Learning and Applying Case Adaptation Rules for Classification: An Ensemble Approach

Abstract

The ability of case-based reasoning systems to solve novel problems depends on their capability to adapt past solutions to new circumstances. However, acquiring the knowledge required for case adaptation is a classic challenge for CBR. This motivates the use of machine learning to generate adaptation knowledge. Much adaptation learning research has studied the case difference heuristic (CDH) approach, which generates adaptation rules from pairs of cases in the case base by ascribing observed differences in case solutions to the differences in the problems they solve, to generate rules for adapting similar problem differences. Extensive research has successfully applied the CDH approach to adaptation rule learning for case-based regression (numerical prediction) tasks. However, classification tasks have been outside of its scope. The work presented in this paper addresses that gap by extending CDH-based learning of adaptation rules to apply to cases with categorical features and solutions. It presents the generalized case value heuristic to assess case and solution differences and applies it in an ensemble-based casebased classification method, ensembles of adaptations for classification (EAC), built on the authors' previous work on ensembles of adaptations for regression (EAR). Experimental results support the effectiveness of EAC.

Case Based Reasoning as a Model for Cognitive Artificial Intelligence

“…This is based on learning adaptation knowledge by understanding relationships among cases; e.g. ensemble learning for adaptation [19], or gradient learning for adaptation [20].…”

Case-based prediction – A survey

“…these rules can be written (σ S = α) → (σ R = β) and can be expressed in various forms, such as adaptation rules [7,27,73,90], dependencies between problem and solution features [42], co-variations [5] or fuzzy rules [16] to name a few. The prediction strategy consists in triggering the rules on pairs of cases involving 405 the new case using a kind of similarity-based inference, as detailed below, in order to derive potential outcomes for the new case.…”

“…It is often the case that the similarity measures σ S and σ R are unknown, or Some case-based prediction approaches such as [73] include a rule selection 420 step prior to the inference, while others such as proportion-based analogical classifiers (see Sec. 6.4) trigger only one rule.…”

A Factorial Study of Neural Network Learning from Differences for Regression

“…These works are based on the exploitation of the case base, following the idea that adaptation knowledge can be acquired from differences between pairs of cases using the case difference heuristic (CDH) [2]. The CDH has been applied first outside of the context of neural networks to produce adaptation rules with various approaches [2,3,6].…”