Incremental Learning from Noisy Data
Machine Learning
Knowledge Acquisition Via Incremental Conceptual Clustering
Machine Learning
Version spaces: an approach to concept learning.
Version spaces: an approach to concept learning.
| Hi-index | 0.00 |
Bias plays a significant role in inductive inference. In the framework of inductive concept learning from examples, there is an unknown target concept to be learned and a set of instances classified as positive or negative examples of the target concept. If learning is incremental, hypotheses (usually expressed in terms of instance features) are formed and then modified to remain consistent with the growing set of known instances. Generalization and specialization are frequently used for making modifications. A hypothesis is consistent with the instances if it logically implies all known positive instances and no known negative instances. If learning is empirical and the concept language is rich, the number of hypotheses consistent with the instances may be quite large. Since the purpose of each hypothesis is to predict over future instances, a judicious choice of one hypothesis over others ("bias")1 can improve these predictions, thereby enhancing system performance.