Abstract: A method for learning a person movement model includes obtaining a data stream having time series of measured multi-dimensional movement data of the person from at least one sensor; segmenting the data into time segments, segment corresponding to one movement step; storing each segment into a first storage section with a predetermined memory size, performing clustering of the segments by computing distances between the segments, learning a two-dimensional non-linear topology preserving embedded feature vector based on the computed distances, clustering the vectors, aligning segments of each cluster and averaging the aligned segments of for generating a prototype segment for each cluster; generating and storing a movement model by comparing the determined number of clusters of a current clustering with the determined number of clusters from the previous clustering stored in a second storage section.

Abstract: A method for learning a person movement model includes obtaining a data stream having time series of measured multi-dimensional movement data of the person from at least one sensor; segmenting the data into time segments, segment corresponding to one movement step; storing each segment into a first storage section with a predetermined memory size, performing clustering of the segments by computing distances between the segments, learning a two-dimensional non-linear topology preserving embedded feature vector based on the computed distances, clustering the vectors, aligning segments of each cluster and averaging the aligned segments of for generating a prototype segment for each cluster; generating and storing a movement model by comparing the determined number of clusters of a current clustering with the determined number of clusters from the previous clustering stored in a second storage section.

Abstract: A Neural Gas network used for pattern recognition, sequence and image processing is extended to a supervised classifier with labeled prototypes by extending a cost function of the Neural Gas network with additive terms, each of which increases with a difference between elements of the class labels of a prototype and a training data point and decreases with their distance. The extended cost function is then iteratively minimized by adapting weight vectors of the prototypes. The trained network can then be used to classify mass spectrometric data, especially mass spectrometric data derived from biological samples.

Abstract: A Neural Gas network used for pattern recognition, sequence and image processing is extended to a supervised classifier with labeled prototypes by extending a cost function of the Neural Gas network with additive terms, each of which increases with a difference between elements of the class labels of a prototype and a training data point and decreases with their distance. The extended cost function is then iteratively minimized by adapting weight vectors of the prototypes. The trained network can then be used to classify mass spectrometric data, especially mass spectrometric data derived from biological samples.