347 results about "Transition probability matrix" patented technology

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.

The invention provides a comment analysis method based on word vectors and syntactic characteristics in the field of data analysis. The comment analysis method comprises the steps of obtaining commodity page comment data of an e-commerce website; preprocessing the acquired target data set; extracting a appendix lexical set provided by Hownet and NTU to form a basic emotion dictionary; carrying outword vector training on the obtained preprocessed data set through a Word2Vec tool; establishing a probability transfer matrix by using the semantic similarity matrix; carrying out core sentence rule-based processing on the obtained commodity comment text; carrying out preprocessing on the obtained text without the redundancy; performing part-of-speech extraction (commodity attributes, negative words, degree words and sentiment words) evaluation matching on the obtained dependency relationship pairs; combining the evaluation matching pair with an emotion dictionary, subjecting evaluation objects to appraisal value calculation and quality sorting, and finally, realizing the evaluation objects through a visual interaction interface, so that accurate, real-time, automatic and convenient processing and analysis on commodity comment data are realized, and the method can be used in an e-commerce platform.

A method for generating an optimized transition probability matrix (OTPM) is provided. The method is performed using a computer system coupled to a database. The method includes storing in the database financial data including obligor credit ratings, generating multi-period empirical transition probability matrices (ETPMs) for a selected time horizon using the financial data stored within the database, generating a mathematical expression to minimize a difference between target ETPM values and candidate OTPM values, and calculating the OTPM from the generated mathematical expression and the financial data stored within the database, wherein the calculated OTPM includes a first set of optimized transition probability values for predicting a likelihood that a credit rating of an obligor will migrate from one credit state to another credit state during a first time interval in the future.

The invention discloses a method for predicting a time sequence of number of people served by base stations based on space-time transfer probabilities of mobile phones. The method comprises the following steps: calculating the total number of people within base station service areas of the mobile phones within an equal time period by using space-time orbit data of the mobile phones; dividing people moving orbits by using the space-time orbit data of the mobile phones, and calculating the number of people coming back and forth between the base stations within adjacent time periods in a research area; based on related theory of Bayesian and Markov chains, calculating the transfer probabilities of mobile phone users within target base stations to occur in the base stations at a current moment according to historical data; calculating the transfer probabilities of the mobile phone users within the target base stations to occur in the base stations within different time periods to construct a complete space-time transfer probability matrix in the research area; and predicting the number of people served within the base station ranges of the mobile phones in the research area with the relatively stable total number of people by using the complete space-time transfer probability matrix. The method disclosed by the invention has the advantages of low data acquisition cost, simple model structure and high prediction efficiency.

The invention discloses a positioning method. A wireless service terminal collects positions of a plurality of sample points and corresponding wireless signal feature values and builds a probability distribution model of the wireless signal feature values on the positions of the plurality of sample points, wherein the number of the sample points N is a positive whole number. A Markov single-step position transition probability matrix is built. The wireless service terminal conducts a m-time positioning according to the probability distribution model and the Markov single-step position transition probability matrix, wherein m is more than or equal to 1. Further, the Markov single-step position transition probability matrix can achieve necessary updates according to practical situations. The positioning method effectively uses information of the wireless signals in the environment, achieves real-time positioning of pre-positioning terminals and further improves accuracy of the positioning.

The invention provides a self-adaptive target tracking information filtering algorithm of a maneuvering strategy. The self-adaptive target tracking information filtering algorithm comprises the following steps: firstly, establishing a target tracking model of multiple maneuvering strategies and multiple motion models, and then, carrying out the target tracking information filtering algorithm under the multiple maneuvering strategies and multiple motion models to obtain a target tracking trajectory. According to the self-adaptive target tracking information filtering algorithm, a maneuvering strategy concept is introduced, the target tracking model of the multiple maneuvering strategies and multiple motion models is established, the maneuvering strategy transition probability matrix is corrected in real time by the error compression rate of unmatched maneuvering strategies and posteriori information, the matching degree of the maneuvering strategies in the target tracking process is improved, and further, the matching degree of a motion model is further improved. Meanwhile, by combining the self-adaptive structure model and Kalman information filtering, measurement information of multiple sensors is effectively fused, and the tracking precision and the stability of target tracking are remarkably improved.

