This paper concerns a greedy EM algorithm for t-mixture modeling, which is more robust than Gaussian mixture modeling when a typical points exist or the set of data has heavy tail. Local Kullback divergence is used to...
详细信息
This paper concerns a greedy EM algorithm for t-mixture modeling, which is more robust than Gaussian mixture modeling when a typical points exist or the set of data has heavy tail. Local Kullback divergence is used to determine how to insert new component. The greedy algorithm obviates the complicated initialization. The results are comparable to that of split-and-merge EM algorithm while the proposed algorithm is faster. Also the by product of a sequence of mixture models is useful for model selection. Experiments of synthetic data clustering and unsupervised color image segmentation are given.
This paper investigates the problem of how to carry out 3D scene reconstruction from multiple views. It improves the algorithms of projective reconstruction based on the homography induced by the infinite plane which ...
详细信息
ISBN:
(纸本)0780378652
This paper investigates the problem of how to carry out 3D scene reconstruction from multiple views. It improves the algorithms of projective reconstruction based on the homography induced by the infinite plane which should have 4 points on a reference plane visible in all views given by Hartley and Rother et al., and proposes a new linear algorithm based on 3 points on a reference plane visible in all views. It avoids the difficult task of determining whether 4 object points are coplanar or not, because 3 points which are not collinear just determine a plane.
The patterns of EEG changes with the mental tasks performed by the subject. In the field of EEG signal analysis and application, the study to get the patterns of mental EEG and then to use them to classify mental task...
详细信息
ISBN:
(纸本)953184061X
The patterns of EEG changes with the mental tasks performed by the subject. In the field of EEG signal analysis and application, the study to get the patterns of mental EEG and then to use them to classify mental tasks has the significant scientific meaning and great application value. But for the reasons of different artifacts contained in EEG, the pattern detection in EEG produced from normal mental states is a very difficult problem. In this paper, independent component analysis is applied to EEG signals collected from different mental tasks .The experiment results show that when one subject performs a single mental task in different trails, the independent components of EEG are very similar. It means that the independent components can be used as the mental EEG patterns to classify the different mental tasks.
It usually needs complicated nonlinear operations to get the characteristics from the raw information inputted, and it is very difficult to find this kind of algorithm directly. The geometrical meaning of the multilay...
详细信息
ISBN:
(纸本)0780375084
It usually needs complicated nonlinear operations to get the characteristics from the raw information inputted, and it is very difficult to find this kind of algorithm directly. The geometrical meaning of the multilayer perceptron's neuron model indicates that classifying samples according to the requirements by constructing neural networks is equal to finding a collection of domains with which vectors of the preset sample sets are partitioned. But in some applications, such as time series forecasting including stock share forecasting, due to their preset sample sets may contain some exceptions and erroneous results, it is desired to introduce some self-adjusting and probabilistic decision-making mechanism to enhance the accuracy of classification. At the same time the mechanism can reduce the size of neural networks and speed up the recognition process. We discuss a self-adjusting and probabilistic decision-making mechanism for the covering algorithm. Based on the method, we developed a self-adjusting and probabilistic decision-making classifier and applied the software package to forecast the share index of Shanghai's stock market.
Panoramic mosaics methods based on an 8-parameter planar homography matrix have to overcome the accumulated errors when a sequence of images loops back on itself. Usual methods are computationally intensive, and canno...
详细信息
ISBN:
(纸本)9628576623
Panoramic mosaics methods based on an 8-parameter planar homography matrix have to overcome the accumulated errors when a sequence of images loops back on itself. Usual methods are computationally intensive, and cannot ensure complete consistency of homographies. The paper presents a simple method which does not require the consistency of homographies. The method mainly exploits an un-calibrated image perspective interpolation technique (B. Yuan et al., 1998). It is therefore simple to calculate and easy to implement.
暂无评论