Sapient Agents Seven Approaches - PowerPoint PPT Presentation

About This Presentation
Title:

Sapient Agents Seven Approaches

Description:

Manually designed vs. automatically derived/adjusted/evolved ... SA searches for novel design concepts. SA uses several representation spaces? ... – PowerPoint PPT presentation

Number of Views:131
Avg rating:3.0/5.0
Slides: 19
Provided by: zs9
Learn more at: https://cs.gmu.edu
Category:

less

Transcript and Presenter's Notes

Title: Sapient Agents Seven Approaches


1
Sapient AgentsSeven Approaches
  • Zbigniew Skolicki
  • Tomasz Arciszewski

2
Agents vs. Agents
  • A wide spectrum of possibilities
  • Swarm agents vs. complex stand-alone agents
  • Manually designed vs. automatically
    derived/adjusted/evolved
  • Agents, Intelligent Agents, Sapient Agents

3
Agent
  • An autonomous system situated within an
    environment, which senses its environment,
    maintains some knowledge and learns upon
    obtaining new data and, finally, which acts in
    pursuit of its own agenda to achieve its goals,
    possibly influencing the environment

4
Agent
agent
environment
5
Agent
Acting
Reasoning
Sensing
environment
6
Intelligent Agent (IA)
Acting
Reasoning
Sensing
environment
7
Sapient Agent (SA)
Acting
Reasoning
Wisdom
Sensing
environment
8
Seven approaches
  • Difficult to give a precise definition at this
    stage
  • a definition by coverings

9
1. Knowledge Representation
  • SA utilizes meta-rules, complex models
  • Heuristic rules
  • SA understands the whole structure of a knowledge
    graph/ontology

10
2. Emergence
  • Understanding at global, macroscopic level
  • Line of Evolution approaches
  • SA recognizes emergent patterns

11
3. Exploitation/Exploration
  • SA searches for novel design concepts
  • SA uses several representation spaces?
  • SA focuses on exploration

12
4. Evolutionary Computation
  • SA makes strategic decisions
  • Diversity maintained
  • Slow convergence
  • Possible short-term diminishment of overall
    performance

13
5.Time
  • SA has wider (long-term) perspective
  • Hierarchical planning
  • Holistic understanding in the case of games
  • SA makes long term decisions

14
6. Domain Dependence
  • SA has inter-domain knowledge
  • Structural adaptation (not just adjusting
    parameters)
  • SA can abstract knowledge at very high level

15
7. Chaos
  • Identification of attractors
  • Multi-solution/stream tracking (as far as
    possible)
  • SA understands constraints of real world and
    chooses the safest solution

16
Summary
17
Conclusions
  • First attempt to classify Sapient Agents
  • All approaches separate but interrelated
  • Looking for Big Picture absolutely crucial for SA
  • Good starting point for a discussion
  • Much more work necessary

18
Thank you
  • Questions?
Write a Comment
User Comments (0)
About PowerShow.com