374 results about "Transition matrices" patented technology

Systems and methods to process and translate pinyin to Chinese characters and words are disclosed. A Chinese language model is trained by extracting unknown character strings from Chinese inputs, e.g., documents and/or user inputs/queries, determining valid words from the unknown character strings, and generating a transition matrix based on the Chinese inputs for predicting a word string given the context. A method for translating a pinyin input generally includes generating a set of Chinese character strings from the pinyin input using a Chinese dictionary including words derived from the Chinese inputs and a language model trained based on the Chinese inputs, each character string having a weight indicating the likelihood that the character string corresponds to the pinyin input. An ambiguous user input may be classified as non-pinyin or pinyin by identifying an ambiguous pinyin/non-pinyin ASCII word in the user input and analyzing the context to classify the user input.

A method for categorizing multiple users of a website based on their respective transition patterns between successive types of user activities is implemented at a computing device having one or more processors and memory. After receiving one or more log files from a data source, the computing device generates respective event transition matrixes for multiple users in accordance with the users' event records such that each element of each respective event transition matrix includes a probability of an occurrence of two consecutive event types A and B for the same user. Based on an analysis of the multiple users' event transition matrixes, the computing device categorizes the multiple users into at least two distinct groups of users and performs one or more operations to the website's user accounts associated with one or more users from one of the groups.

Systems, methods, and computer readable storage media are described that can provide a multi-level hierarchical framework to progressively associate detection responses, in which different methods and models are adopted to improve tracking robustness. A modified transition matrix for the Hungarian algorithm can be used to solve the association problem that considers not only initialization, termination and transition of tracklets but also false alarm hypotheses. A Bayesian inference approach can be used to automatically estimate a scene structure model as the high-level knowledge for the long-range trajectory association.

Methods and apparatus are provided for generating a set of restrictive costs associated with directional paths between selected nodes in a group of nodes of a network system wherein each node in the group is connected by a link to at least one other node in the group, and each link has a pair of directional costs associated therewith such that each link can be considered as a pair of oppositely-directed, unidirectional links each of which extends from a source node to a destination node and has a said directional cost associated therewith. The method allows asymmetric costs, rather than merely symmetric costs, to be considered, and can be applied, for example, to generate the transition matrix for a PNNI peer group.

The invention brings forward a rapid and automatic mosaic technology of aerial video in a search and tracking system. The technology provided by the invention comprises the following steps: two continuous frames of color images in an aerial video sequence are converted to grayscale images; three-valued processing is carried out on the grayscale images by an image three-valued method, edge feature information images are extracted by an edge detection sobel operator and feature points are extracted by an Harris algorithm; feature point matching is completed by an optical flow pyramid algorithm; optimal homography matrix of the color images is calculated by combining a DLT algorithm and a RANSAC algorithm; panoramic image mosaics is completed after image correction of the color images by correction geometric transition matrix; and dynamic regulation of image three-valued upper and lower thresholds and a threshold of the edge detection sobel operator is carried out during the process of panoramic image mosaics. According to the invention, image mosaic accuracy and robustness are raised.

The present invention provides a method and an apparatus for generating pseudo-random numbers with very long periods and very low predictability. A seed random sequence is extended into a much longer sequence by successive iterations of matrix operations. Matrices of candidate output values are multiplied by non-constant transition matrices and summed with non-constant offset matrices; the result is then processed through one or more modulus operations, including non-constant modulus operators, to generate the actual output values. The invention also includes the possibility of introducing non-invertible matrices into the operations. The invention creates final results that are equidistributed over large samples. Secondary pseudo-random and other processes determine the non-constant transition matrices, offset matrices, and modulus operators.

