David Patterson

IJCAI Conference 2007 Conference Paper

• Niall Rooney
• David Patterson

In this paper we present a novel method that fuses the ensemble meta-techniques of Stack-ing and Dynamic Integration (DI) for regres-sion problems, without adding any major computational overhead. The intention of the technique is to benefit from the varying per-formance of Stacking and DI for different data sets, in order to provide a more robust technique. We detail an empirical analysis of the technique referred to as weighted Meta-Combiner (wMetaComb) and compare its per-formance to Stacking and the DI technique of Dynamic Weighting with Selection. The em-pirical analysis consisted of four sets of ex-periments where each experiment recorded the cross-fold evaluation of each technique for a large number of diverse data sets, where each base model is created using random fea-ture selection and the same base learning al-gorithm. Each experiment differed in terms of the latter base learning algorithm used. We demonstrate that for each evaluation, wMeta-Comb was able to outperform DI and Stack-ing for each experiment and as such fuses the two underlying mechanisms successfully

IJCAI Conference 2005 Conference Paper

Sophia: A novel approach for Textual Case-based Reasoning

• David Patterson
• Niall Rooney
• Vladimir Dobrynin
• Mykola

In this paper we present a novel methodology for textual case-based reasoning. This technique is unique in that it automatically discovers case and similarity knowledge, is language independent, is scaleable and facilitates semantic similarity between cases to be carried out inherently without the need for domain knowledge. In addition it provides an insight into the thematical content of the casebase as a whole, which enables users to better structure queries. We present an analysis of the competency of the system by assessing the quality of the similarity knowledge discovered and show how it is ideally suited to case-based retrieval (querying by example).

AAAI Conference 2002 Conference Paper

A Regression Based Adaptation Strategy for Case-Based Reasoning

• David Patterson
• and Mykola Galushka

Adaptation is the least well studied process in casebased reasoning (CBR). The main reasons for this are the potentially complex nature of implementing adaptation knowledge and the difficulties associated with acquiring quality knowledge in the first place and competently maintaining it over time. For these reasons most CBR systems are designed to leave the adaptation component to the expert and therefore function simply as case retrieval systems as opposed to truly reasoning systems. Here we present a competent adaptation strategy, which uses a modified regression algorithm to automatically discover and implement locally specific adaptation knowledge in CBR. The advantages of this approach are that the adaptation knowledge acquisition process is automated, localised, guaranteed to be specific to the task at hand, and there is no adaptation knowledge maintenance burden on the system. The disadvantage is that the time taken to form solutions is increased but we also show how a novel indexing scheme based on k-means clustering can help reduce this overhead considerably.