Issue 2 • Date May 1998
Cited by: Papers (6)
A neural network based fuzzy set model is proposed to support organizational decision making under uncertainty. This model incorporates three theories and methodologies: classical decision-making theory under conflict, as suggested by Luce and Raiffa (1957), the fuzzy set theory of Zadeh (1965, 1984), and a modified version of the backpropagation (BP) neural network algorithm originated by Rumelhart et al. (1986). An algorithm that implements the model is described, and an application of the model to a real data example is used to demonstrate its use View full abstract»
Cited by: Papers (8)
Electrical drives are usually modeled using circuit theory, with currents or linking fluxes chosen as state variables for its electrical part and rotor speed or position chosen for its mechanical part. Often, its internal structure contains nonlinear relations which are difficult to model such as dead-time, hysteresis, and saturation effects. On the contrary, if the available model is accurate enough, its parameter values are generally difficult to obtain and/or be estimated in real time. Therefore, the paper investigates the use of fuzzy logic for the automatic modeling of electrical drive systems. An experimental system composed of a DC motor supplied from a DC-DC converter is used. The authors underline the unsupervised learning characteristics of the fuzzy algorithm, its memory and generalization capabilities. Some learning situations with critical effects on model performance are presented and discussed, pointing out some results and conclusions concerning the fuzzy modeling process in practice View full abstract»
Cited by: Papers (67)
The paper presents a learning procedure for optimizing the parameters in the evidence-theoretic k-nearest neighbor rule, a pattern classification method based on the Dempster-Shafer theory of belief functions. In this approach, each neighbor of a pattern to be classified is considered as an item of evidence supporting certain hypotheses concerning the class membership of that pattern. Based on this evidence, basic belief masses are assigned to each subset of the set of classes. Such masses are obtained for each of the k-nearest neighbors of the pattern under consideration and aggregated using Dempster's rule of combination. In many situations, this method was found experimentally to yield lower error rates than other methods using the same information. However, the problem of tuning the parameters of the classification rule was so far unresolved. The authors determine optimal or near-optimal parameter values from the data by minimizing an error function. This refinement of the original method is shown experimentally to result in substantial improvement of classification accuracy View full abstract»
Cited by: Papers (21)
An adaptive fusion algorithm is proposed for an environment where the observations and local decisions are dependent from one sensor to another. An optimal decision rule, based on the maximum posterior probability (MAP) detection criterion for such an environment, is derived and compared to the adaptive approach. In the algorithm, the log-likelihood ratio function can be expressed as a linear combination of ratios of conditional probabilities and local decisions. The estimations of the conditional probabilities are adapted by reinforcement learning. The error probability at steady state is analyzed theoretically and, in some cases, found to be equal to the error probability obtained by the optimal fusion rule. The effect of the number of sensors and correlation coefficients on error probability in Gaussian noise is also investigated. Simulation results that conform to the theoretical analysis are also presented View full abstract»
Many real-world decision-making problems fall into the general category of classification. Algorithms for constructing knowledge by inductive inference from example have been widely used for some decades. Although these learning algorithms frequently address the same problem of learning from preclassified examples and much previous work in inductive learning has focused on the algorithms' predictive accuracy, little attention has been paid to the effect of data factors on the performance of a learning system. An experiment was conducted using five learning algorithms on two data sets to investigate how the change in labeling the class attribute can alter the behavior of learning algorithms. The results show that different preclassification rules applied on the training examples can affect either the classification accuracy or classification structure View full abstract»
Cited by: Papers (5)
In their previous work, the authors have developed a method for selecting features based on the analysis of class regions approximated by hyperboxes. They select features analyzing class regions approximated by ellipsoids. First, for a given set of features, each class region is approximated by an ellipsoid with the center and the covariance matrix calculated by the data belonging to the class. Then, similar to their previous work, the exception ratio is defined to represent the degree of overlaps in the class regions approximated by ellipsoids. From the given set of features, they temporally delete each feature, one at a time, and calculate the exception ratio. Then, the feature whose associated exception ratio is the minimum is deleted permanently. They iterate this procedure while the exception ratio or its increase is within a specified value by feature deletion. The simulation results show that the current method is better than the principal component analysis (PCA) and performs better than the previous method, especially when the distributions of class data are not parallel to the feature axes View full abstract»
An experimental approach to robotic grasping using a connectionist architecture and generic grasping functionsPublication Year: 1998 , Page(s): 239 - 253
Cited by: Papers (9)
An experimental approach to robotic grasping is presented. This approach is based on developing a generic representation of grasping rules, which allows learning them from experiments between the object and the robot. A modular connectionist design arranged in subsumption layers is used to provide a mapping between sensory inputs and robot actions. Reinforcement feedback is used to select between different grasping rules and to reduce the number of failed experiments. This is particularly critical for applications in the personal service robot environment. Simulated experiments on a 15-object database show that the system is capable of learning grasping rules for each object in a finite number of experiments as well as generalizing from experiments on one object to grasping from another View full abstract»
Cited by: Papers (12) | Patents (1)
