ISL and Creativity - A production-oriented model

Segmentation, Chunking, Attention and Creative generations

Main modules (in creativitips.CTPs):

• pparser.py (implementation of PARSER)
• tps.py (modul for tracking Transitional Probabilities)
• computation.py (for computing sequences iteratively: using previous modules)

Entry points:

• parser_main.py: for experiments using PARSER only
• tps_main.py: for experiments combining TPs and PARSER mechanisms

---

Issues:

• segmentation
  • TPS (forward, backward, MI). How to select the right one??
  • With or without chunking?
• chunking
  • with or without TPS as cues? what about using MDL instead?
• abstraction and generalization:
  • algebraic patterns for abstraction?

What's next?

• Replay of experiences
• Retrieval
• state uncertainty
• Inference and error-driven learning
• Multi-modal integration