Search

In this paper we introduce a new fuzzy system using adaptive fuzzy pattern classification (AFPC) for data-based online evolvement. The fuzzy pattern concept represents an efficient tool for handling uncertainty in multi-dimensional data streams and combines powerful performance, flexibility and meaningful interpretability within one consistent framework. We outline AFPC for non-linear, multi-dimensional transition processes, namely, for the identification of lane change intention in car driving. While lane changes are rare, they are highly safety-relevant transition processes, showing high fuzziness and large individual and inter-individual variations (e.g., in lane change duration). The method employs a combined knowledge- and data-based approach, and the underlying fuzzy potential membership function concept models expert knowledge, closely mirroring human cognition. The design of AFPC comprises (I) an initial training phase (off-line and supervised), which generates a meaningful start-classifier, (II) an online application phase, and finally (III) an evolvement phase (online and unsupervised). Here we consider parametric and structural adaptations and discuss prospects and future challenges. Furthermore, we present specific modeling results for such online data from a real driving study. Next-generation advanced driver assistance systems, as well as autonomously driven vehicles need to evolve, in terms of parameters and structure, based on online real-time data. AFPC presents an efficient tool for application in this area and others (e.g., medicine).

The article describes a general two-step procedure for the numerical translation of vague linguistic terms (LTs). The suggested procedure consists of empirical and model components, including (1) participants' estimates of numerical values corresponding to verbal terms and (2) modeling of the empirical data using fuzzy membership functions (MFs), respectively. The procedure is outlined in two studies for data from N = 89 and N = 109 participants, who were asked to estimate numbers corresponding to 11 verbal frequency expressions (e.g., sometimes). Positions and shapes of the resulting MFs varied considerably in symmetry, vagueness, and overlap and are indicative of the different meanings of the vague frequency expressions. Words were not distributed equidistantly across the numerical scale. This has important implications for the many questionnaires that use verbal rating scales, which consist of frequency expressions and operate on the premise of equidistance. These results are discussed for an exemplar questionnaire (COPSOQ). Furthermore, the variation of the number of prompted LTs (5 vs. 11) showed no influence on the words' interpretations.

In some nonstationary time series, where a global model is neither available nor applicable, we may observe recurring patterns that can be extracted to create several local models instead. This article proposes knowledge-based short-time prediction methods for multivariate streaming time series that rely on the early recognition of such local patterns. A parametric fuzzy model for patterns is presented, along with an online, classification-based recognition procedure, which will introduce the notion of evolving classification results. Subsequently, two options are discussed to predict time series employing the fuzzified pattern knowledge, accompanied by an example. Special emphasis is placed on comprehensible models and methods, as well as a seamless interface to data mining algorithms.

This article proposes an individual fuzzy modelling and treatment of data allowing for the specific uncertainty of each datum. A modelling approach based on parametric fuzzy sets is being introduced which can be employed to model both data with individual uncertainties as well as abstract phenomena in a feature space (classes). An aggregation procedure is being presented which takes the individual characteristics of all learning objects into account, resulting in a fuzzy description of the same, interpretable structure. Examples are given using both synthetic and real-world data.

This article contributes to clustering and fuzzy modelling of data such that specific characteristics of each datum can be incorporated. Particularly, each object may exhibit an individual area of influence in its feature space, for which it is representative. For such objects, a similarity measure is introduced, which is used to modify common clustering algorithms to take each object’s extent into account when finding clusters. A real-world example demonstrates the practical usability of the presented methods, which deliver results in accordance to findings of experts in that field.

This article proposes knowledge-based short-time prediction methods for multivariate streaming time series, relying on the early recognition of local patterns. A parametric, well-interpretable model for such patterns is presented, along with an online, classification-based recognition procedure. Subsequently, two options are discussed to predict time series employing the fuzzified pattern knowledge, accompanied by an example. Special emphasis is placed on comprehensible models and methods, as well as an easy interface to data mining algorithms.

This work is dedicated to a network oriented modelling approach based on the interconnection of multivariate Fuzzy Pattern Classifier nodes. The main point is the elaboration of a data driven and hierarchical design strategy for such fuzzy classifier network models. In detail two essential issues will be dealt with: the automatic layout of the network and the configuration of the classifier nodes. Both issues are addressed using a multiple cluster analysis approach. Finally the network design and operation ARE illustrated within the scope of an practical example.

A modeling method is suggested in this paper, which permits building multidimensional fuzzy models of time series consisting of fuzzy prototypes. These models have to be trained in a so-called period of learning and are suitable for short, medium and long range forecasts. The prediction of an incomplete time series is based on fuzzy classification to the prototypes. The results are grades of membership. In principle, these grades and further courses of prototypes are used to forecast the time series.