The invention discloses a social network based mobile terminal user grouping method. The method comprises: according to history of communication between terminal users, quantizing communication contact to generate a social relational graph (STG); in combination with preference attributes of the terminal users, generating an attribute relational graph (ARG) taking preference degrees between the terminal users and attributes as weights; generating a social relation-attribute graph in combination with the STG and the ARG, designing an SAPLA algorithm to predict unknown attributes of the terminal users, and adjusting preference degrees of known attributes; and proposing an SARA algorithm by utilizing a random walk model, combining transfer probabilities between the terminal users and between the terminal users and the attributes, giving out a transfer probability matrix between the terminal users, with relatively low complexity, giving out a random walk distance matrix Rl by utilizing the transfer probability matrix, setting a target function in combination with the matrix Rl, and grouping the terminal users until the target function is converged. According to the method, the complexity of operation is lowered and the accuracy of grouping is improved.

The invention discloses a load characteristics comprehensive classification method based on Markov Monte Carlo. The method includes the following steps: finding the voltage drop time point, carrying out load dynamic characteristics extraction and classification at the disturbance moment corresponding to the voltage drop time point; judging whether the change between the load classifications has a Markov property or not; dividing all data into uniform segments by time; establishing Markov chain's probability transfer matrix based on the maximum likelihood thought for each data segment; judging whether the numerical characteristics are changed or not: if no, go to step V; if yes, carrying out clustering on the load data in the time segment according to the numerical characteristics corresponding to the matrix, and obtaining the probability transfer matrix of the load data with changed numerical characteristics in each time segment ; carrying out Markov Monte Carlo simulation and describing the load change situation; processing the sequence reflecting the load classification conversion using the Hidden Markov Model (HMM). The method provided by the invention improves the Markov chain Monte Carlo simulation and effectively reduces the possibility of the matrix entering the stable state after iteration.

The invention relates to a prediction method and device for an opportunity network link, and a readable storage medium. The method comprises the steps: reconstructing a Markov transition probability matrix based on a random walk algorithm and second-order neighbor information, and obtaining an improved restart random walk similarity index according to the Markov transition probability matrix; segmenting the opportunity network data to obtain multiple network snapshots according to a time series analysis method, and performing the randomized walk of each of the segmented network snapshots to obtain a corresponding similarity matrix according to the improved restart random walk similarity index; establishing a deep belief network prediction model, taking the similarity matrix as the model input, and obtaining a lowest-energy deep belief network prediction model to predict the opportunistic network link after the iterative training test. The method provided by the invention can achieve the precise prediction of the network link.

The invention discloses a wind power probabilistic forecasting method based on a longitudinal moment Markov chain model. Wind power historical data of corresponding moments are recorded through reasonable longitudinal moment and state partition to form a longitudinal moment Markov chain probability transfer matrix which embodies state transition probability characteristics of longitudinal moments; on the basis of wind power probabilistic forecasting results, forecast is carried out by recording the probability distribution of wind power output variation of an adjacent moment and setting expectation of a confidence interval over probabilistic forecasting, probabilities exceeding the confidence interval are amended through the expectation of variation, and precision of determinacy predicted values is improved. Through verification of actual wind field data and comparison of various error indicators, the effectiveness of the model and the forecasting method is confirmed.

The invention discloses a geographical location prediction method based on a continuous time sequence Markov model. The method comprises the following steps: step 1, filtering and clustering the original user trajectory data to generate a series of candidate locations; step 2, converting the user trajectory data into a [time T, location L] sequence according to the information of the candidate locations; step 3, implementing Gaussian mixture modelling on the sequence density of each location, and improving an original Markov model in combination with a transition probability matrix, a sequence point probability and other information to establish a Markov model based on a continuous time sequence; and step 4, predicting the geographical location of a target time point by using the Markov model based on the continuous time sequence. By adopting the technical scheme disclosed by the invention, the accuracy of prediction can be improved.

