At the end of this book, we want to conclude with an assessment of areas where we hope this book will make a contribution, a description of problems that are still open, and a discussion of possible future work.
KeywordsReinforcement Learning Concept Formation Action Selection Inductive Logic Programming Concept Hierarchy
Unable to display preview. Download preview PDF.
- 1.As pointed out in [Wrobel, 1991], a possible choice for action selection would be a hierarchy of situation-action descriptions into which a new situation is classified.Google Scholar
- 2.The initial vocabulary could either be a maximally coarse, i.e., binary segmentation, or a specific pre-determined segmentation that the agent could have inherited through evolution-like processes. This opens up the interesting possibility of using genetic algorithms [DeJong, 1988] to study the development of elementary feature sets.Google Scholar
- 3.A distinction is non-informative if the conditional probability of a (predicted) effector value given an attribute value is identical or nearly identical for both values. If a probabilistic concept hierarchy is used, this can be computed from the attribute probabilities stored with the concepts.Google Scholar