Trevor P. Martin

Transitioning from recognition to understanding in vision using additive Cartesian granule feature models

Proposes an approach to object recognition that facilitates the transition from recognition to understanding. The proposed approach begins by segmenting the images into regions using standard image processing approaches, which are subsequently classified using a discovered fuzzy Cartesian granule feature classifier. Understanding is made possible through the transparent and succinct nature of the discovered models. The recognition of roads in images is taken as an illustrative problem in the vision domain. The discovered fuzzy models, while providing high levels of accuracy (97%), also provide understanding of the problem domain through the transparency of the learnt models. The learning step in the proposed approach is compared with other techniques, such as decision trees, naive Bayes methods and neural networks.

An Intelligent Method for Inferring Information about the User/Client

Having in mind todays growth of information sources, both in terms of their number and of their size, whether we are referring to the Internet, a corporate intranet, or a library information retrieval system, we can say that manipulating information is not a trivial task. The user is not often being catered for in distributed information systems. He/ she seems to be interacting with systems that do not recognize his/her uniqueness and thus do not offer an individualized treatment. As a result, User Modeling is a core, essential factor in achieving personalization. We present here an intelligent way of inferring user related information that is not available, a situation that is very likely to occur due to sparseness of relevant data. This method can be very useful in recommender systems and this is illustrated with an example.

Knowledge discovery using Cartesian granule features with applications

Current approaches to knowledge discovery can be differentiated based on the discovered models using the following criteria: effectiveness, understandability (to a user or expert in the domain) and evolvability (the ability to adapt over time to a changing environment). Most current approaches satisfy understandability or effectiveness, but not simultaneously while tending to ignore knowledge evolution. We show how knowledge representation based upon Cartesian granule features and a corresponding induction algorithm can effectively address these knowledge discovery criteria (in this paper, the discussion is limited to understandability and effectiveness) across a wide variety of problem domains, including control, image understanding and medical diagnosis.