The invention discloses a sample categorization method based on a depth neural network. The open set category mining and extending method based on a depth neural network comprises steps of using a sample set comprising defined category samples to train a categorized model to be extended, obtaining categorization threshold value information, sending a sample set comprising undefined category samples into the categorization model to be extended, determining at least part of the undefined category samples according to the categorization threshold value information of the categorization model to be extended, artificially marking the undefined category samples, adding a number of columns of a weight transfer matrix in a categorization layer of the depth nerve network in order to increase a total number of model recognition categories, wherein the added weight columns comprise first information associated with global categorization and second information associated with connection between categories and using the undefined category samples which are artificially marked to increase the models which already finish training and updating. The open set category mining and extending method based on the depth neural network and the device thereof extend the depth neural network through modifying a depth neural network categorization layer weight transfer matrix, dynamically increases the number of the recognized categories so as to process the open set recognition problem and can be applied to a scene which is closer to a real scene.

A computer implemented method, system, and computer usable program code for classifying a data stream using high-order models. The data stream is divided into a plurality of data segments. A classifier is selected for each of the plurality of data segments. Each of a plurality of classifiers is clustered into states. A state transition matrix is computed for the states. The states of the state transition matrix specify one of the high-order models for classifying the data stream.

The image process method comprises: A. multiplying these linear-calculation matrixes, including the transition matrix to transfer pixel RGB data into color space with separated brightness and chroma into a unique processing matrix; and B. multiplying the said unique matrix to pixel EGB data to obtain the pixel point data with separated brightness and chroma, then taking Gamma correction for brightness component. This invention increases processing speed, reduces difficulty on chip design and production, and decreases power consumption and cost.

The invention discloses a transition probability adaptivity-based interacting multiple model target tracking method. A motion track measurement value of a target is collected through a sensor and a motion state model set of the target is built; according to priori knowledge, a probability of an initial model and a transition probability matrix of the model are set; a state value is subjected to input interaction; an interactive value serves as an input value of filtering in the next step; parallel filtering is performed through filters under sub-models to obtain filter values under different models, and a probability of each model is updated; according to updated model change rates, a state transition matrix is corrected by adopting a hyperbolic sine inverse function to realize adaptivityof the transition probability matrix; and finally the filter values of the sub-models are subjected to weighted summation, thereby realizing target tracking. The adaptivity of the state transition matrix of an interacting multiple model algorithm is realized; and maneuvering and non-maneuvering target tracking can be realized to obtain a real motion track of the target, thereby improving trackingperformance of the interacting multiple model-based target tracking method.

For information encryption and decryption the apparatus uses the same hardware keys, which have the transition matrix of a key-automaton with no output signal and with an initial state and a final state burnt in. To each character in the character set of the plaintext message there is one or more final states of the key-automaton assigned. During encryption the message is read in sequentially character by character and the key automaton assigns to each character a random character string, whose length is adjustable length within a given length range. The process is the following: for each character in the message the apparatus generates a character string of adjustable length and with no initial- and end markers, which takes the apparatus from its current state into the final state that corresponds to the subsequent character of the message. The apparatus creates the encrypted message by linking these character strings together. The encrypted message can be decrypted using the same apparatus.

The invention provides a panoramic animation jumping method and a system. The method comprises steps: when scene jumping is triggered, a transition matrix generated in advance is used to carry out model modification processing on the model matrix of the current scene, and transparency modification processing and stretch processing are carried out on the current scene; when the current scene starts to move or stretch, a thumbnail of a target scene and pictures within the visible range are loaded, and the target scene is rendered after loading is completed; and after moving and stretching of the current scene are completed and rendering of the target scene is completed, the target scene is shown. According to the technical scheme provided by the invention, natural and smooth switching from the current scene to the target scene can be realized.

A system for decoding digital signals subjected to block coding includes a post-processor that corrects the codewords affected by error, identifying them with the most likely sequence that is a channel sequence and that satisfies a syndrome check. The post-processor is a finite-state machine described by a graph that represents the set of error events. The post-processor evolves in steps through subsequent transition matrices, deleting at each step the paths that accumulate an invalid number of error events or an excessive number of wrong bits, paths that accumulate a total reliability higher than a given threshold, paths with an invalid check on the received sequence, and paths that reveal an invalid syndrome after having reached a maximum number of events.

