36 results about "Probabilistic graph" patented technology

A location system for locating and determining the motion and velocity of a wireless device. The methods include direct inferences about whether a device is in motion versus static based on a statistical analysis of the variation of radio signal strengths over time. The system is trained according to a sparse set of identified locations from which signal strengths are measured. The system uses the signal properties of the identified locations to interpolate for a new location of the wireless device. The system uses a probabilistic graph where the identified locations of the floor plan, expected walking speeds of pedestrians, and independent inference of whether or not the device is in motion are used to determine the new location of the device.

A clustering apparatus of probabilistic graphs, includes a center selection unit configured to select one or more centers among the probabilistic graphs. The clustering apparatus further includes a center determination unit configured to calculate a minimum bound and a maximum bound of a distance with respect to each of the centers, for each of the probabilistic graphs, and determine a center, among the centers, to which the probabilistic graphs are to be allocated based on the minimum and maximum bounds. The clustering apparatus further includes a clustering unit configured to allocate the probabilistic graphs to the center to generate one or more clusters.

A clustering apparatus of probabilistic graphs, includes a center selection unit configured to select one or more centers among the probabilistic graphs. The clustering apparatus further includes a center determination unit configured to calculate a minimum bound and a maximum bound of a distance with respect to each of the centers, for each of the probabilistic graphs, and determine a center, among the centers, to which the probabilistic graphs are to be allocated based on the minimum and maximum bounds. The clustering apparatus further includes a clustering unit configured to allocate the probabilistic graphs to the center to generate one or more clusters.

The invention provides an environment modeling method applicable to navigation of automatic piloting vehicles. One of the key problems needing to be overcome for the navigation of automatic piloting vehicles is to modeling an environment in which a vehicle pilots, to distinguish sceneries in the environment and to convert environmental information into parameterized information which can be used for intelligent obstacle avoidance and path planning of an automatic piloting vehicle. According to the invention, a laser sensor is provided at the front of the automatic piloting vehicle; a series of steps like measurement of spatial distance between the center of the laser sensor and the center of the vehicle are carried out; the whole environment is modeled by utilizing laser point sequences acquired in the driving process of the vehicle. Displacement and course angles of the vehicle are calculated by registering observation of the laser sensor at adjacent sampling time, which is a self-contained scheme and can effectively avoid the problem of LOS (lost of signals) in extreme environments in similar methods which employ a scheme bases on a constellation system; in the method provided in the invention, the laser point sequences in laser beams are processed with a method of inference based on a probabilistic graph model, which enables geometrical characteristics of scenery contours to be utilized and managed intelligently, and therefore, higher accuracy in environment modeling is obtained in the invention.

The invention discloses a game-based link prediction method and system in an exchangeable graph. The method adopts steps: (1) Obtain a data set, and represent the elements in the data set through a graph. The graph is composed of a node set and an edge set; The nodes represent the entities to be predicted, and the edges represent the relationship between the entities to be predicted. The nodes and edges of the data set meet the requirements of the exchangeable graph; (2) use a game-based network evolution model to filter the For edges that do not meet the set requirements, use the filtered graph as the training set; (3) use the training set to iteratively train the probability graph model to obtain a probability graph model with optimal model parameters; (4) use (3) A probabilistic graph model with optimal model parameters performs link prediction. The present invention can be applied in recommendation systems, risk assessment, system planning or social networks. The relationship between entities and entities meets the requirements of exchangeable graphs, and the relationship between entities The relationship prediction is fast and accurate.