Report de Recerca IIIA 90/4

This is the draft for discussing the philosophy underlying the research on learning systems when the IIIA started a long term research line in Machine Learning. The contents of this paper owe a lot to discussions with Ramon Lopez de Mantaras and Walter van de Velde, but the opinions expressed in the draft are those of the author need not be, and indeed not all are, shared by these or other persons.

The Massive Memory Project

This is a White Paper on the AMP project

Enric Plaza

IIIA - Institut d'Investigació en Intel.ligència Artificial
CSIC - Spanish Scientific Research Council,
Campus Universitat Autonòma de Barcelona,
08193 Bellaterra, Catalonia, Spain.
Voice: +34 3 5809570;
Fax: +34 3 5809661;
Email: plaza@iiia.csic.es
WWW: http://www.iiia.csic.es

• Epigraph
• 1. Introduction
  • A Meta-Theoretical Statement:
  • A Second Meta-Theoretical Statement
• 2. The Experience-based Paradigm
  • 2.1 Experience-based Artificial Intelligence
    • 2.1.1 Schank's Dynamic Memory approach
    • 2.1.2 The Society of Mind and K-lines, a theory of memory
    • 2.1.3 Memory-based Architectures
    • 2.1.4 Common-sense as memory of world knowledge in CYC
  • 2.2 Experience-based machine learning
    • 2.2.1. The dimensions of ML approaches
    • 2.2.2. Case-based and Analogical learning
    • 2.2.3. Chunking and compilation
    • 2.2.4. Self-improving system learning in THEO
• 3. Massive Parallelism and Memory-based Approaches
• 4. PROPOSAL 1. Memory is an active, central notion
• 5. PROPOSAL 2. Memory allows parallelism in problem solving
  • 5.1 Case-based Reasoning and Marker Passing
  • 5.2 The scope of spreading activation tasks
• 6. PROPOSAL 3. Representation and Memory
  • 6.1 Representation as Inference
  • 6.2 Spontaneous and Deliberate Inference
    • 6.2.1 Spontaneous Inference
      • Perception/Recognition as Non-Mediated Reminding/Indexing
    • 6.2.2 Deliberate inference as problem solving:
  • 6.3 Learning as Representation Change
• 7. Theoretical Commitments
  • 7.1 Situated Learning of an Agent
  • 7.2 Cognitive Clichés and Learning
    • 7.2.1 Learning as problem-solving tasks
    • 7.2.2 Meta-level inference and learning
    • 7.2.3 Clichés and meta-level inference
    • 7.2.4 Clichés and Spontaneous Inference
• References