edu.iastate.jrelm.core
Interface Feedback<O>
- All Known Implementing Classes:
- FeedbackDoubleValue, SimpleFeedback
public interface Feedback<O>
Interface for classes representing feedback for reinforcement learning
algorithms. Such classes will encapulate data for passing learners
information regarding their past choices. In addition, Feedback classes may
contain methods for special processing of feedback information or for
comparison of Feedback objects.
Feedback may be simple, such as a single reward value resulting from the
last action chosen, or more complicated, such as a weighted reward
distribution over a recent history of multiple action choices. The Feedback
interface allows us to build appropriately complex classes for different
types of feedback.
- Author:
- Charles Gieseler
getValue
O getValue()
- Retrieve the
- Returns: