Python code for the Implementation of DEvelopmentAl Learning (IDEAL) Course

Overview

The course is composed of 7 lessons:

1. The lesson builds a simple agent that memorizes interactions and uses them to predict the results when repeating experiments;
2. The lesson extends the agent attaching a valence to each interaction and uses these valences to avoid negative interactions;
3. The lesson extends the agent to learn composite interactions and uses them to exploit regularities in the sequences of interactions;
4. The lesson extends the agent with self-programming using abstract experiments and uses them to exploit longer sequences of interactions;
5. The lesson simplifies the agent with recursive learning and achieves complex behavior in a maze environment;
6. The lesson contains discussion and demonstrations about cognitive architectures;
7. The lesson contains discussion about research into developmental artificial intelligence.

Files

The source files implement the agents for each lesson.

The trace files are the outputs generated by running the agents.