K. Pawelzik, J. Kohlmorgen, and K.-R. Muller. Annealed competition of experts for a segmentation and classification of switching dynamics. Neural Computation, 8(2):340--356, 1996.  Home/Search   Document Details and Download   Summary   Related Articles   Check

....then we nd out f 1 ; fm for modeling the time series. That is, y t = f r t (x t ) 8t = 1; l; 1) where l is the length of the time series. The index r t 2 f1; mg means functions alter according to r t . Many papers have tackled this issue by using neural networks. See [10, 9, 4] and references therein. Recently an approach using support vector machines is [3] One more dicult problem is how to model time series from drifting dynamics. For this type of time series, y t can not be determined by one single function but a mixture of several functions. We can assume that ....

....problems. Unfortunately, as the goal is still to solve a non convex problem, sometimes our algorithm can fall into local minima. A possible solution is to consider in the beginning that series are not mixed but only switching dynamics. Then by the competition of experts using approaches in [3, 10, 9, 4], we obtain a better initial set of p i . Then the algorithm is less likely to be trapped into local minima. 4.2 Scaling Models (Functions) Another possible problem of our approach is that f i (x) may have values in a larger range than those of y t provided. To make f i (x) in the same range of ....

K. Pawelzik, J. Kohlmorgen, and K.-R. Muller. Annealed competition of experts for a segmentation and classi cation of switching dynamics. Neural Computation, 8(2):340-356, 1996.

....106, Taiwan (cjlin csie.ntu.edu.tw) Ruby C. Weng Department of Statistics National Chenechi University Taipei 116, Taiwan Abstract We present a framework for the unsupervised segmentation of switching dynamics using support vector machines. Following the architecture by Pawelzik et al. [21] where annealed competing neural networks were used to segment a non stationary time series, in this article we exploit the use of support vector machines, a well known learning technique. First, a new formulation of support vector regression is proposed. Second, an expectation maximization ....

.... the fuzzy c regression models (FCRM) 5] with applications to models with known forms and known distribution of the noise term, a combination of supervised and unsupervised learning using hidden Markov models based on neural networks [4] and other competing neural networks approaches in [14] [21], 8] 11] 9] In [21] 14] annealed competing neural networks were used to segment a non stationary time series, where non stationarities are caused by switching dynamics. This method is called Annealed Competition of Experts (ACE) Unlike the mixtures of experts architecture [7] which ....

[Article contains additional citation context not shown here]

K. Pawelzik, J. Kohlmorgen, and K.-R. M uller. Annealed competition of experts for a segmentation and classification of switching dynamics. Neural Computation, 8(2):340--356, 1996.

....of multiple models methods. There is special interest in the development of clustering, classification, prediction and parameter estimation algorithms for time series ( dynamic ) problems. Some remarkable efforts in this direction include partition algorithms [10, 19] mixtures of experts [5, 12, 13, 14, 15, 16, 25], ensembles of neural networks [3, 7, 26] trees of neural networks [17, 32] threshold models [35] Takagi Sugeno fuzzy models [34] and much more. For an extensive bibliographical coverage see the books [22, 31] The predictor architecture proposed in this paper is modular in the sense that it ....

K. Pawelzik, J. Kohlmorgen and K.-R. Muller, "Annealed competition of experts for a segmenta- tion and classification of switching dynamics", Neural Computation, 1996, vol.8, pp.340-356.

....general modular systems. The main feature of such systems is that each module is specialized in describing the input output behavior of a particular source. Methods of this type are used extensively in the neural and control literature and are presented under various names such as local experts [7, 8, 17], local controllers [2] multiple models [13, 15, 16] or regime models [3, 4, 5, 9] The unsupervised problem is considerably harder than the supervised one. In the latter case, labeled data are available for an off line training phase. i.e. two sequences of data points are available; the first ....

....a very rapid rate. In such a case, at the initial stages of the data allocation process, before a predictor has the chance to specialize in a single source, it will collect a mixed dataset. It should be added that slow switcAing is essential in time series modeling and is usually assumed to hold [11, 17]. Of course, this is also a relative issue, i.e. in situations where one may suspect fast switching, it will be necessary to increase the sampling rate, so that the time series switches slowly in relation to the learning algorithm. 4 Data Allocation Algorithms We now present a parallel data ....

K. Pawelzik, J. Kohlmorgen and K.R. Muller. "Annealed competition of experts for a segmentation and classification of switching dynamics", Neural Computation, vol.8, pp.357-372, 1996.

....receptive field locally weighted regression was presented by Schaal and Atkeson [210] This algorithm consists of linear regression experts and a set of Gaussian receptive field units which gate the expert outputs. Each of the experts are trained independently in this algorithm. Pawelzik et al. [171] consider the problem of segmentation of a time series by an ensemble of radial basis functions, which they describe as a mixture of experts without a gating network. They derive a weighting factor for each expert in the ensemble based on its likelihood of explaining the current sample in the time ....

Pawelzik, K., Kohlmorgen, J. and Muller, K. R. [1996], `Annealed competition of experts for a segmentation and classification of switching dynamics', Neural Computation 8(2), 340--356.