The invention discloses a circular synthetic aperture radar imaging method, belonging to the technical field of electronic signal processing. The invention relates to a spatial remote sensing and air-to-ground observation information processing technology, in particular to an onboard circular synthetic aperture radar imaging technology. According to the imaging method, an original oblique plane echo is mapped to the ground plane by solving the inverse of a system kernel function, and the distribution of signals in a wave number domain is obtained by multiplying an azimuth frequency domain with a reference function, so that plane wave approximation is avoided, the problem that the size of an imaging area of the traditional polar format algorithm is limited can be solved, and the imaging of a large imaging scene can be realized; pseudo-polar coordinates are used for replacing rectangular coordinates as an intermediate interpolation transition matrix, and the distribution density characteristic of the signals in the wave number domain is considered, so that the interpolation precision is higher, the obtained image resolution is higher, and the high resolution imaging can be realized; the imaging process only involves one-dimensional interpolation, and fast algorithms such as chirp z transform (CZT) and inverse fast Fourier transform (IFFT) are adopted, so that the method has very high computational efficiency and can realize rapid imaging.

The invention discloses a semi-supervised dimensionality reduction method for high dimensional data clustering. The method comprises the following steps: (1) constructing a sample characteristic matrix; (2) constructing a constraint matrix; (3) constructing an iterative equation set and an iterative output transition matrix; and (4) obtaining a sample characteristic matrix after an operation of dimensionality reduction is performed. In the invention, through adding part of known class information as a constraint in the process of decomposing the sample characteristic matrix, and using an idea of concept decomposition, a coefficient matrix obtained by decomposition is used as the low-dimensionality representation of a high-dimensionality sample characteristic matrix, and when the low-dimensionality matrix is applied to clustering analysis, the clustering analysis becomes simple and effective; meanwhile, in the invention, data subjected to dimensionality reduction has a good interpretability; and compared with a dimensionality reduction method in the prior art, by using the dimensionality reduction method disclosed by the invention, the discrimination capacity of the clustering analysis can be further improved.

Methods and systems are provided for efficiently computing page rankings of web pages or other interconnected objects. The rankings are produced by efficiently computing a principal eigenvector of a page ranking transition matrix. The methods and systems provided herein can be used to produce page rankings in a distributed and/or incremental manner, and can be used to allocate computing resources to processing page rankings for those pages that most demand them.

A perceptive global positioning device is disposed on or carried by a carrier (vehicle, pedestrian, etc.) for receiving positioning signals transmitted from satellites, to detect various behavioral states related to movement and correct errors in positioning data for positioning the carrier. The positioning device includes a sensor for receiving positioning data related to the global position of the carrier transmitted from satellites, a database for storing the positioning data received by the sensor and a preset behavioral transition matrix of the carrier, an analysis module for analyzing the positioning data and behavioral transition matrix to obtain predictable behavior data, and a correction module for comparing the predictable behavior data with the positioning data so as to correct errors of the positioning data of the carrier, thereby overcoming the defect of inaccurate data as encountered in the prior art.

A method segments a video. Audio frames of the video are classified with labels. Dominant labels are assigned to successive time intervals of consecutive labels. A semantic description is constructed for sliding time windows of the successive time intervals, in which the sliding time windows overlap in time, and the semantic description for each time window is a transition matrix determined from the dominant labels of the time intervals. A marker is determined from the transition matrices, in which a frequency of occurrence of the marker is between a low frequency threshold and a high frequency threshold. Then, the video is segmented at the locations of the markers.

Predicting a malfunction of a component of a unit includes providing a transition matrix of a parameter of the component, wherein the transition matrix includes for a number of discrete value states of the parameter probabilities to switch from one discrete value state to another within a certain time period; providing the conditional probability distribution for the malfunction given the discrete value states; providing a current discrete value state of the parameter; determining a conditional probability distribution of the discrete value states given the current discrete value state for a future point in time based on the current discrete value state and on the transition matrix by use of a Markov chain; and determining a probability for the malfunction for the future point in time based on the conditional probability distribution of the discrete value states for the future point in time and the conditional probability distribution for the malfunction.

