Search

A novel method of classification and current progress of work is presented in this paper. Its concept is based on application of curved and smooth surface being Riemannian manifold as the classifier. The model in the form of such a manifold is created using samples from a training set which are points in multidimensional linear space. This model is a structure where the greater group of points curves the surface around it more than a smaller group. Therefore if a new, unknown sample appears in such a surface, it will "fall" on an object and gets a label of object which it has fallen on. In this paper we present the progress of work leading to implementation of the first stage of the training phase: finding the best metric tensor for the set of points being feature vectors of samples from the training set reduced to minimization problem. Finally, the perspective of development of Riemannian Classifier was shown. [ABSTRACT FROM AUTHOR]

The authors present an application of Grassmann manifold to the evaluation of human dynamic balance based on the time series representing movements of hip, knee and ankle joints in the sagittal, frontal and transverse planes. Time series were extracted from gait sequences which were recorded in the Human Motion Laboratory (HML) of the Polish-Japanese Academy of Information Technology in Bytom, Poland using the Vicon system. [ABSTRACT FROM AUTHOR]

Medical centres are obliged to store past medical records, including the results of quality assurance (QA) tests of the medical equipment, which is especially useful in checking reproducibility of medical devices and procedures. Analysis of multivariate time series is an important part of quality control of NMR data. In this work we proposean anomaly detection tool based on Reiteration of Hankel Singular Value Decomposition method. The presented method was compared with external software and authors obtained comparable results. [ABSTRACT FROM AUTHOR]

Algorithms of tracking and action recognition are still under development and many problems still have to be solved. New methods are usually tested on available benchmarks with defined actions and human behavior however such approach has many limitations. For that reason the authors proposed new procedure of generating random action instances using on graph-based scenarios. Such idea can be applied in creation of different datasets as well as in simulations. [ABSTRACT FROM AUTHOR]