorithm (GA)
- Ant colony optimization (ACO)
- Particle swarm optimization (PSO)
- Simulated annealing (SA)
Swarm intelligence and metaheuristics
Why is it called simulated annealing?
Name comes from cooling of metals - annealing.
Swarm intelligence and metaheuristics
Describe the idea of simulated annealing (SA).
When running SA, new solution spaces are selected for exploration with a probability based on the temperature (parameter). The temperature decays as time goes by.
Swarm intelligence and metaheuristics
Describe the idea of ant colony optimization (ACO).
ACO is an algorithm where agents explore the search space. They deposit pheromones to modify the environment (called stigmergy) as they explore, which probabilistically promotes other ants to explore that path.
Swarm intelligence and metaheuristics
Describe the three components of particle swarm optimization (PSO).
- Momentum/inertia
- Own best position history
- Global best position
Swarm intelligence and metaheuristics
How do you update a PSO particle’s position?
(Image)
Swarm intelligence and metaheuristics
What is the ideal fitness function (mathematical definition)?
A convex function without local minima (Unless that local minima is a global minima).
Swarm intelligence and metaheuristics
What is the easiest way to have a multi-objective fitness function?
Weight and sum each objective’s fitness.
