Please enable JavaScript.
Coggle requires JavaScript to display documents.
Knowledge-Based & Multi-Agent Systems - Coggle Diagram
Knowledge-Based & Multi-Agent Systems
Knowledge-Based Systems (KBS)
Logical Agents
Act based on reasoning
Not just reflex reactions
Knowledge Base (KB)
Represented using logic languages
AXIOMS = predefined truths
Stores facts and rules (sentences)
TELL & ASK interface
Sequence of Interaction
Perception → Reasoning → Action
Make Action Query
Make Action Sentence
Make Percept Sentence
Knowledge Representation
Declarative (facts, rules)
Procedural (coded behavior)
Knowledge vs Implementation Level
Wumpus World Example
P.E.A.S. Framework
Performance: +1000 (gold), -1000 (pit/Wumpus), -1 (step), -10 (arrow)
Environment: 4x4 grid, gold, pits, Wumpus
Actuators: Move, Turn, Grab, Shoot, Climb
Sensors: Stench, Breeze, Glitter, Bump, Scream
World Characteristics
Sequential, Partially Observable
Multi-Agent Environments
Conventions & Communication
Example: Tennis, Driving
Communication
Plan Recognition
Predict actions of others based on behavior
Coordination & Planning
Agents make their own plans
Must cooperate to reach shared goals
Evolved Behaviors (Biological Inspiration)
Sharks (feeding), Bees (swarm), Ants (self-organization)
Zebras, Bats, Starlings (flocking/swarming)
Flocking & Emergent Behavior
Emergence
Simple rules → Complex behavior
Macro patterns from micro actions
Craig Reynolds’ Boids
Alignment (move in same direction)
Cohesion (move together)
Separation (avoid crowding)
Applications
Animation (eg: Lion King)
Simulation (fish, insects)
Robotics
Knowledge Representation
Techniques
Semantic networks
Frames
Logic (Propositional, Predicate)
Ontologies
Must be consistent, complete, efficient
What knowledge the agent knows & how it’s stored
References
https://www.sciencedirect.com/topics/computer-science/knowledge-based-system
https://www.geeksforgeeks.org/knowledge-representation-in-ai/
https://www.sciencedirect.com/science/article/abs/pii/S0278612518300414
Multi-Agent Systems (MAS)
Multiple agents interacting in an environment
Agents can:
Cooperate
Compete
Negotiate
Distributed problem-solving
Examples
Traffic systems
Game AI
Swarm robotics