The Good, the Bad and the Incorrectly Classified: Profiling Cases for Case-Base Editing

Abstract

Abstract. Case-based approaches to classification, as instance-based learning techniques, have a particular reliance on training examples that other supervised learning techniques do not have. In this paper we present the RDCL case profiling technique that categorises each case in a casebase based on its classification by the case-base, the benefit it has and/or the damage it causes by its inclusion in the case-base. We show how these case profiles can identify the cases that should be removed from a case-base in order to improve generalisation accuracy and we show what aspects of existing noise reduction algorithms contribute to good performance and what do not.

On Dataset Complexity for Case Base Maintenance

“…Finally, we think the ideas in [4] may be relevant to future work. In [4], Delany identifies 8 types of case profile, depending on the non-emptiness or otherwise of four sets associated with each case (the reachability set, the coverage set, the liability set and the dissimilarity set).…”

“…In [4], Delany identifies 8 types of case profile, depending on the non-emptiness or otherwise of four sets associated with each case (the reachability set, the coverage set, the liability set and the dissimilarity set). Her paper looks empirically at what happens when cases with certain profiles are deleted.…”

Adaptive case-based reasoning using retention and forgetting strategies

“…Thus, for example, Brighton and Mellish [50] use the case competence properties of cases in their Iterative Case Filtering (ICF) algorithm. On the other hand, Delany [61] presented a case profiling technique that categorizes each case in the case-base. The profile is based on three characteristics that are derived from constructing a competence model of the case base.…”

Similarity assessment for removal of noisy end user license agreements

“…Noise reduction has been researched extensively, especially with regard to finding remedies for specific classes of problems. Recently, a thorough review of noise reduction approaches for instance-based learning algorithms has been carried out [10]. This review identifies problems associated specifically with instance-based learners and presents an approach, based on case-based reasoning, to detect instances that are either contributing negatively or positively to a particular case or problem.…”