....of Statistics National Chenechi University Taipei 116, Taiwan Abstract We present a framework for the unsupervised segmentation of time series using support vector regression. It applies to non stationary time series which alter in time. We follow the architecture by Pawelzik et al. [13] which consists of competing predictors. In [13] competing Neural Networks were used while here we exploit the use of Support Vector Machines, a new learning technique. Results indicate that the proposed approach is as good as that in [13] Di#erences between the two approaches are also ....

....Taipei 116, Taiwan Abstract We present a framework for the unsupervised segmentation of time series using support vector regression. It applies to non stationary time series which alter in time. We follow the architecture by Pawelzik et al. 13] which consists of competing predictors. In [13] competing Neural Networks were used while here we exploit the use of Support Vector Machines, a new learning technique. Results indicate that the proposed approach is as good as that in [13] Di#erences between the two approaches are also discussed. I. Introduction Recently support vector ....

[Article contains additional citation context not shown here]

K. Pawelzik, J. Kohlmorgen, and K.-R. Muller. Annealed competition of experts for a segmentation and classification of switching dynamics. Neural Computation, 8(2):340--356, 1996.

No context found.

Pawelzik, K., Kohlmorgen, J., Muller, K.-R. (1996). Annealed Competition of Experts for a Segmentation and Classification of Switching Dynamics. Neural Computation 8(2), 340--356.

No context found.

Pawelzik, K., Kohlmorgen, J., Muller, K.-R. (1996). Annealed Competition of Experts for a Segmentation and Classification of Switching Dynamics. Neural Comput 8(2), 340--356.

No context found.

Pawelzik, K., Kohlmorgen, J., Muller, K.-R. (1996). Annealed Competition of Experts for a Segmentation and Classification of Switching Dynamics. Neural Computation 8(2), 340--356.

No context found.

Pawelzik, K., Kohlmorgen, J., Muller, K.-R. (1996). Annealed Competition of Experts for a Segmentation and Classification of Switching Dynamics. Neural Comput 8(2), 340--356.

No context found.

Pawelzik, K., Kohlmorgen, J., Muller, K.-R. (1996). Annealed Competition of Experts for a Segmentation and Classification of Switching Dynamics. Neural Computation 8(2), 340--356.

No context found.

K. Pawelzik, J. Kohlmorgen and K.-R. Muller, "Annealed competition of experts for a segmentation and classification of switching dynamics," Neural Computation, vol. 8, pp. 340--356, 1996.

No context found.

Pawelzik, K., Kohlmorgen, J., Muller, K.-R. (1996). Annealed Competition of Experts for a Segmentation and Classification of Switching Dynamics, Neural Computation, 8:2, 340-356.

No context found.

Pawelzik, K., Kohlmorgen, J., Muller, K.-R. (1996). Annealed Competition of Experts for a Segmentation and Classification of Switching Dynamics, Neural Computation, 8:2, 342-358.

No context found.

Pawelzik, K., Kohlmorgen, J., Muller, K.-R. (1996). Annealed Competition of Experts for a Segmentation and Classification of Switching Dynamics. Neural Computation 8(2), 340--356.

No context found.

Pawelzik, K., Kohlmorgen, J., Muller, K.-R. (1996). Annealed Competition of Experts for a Segmentation and Classification of Switching Dynamics. Neural Computation 8(2), 340--356.

No context found.

Pawelzik, K., Kohlmorgen, J., Muller, K.-R., Annealed Competition of Experts for a Segmentation and Classifi- 11 cation of Switching Dynamics, to appear in Neural Computation (1995)

No context found.

K. Pawelzik, J. Kohlmorgen, and K.-R. Muller. Annealed competition of experts for a segmentation and classification of switching dynamics. Neural Computation, 8(2):340--356, 1996.

No context found.

K. Pawelzik, J. Kohlmorgen, and K.-R. Muller. Annealed competition of experts for a segmentation and classification of switching dynamics. Neural Computation, 8(2):340--356, 1996.

No context found.

K. Pawelzik, J. Kohlmorgen, and K.-R. Muller. Annealed competition of experts for a segmentation and classi cation of switching dynamics. Neural Computation, 8(2):342-358, 1996.

No context found.

Pawelzik, K., Kohlmorgen, J., Muller, K.-R. (1995). Annealed Competition of Experts for a Segmentation and Classification of Switching Dynamics, submitted to Neural Computation.

No context found.

K. Pawelzik, J. Kohlmorgen, and K.-R. Muller. Annealed competition of experts for a segmentation and classi cation of switching dynamics. Neural Computation, 8(2):342-358, 1996.

No context found.

K. Pawelzik, J. Kohlmorgen, and K.-R. Muller, "Annealed competition of expert for segmentation and classification of switching dynamics," Neural Computation, vol. 8, no. 2, pp. 340-356, February 1996.

No context found.

K. Pawelzik, J. Kohlmorgen, and K.-R. Muller. Annealed competition of experts for a segmentation and classification of switching dynamics. Neural Computation, 8(2):340--356, 1996.

No context found.

K. Pawelzik, J. Kohlmorgen and K.R. Muller. 1996. "Annealed Competition of Experts for a Segmentation and Classification of Switching Dynamics", Neural Computations, vol.8, pp.340-356.

First 50 documents  Next 50

Online articles have much greater impact   More about CiteSeer.IST   Add search form to your site   Submit documents   Feedback  

CiteSeer.IST - Copyright Penn State and NEC