Detecting Ambiguities in Regression Problems using TSK Models

Regression refers to the problem of approximating measured data that are assumed to be produced by an underlying, possibly noisy function. However, in real applications the assumption that the data represent samples from one function is sometimes wrong. For instance, in process control different strategies might be used to achieve the same goal. Any regression model, trying to fit such data as good as possible, must fail, since it can only find an intermediate compromise between the different strategies by which the data were produced. To tackle this problem, an approach is proposed here to detect ambiguities in regression problems by selecting a subset of data from the total data set using TSK models, which work in parallel by sharing the data with each other in every step. The proposed approach is verified with artificial data, and finally utilised to real data of grinding, a manufacturing process used to generate smooth surfaces on work pieces.