The paper summarizes recent results on both binary and M-ary distributed hypothesis testing problems with decision makers (DMs) organized in structured decision networks. The general problem of finding an optimal organizational structure and decision strategy for such networks is formulated as a functional optimization problem. A normative model to study the effect of interactions between task structure and organizational design on the performance of hierarchical organizations is presented. A binary signal detection model is considered to illustrate the joint impact of organizational design and of task environment on the organizational decision performance. The concept of a congruent organizational structure (i.e., a structure that achieves centralized performance with minimal communication) is introduced, and a graph decomposition algorithm to synthesize congruent structures is discussed View full abstract»
The relationship between quantizability and learning complexity in multilayer neural networks is examined. In a special neural network architecture that calculates node activations according to the certainty factor (CF) model of expert systems, the analysis based upon quantizability leads to lower and also better estimates for generalization dimensionality and sample complexity than those suggested by the multilayer perceptron model. This analysis is further supported by empirical simulation results View full abstract»
Cited by: Papers (10) | Patents (8)
The data exploration task can be divided into three interrelated subtasks: 1) feature selection, 2) discovery, and 3) interpretation. This paper describes an unsupervised discovery method with biases geared toward partitioning objects into clusters that improve interpretability. The algorithm ITERATE employs: 1) a data ordering scheme and 2) an iterative redistribution operator to produce maximally cohesive and distinct clusters. Cohesion or intraclass similarity is measured in terms of the match between individual objects and their assigned cluster prototype. Distinctness or interclass dissimilarity is measured by an average of the variance of the distribution match between clusters. The authors demonstrate that interpretability, from a problem-solving viewpoint, is addressed by the intraclass and interclass measures. Empirical results demonstrate the properties of the discovery algorithm and its applications to problem solving View full abstract»
Cited by: Papers (9)
Modeling nonlinear systems by neural networks and fuzzy systems encounters problems such as the conflict between overfitting and good generalization and low reliability, which requires a great number of fuzzy rules or neural nodes and uses very complicated learning algorithms. A new adaptive fuzzy inference system, combined with a learning algorithm, is proposed to cope with these problems. First, the algorithm partitions the input space into some local regions by competitive learning, then it determines the decision boundaries for local input regions, and finally, based on the decision boundaries, it learns the fuzzy rule for each local region by recursive least squares (RLS). In the learning algorithm, the key role of the decision boundaries is highly emphasized. To demonstrate the validity of the proposed learning approach and the new adaptive fuzzy inference system, four examples are studied by the proposed method and compared with the previous results View full abstract»
Cited by: Papers (11)
The paper describes a framework for task sequence planning for a generalized robotic work cell. The AND/OR net provides a compact, distributed, domain-specific representation of geometric configurations of parts and devices in the work cell. The approach maintains a correspondence from geometric state information to task and motion plans and on-line discrete-event control that is not available in traditional action-based planners. The feasibility criteria for each AND/OR net transition guide the geometric reasoning required in the planning of feasible sequences. The resulting search space for plans is often much smaller (due to explicit representation of geometric constraints) than the state space of an action-based task planner. For purposes of analysis, the AND/OR net is mapped into a Petri net and the resulting Petri net is shown to be bounded and have guaranteed properties of liveness, safeness, and reversibility. In this form, the AND/OR net may be viewed as a Petri net synthesis tool in which the resulting Petri net representation may be used for on-line scheduling and control of the system View full abstract»
Cited by: Papers (5) | Patents (1)
A document retrieval system mainly consists of three components: document representation, user queries, and document evaluation. Each component may involve some uncertainties. Fuzzy set theory is a natural approach to coping with the representation of documents, queries, and the relevance of documents to a given query. The authors propose a fuzzy document retrieval model on the World Wide Web (WWW) environment to support conceptual queries. A flexible query expression is proposed to support different semantics of the queries. A concept network is adopted as the knowledge base to represent the relevance of the concepts. The concept network is explored from the WWW. Moreover, they also support neighborhood queries, which retrieve documents relevant to a document specified by a user. A system is currently being implemented to achieve these functions View full abstract»
Cited by: Papers (3)
Compartmental modeling is an approach to dynamic systems analysis that has proven useful in enhancing the understanding of biological and ecological systems. The compartmental approach emphasizes model conceptualization and is especially appropriate in applications in which quantitative behavioral data are difficult or expensive to obtain and in which qualitative understanding is the primary goal. In these applications, models are often developed by interdisciplinary teams. Domain experts contribute their understanding of component behaviors and interactions, while systems engineers guide the modeling process and insure the rigor of subsequent model-based analyses. The success of these applications clearly demands efficient and effective methods for transferring qualitative knowledge across disciplinary boundaries. The paper examines the expanded roles of knowledge acquisition, verification, and interpretation in the formulation of compartmental models. The objective is to illustrate that knowledge engineering, as currently practiced in the development of expert systems, is an essential ingredient in the compartmental approach to dynamic systems View full abstract»
Aims & Scope
Overview, tutorial and application papers concerning all areas of interest to the SMC Society: systems engineering, human factors and human machine systems, and cybernetics and computational intelligence.
Authors should submit human-machine systems papers to the IEEE Transactions on Human-Machine Systems.
Authors should submit systems engineering papers to the IEEE Transactions on Systems, Man and Cybernetics: Systems.
Authors should submit cybernetics papers to the IEEE Transactions on Cybernetics.
Authors should submit social system papers to the IEEE Transactions on Computational Social Systems.
Meet Our Editors
Dr. Vladimir Marik
(until 31 December 2012)