The invention discloses a word segmentation processing method and device, a mobile terminal and a computer readable storage medium. The method comprises the following steps of: when a to-be-segmentedstatement is obtained, determining a target language type corresponding to the to-be-segmented statement; respectively first feature vectors corresponding individual characters, second feature vectorscorresponding to two words and third feature vectors corresponding to proper nouns in the to-be-segmented statement; determining current fourth feature vectors of the individual characters accordingto the first feature vectors, the second feature vectors and the third feature vectors; and carrying out word segmentation on the to-be-segmented statement according to a preset Chinese character label transfer matrix and the current fourth feature vectors of the individual characters. According to the method, word segmentation is carried out on to-be-segmented statements according to target language types corresponding to the to-be-segmented statements, so that the correctness of carrying out word segmentation on to-be-segmented statements in various language types is improved; and proper resources can be loaded according to requirements, so that storage spaces of mobile terminals are saved and the user experience is improved.

The invention aims to provide a system-level error parameter field calibration method of a strapdown inertial navigation optical fibre gyro. The system-level error parameter field calibration method comprises the following steps: an optical fibre gyro strapdown inertial navigation system is communicated with auxiliary attitude measuring equipment, and the optical fibre gyro strapdown inertial navigation system can carry out initial alignment; the output values of the optical fibre gyro and the auxiliary attitude measure equipment are collected, the output value of the optical fibre gyro is converted into the angular speed of a coordinate system of the gyro relative to an inertial system and the angular speed is substituted into a strapdown inertial navigation equation to obtain a transition matrix of the optical fibre gyro relative to the local geographic coordinate system; according to a real-time attitude angle provided by the auxiliary attitude measuring equipment, a true attitude transition matrix T is obtained and an attitude error matrix and a system real-time attitude error angle are obtained further; and the attitude error matrix and the system real-time attitude error angle are substituted into a system level filtering state calibration equation of the optical fibre gyro to estimate each error parameter so as to finish identification. The system-level error parameter field calibration method provided by the invention has strong flexibility in the selection of a calibration path, and greatly reduces the dimension of a filter state equation.

The invention belongs to a quick processing method for a sensor antenna array received signals. The quick processing method comprises dividing subarrays, establishing signal matrixes of the subarrays, establishing autocorrelation matrixes and cross correlation matrixes, establishing signal subspace matrixes, establishing signal characteristic matrixes and inverse matrixes thereof, establishing transition matrixes, establishing rotation invariable relation matrixes, and finally obtaining incoming wave angles of different signals. A method of directly establishing the rotation invariable relation matrixes is adopted, and a method of processing and establishing an array rotation invariable relation through the multiplication and addition of the matrixes is only used, so that a sensor antennaarray quickly tests incoming wave directions of a single or a plurality of wireless transmission signals; and compared with the prior art, the method has the advantages that the processing speed is high, the accuracy of testing the incoming wave direction angles is obviously improved particularly when the signal to noise ratio is low, and the method has the characteristics of low requirements forequipment and hardware performance, quick and accurate test on target incoming wave directions, high accuracy, short flow, low operating cost, wide application range and the like.

The invention provides a method for determining the coordinates of the tip of a bucket through using a GNSS and a tilt sensor. The method comprises the following steps that: step (1) main antenna GNSSgeodetic latitude and longitude coordinates and a geodetic azimuth are acquired; step (2) the geodetic latitude and longitude coordinates are converted into engineering coordinates, and the geodeticazimuth is converted into an engineering coordinate azimuth; step (3) a vehicle body coordinate system-to-engineering coordinate system transition matrix is calculated according to the numerical values of vehicle body sensors and the coordinate azimuth; step (4) the vehicle body coordinate system coordinates of the tip of the bucket are calculated according to the numerical values of the sensors on an upper arm, a forearm and a rocker and a mechanical model; and step (5) the engineering coordinates of the tip of the bucket are calculated according to the transition matrix and the vehicle bodycoordinate system coordinates. According to the method of the invention, high-accuracy Beidou navigation and positioning technologies are adopted; a GNSS coordinate-to-engineering coordinate transformation algorithm is adopted to convert the GNSS coordinates (B, L, H) into the engineering coordinates (N, E, H), so that the engineering coordinates of a GNSS antenna can be obtained in real time.