Understanding Dubious Future Problems

Abstract

Being able to predict the performance of a Case-Based Reasoning system against a set of future problems would provide invaluable information for design and maintenance of the system. Thus, we could carry out the needed design changes and maintenance tasks to improve future performance in a proactive fashion. This paper proposes a novel method for identifying regions in a case base where the system gives low confidence solutions to possible future problems. Experimentation is provided for RoboSoccer domain and we argue how encountered regions of dubiosity help us to analyse the case base and the reasoning mechanisms of the given Case-Based Reasoning system.

Fast anytime retrieval with confidence in large-scale temporal case bases

“…Especially, low confidence zones in a CB's solution space (e.g. categorized as dubiosity patterns in [48]) can help ALK further tune its retrieval performance. For example, if the solutions of best-so-far kNNs exhibit a ''border'' pattern where two competing solutions (say classes) exist; then the algorithm may use this information to encourage resuming the search expecting a possible change in kNNs favoring one of the competing solutions.…”

On Deriving Adaptation Rule Confidence from the Rule Generation Process

“…Reilly et al [6] propose a featurebased confidence model for assessing confidence of the proposed values for a feature by recommender systems. Mulayim and Arcos [7] propose a method for identifying areas of the problem space for which cases give uncertain solutions, identifying regions of the case base in which those problems are located, to guide maintenance. Hullermeier's [8] Credible Case-based Inference (CCBI), for regression tasks, estimates solutions based on "credible sets" of cases, i.e., sets of high confidence cases.…”