Abstract. Storing and using specific instances improves the performance of several supervised learning algorithms. These include algorithms that learn decision trees, classification rules, and distributed networks. However, no investigation has analyzed algorithms that use only specific instances to solve incremental learning tasks. In this paper, we describe a framework and methodology, called instance-based learning, that generates classification predictions using only specific instances. Instance-based learning algorithms do not maintain a set of abstractions derived from specific instances. This approach extends the nearest neighbor algorithm, which has large storage requirements. We describe how storage requirements can be significantly reduced with, at most, minor sacrifices in learning rate and classification accuracy. While the storage-reducing algorithm performs well on several realworld databases, its performance degrades rapidly with the level of attribute noise in training instances. Therefore, we extended it with a significance test to distinguish noisy instances. This extended algorithm's performance degrades gracefully with increasing noise levels and compares favorably with a noise-tolerant decision tree algorithm.

Abstract. An important step in the solution of a target problem in case-based reasoning (CBR) is the retrieval of similar previous cases that can be used to solve the target problem. We review a selection of papers from the CBR literature on aspects of retrieval, such as approaches to the assessment of surface and structural similarity and techniques for automating the construction and maintenance of similarity measures. We also examine a number of retrieval techniques that have been developed to address the limitations of retrieval based purely on similarity. 1

We present a prototype of the AICronus system, an argumentation system that automates a challenging reasoning process used by experts in cosmogenic isotope dating. The architecture of the system is described and preliminary results are discussed. 1.

ults the other as needed. Scenario (3) refers to hybrids in which the learning system consults with the expert knowledge before producing output, such as when CN2 consults with qualitative models to ensure that its induced rules are explainable (Clark z Matwin, 1993), or when knowledge on representational biases is used to select which learning algorithm to apply (Brodley, 1993). Similarly, expert systems in Scenario (4) consult with learning systems before producing output, so as to support such behaviors as knowledge-based refinement. In Proceedi,gs of the First I,ter,atio,al New Zeala,d Two-S'tream Co,fere,ce o Artificial Neural Networks a,d Ex'pert 3'gstems. (pp. 150 151). Dunedin, NZ: IEEE Press. 2Current address: NRL Center for Applied AI, Washington, DC 20375 USA, aha@aic.nrl.navy. mil Other hybrids exist (i.e., Scenario (5)) that have no such master-slave relationship. For example, experts using Clark's (1989) cooperative expert system modify parameters used by the learning

Scientific reasoning is a complex process, alternately requiring flashes of insight and tedious analysis. This dichotomy is evident in constructing a geologic timeline for a landform using cosmogenic isotope dating. Experts in this field frequently spend months on repetitive mathematical tasks, until they have gathered enough information to suddenly understand the