Task 9-Collective intelligence, swarm intelligence

Question 1

Collective group behavior of social animals (bees, ants, fish, birds).

Answer 1

These bahviors are applicaple to systems and companies.
Flexibility: adaptability to changing environments
Robustness: doesn‘t matter if individuals fail
Self-Organization: actions aren‘t supervised
Simple rules -> complex behavior

Question 2

examples

Answer 2

Laying pheromone: efficient tracking for larger colonies (think about SNIF-ACT)
Bucket Brigade: ants are never stationary in a food chain; transfer points aren‘t fixed
Tasks they are specialized in are a priority; unless something else becomes more relevant
Simple fundamental rules

Question 3

agent based models

Answer 3

Use interactions between individuals and their environment to predict and improve collective behavior.

Stigmeric Interaction: indirect consensus about a coordination between agents

Autonomy of Agents: use heuristics and reinforcement learning to change goal/ strategy

Spatially distributed population of agents

Quantitative and objective: based on real world data

Question 4

Processes in ABMs:

Answer 4

Patterns: simulation models to study pattern from human group interactions
Contagion: spread of an influence between individuals via their interactions (trends, emotion: rage, hate, aggressiveness…)
Cooperation: research on social dilemmas

Question 5

expert cognition

Answer 5

adapting to teh choice of a single expert

Question 6

hill climbing

Answer 6

comparing own success to that of a demonstrator and base decision of relative level of success

Question 7

shared mental models

Answer 7

them members have common understanding of the task and who is responsible for wich task

Question 8

swarm intelligence

Answer 8

collective behaviour of self organised individuals /agents in swarm without supervision - more productive and efficient than individual

• Flexibility
• Robustness- group can perform task after an error
• Self-organization- activities are nit centrally controlled

Question 9

tri level hypothesis

Answer 9

for any information processing system to be understood completely , it must be described at three different levels (computational, algorithmic, implementation)