Implementation of DEvelopmentAl Learning (IDEAL) Course

Home » 4. Self-programming » 45. Implementation

Implementing a self-programming agent

If you have no interest in programming then you can skip this page and proceed to the next page.

Project 4 implements the algorithm described on Page 43 and Page 44 to let you run your first self-programming agent.

• main / Main ← Edit Main.java to instantiate Existence040.
• existence / Existence040
• coupling / Experiment040 ← Now, experiments have an intended interaction.
• coupling / interaction / Interaction040 ← Now, interactions have an experiment to perform.

For Lesson 4, your programming activities are:

• 1. Change Main.java to instantiate Existence040 and run it. Make sure that you obtain the trace shown in the next page
• 2. Change Existance040 to instantiate Environment010 and then Environment030 instead of Environment040 and run it. Observe that the modified Existence040 can still learn to be pleased.

Lesson 4 shows that Existence040 can adapt to three different environments (Environment010, Environment030, and Environment040). In fact, it can adapt to environments that require enacting sequences consisting of two steps or less (first-level composite interactions).

However, Existance040 makes bad decisions in environments that require longer sequences. This is because the algorithm does not yet process abstract experiments in the exact same way as primitive experiments, which stops it from being fully recursive. We will address this problem in the next lesson with Radical Interactionism, which removes the notion of primitive experiments and considers all experiments as abstract.

« Previous | Next »

See public discussions about this page or start a new discussion by clicking on the Google+ Share button. Please type the #IDEALMOOC045 hashtag in your post: