Fuzzy models for prediction based on random set semantics

In this paper we propose a random set framework for learning linguistic models for prediction problems. We show how we can model prediction problems based on learning linguistic prototypes defined using joint mass assignments on sets of labels. The potential of this approach is then demonstrated by its application to a model and by benchmark problem and comparing the results obtained with those from other state-of-the-art learning algorithms.

• Type: Conference
• Year: 2003
• Where: EUSFLAT
• Authors: Nicholas J. Randon, Jonathan Lawry
• Keywords: EUSFLAT 2003, EUSFLAT 2007, Prediction Problems, Random Set Framework, State-of-the-art Learning Algorithms