A bulk password grading priori check method includes two steps of grading based on actual password data as well as training and retrieval of a grading password model. The grading includes the steps of firstly sampling and extracting millions of Chinese user password data to obtain a sample training set, then obtaining password probability transfer matrix through a second-order Markov model according to the sample training set and calculating to obtain a strength assessed value of each password, and finally grading all password data according to the password assessed value and a password hierarchy. The training and retrieval of the grading password model includes the steps of storing grading results at a bit vector of a Bloom filter at each grade through hash functions by building a password Bloom filter of different strength for the password set of each grade so as to build a grading password model, and using the password Bloom filters of multiple grades for quickly and efficiently determining the strength of passwords to be checked during retrieval.

An embodiment of the invention provides a method and a device for confirming influence sequencing of users. The method comprises the steps of obtaining first information and second information, wherein the first information is used for indicating the mutual focusing relation of N users in a social network, and the second information is used for indicating the message number sent by the user in N users and schemes of messages sent by the users; confirming N-dimensional weight vectors corresponding to target schemes according to the first information and the second information; obtaining third information, wherein the third information is used for indicating the number of messages forwarded by the users having message forwarding relation and the schemes of the forwarded messages; confirming transition probability matrixes corresponding to the target schemes according to the third information; utilizing a Pecs sorting algorithm to confirm the influence sequencing of N users in the social network in the target scheme field. The actual relevancy of the users in the social network is reflected, so that the accuracy of an influence sequencing result is effectively improved.

The invention discloses a block chain transaction network node importance ranking method. The method comprises the steps that A, a transaction network topological graph is generated; B, background nodes are added; C, edges from non-background nodes to the background nodes are added; D, edges from the background nodes to all the non-background nodes are added; E, a probability transfer matrix is generated, and an initial importance score is given to each node; F, power iteration is performed on the probability transfer matrix H; G, whether convergence is performed is judged, if yes, the step His executed, and if not, the step F is returned to and executed; and H, the background nodes are deleted, and the importance scores of the remaining nodes are obtained. By the adoption of the method,the importance ranking result of the block chain transaction network nodes can better conform to the nature of a block chain transaction network, and the fairness and credibility of importance rankings of the block chain transaction network nodes can be improved.

The invention discloses an optimal coding-based reversible steganography method, which comprises the following steps: generating an original carrier sequence, processing the original carrier sequenceinto a composite carrier sequence in a predetermined manner, and constructing a distortion matrix of the composite carrier sequence; calculating an embedding rate by using a message length to be embedded and the length of the original carrier sequence; calculating the optimal transition probability matrix according to the embedding rate, the distribution probability of the composite carrier sequence, and the distortion matrix; and recursively encoding the composite carrier sequence by using the optimal transition probability matrix so as to reversibly embed a secret message into the compositecarrier sequence with minimal distortion to obtain a secret carrying sequence. The method draws on a distortion model in the steganography field, defines accurate modifying distortion for the reversible hidden carrier, and then uses the optimal coding technique in the reversible hidden field to complete the information embedding with minimal distortion so as to achieve higher security.

The invention discloses a transfer traffic predicting method based on a Markov predicting method. The transfer traffic predicting method comprises the following steps: firstly, calculating a state transfer probability matrix; secondly, predicting a traffic transfer rate through an OD (origin destination) subregion; thirdly, determining a relevant road traffic transfer coefficient through determining relevant road traffic transfer coefficients; finally, by multiplying relevant road traffic prediction with corresponding transfer coefficients, predicting each traffic transfer. Compared with the prior art, the transfer traffic predicting method is improved on the basis of the Markov predicting method, thus quickly and effectively calculating all kinds of predicted transfer traffic, simplifying the conventional predicting calculating steps, improving the work efficiency, and making prediction more smooth and effective; therefore, the transfer traffic predicting method has popularization and application values.

The invention relates to an information classification method and device. The method comprises the steps that intention classification log records of text data information corresponding to historical voice data information input by a user are obtained; text data information corresponding to a plurality of similar inquiry requests is obtained from the intention classification log records; according to the text data information corresponding to the multiple similar inquiry requests, a preset convolution nerve network model and a preset transition probability matrix, a user intention classification model and a target transition probability matrix are determined; the user intention classification model and the target transition probability matrix are used for determining the target intention category to which the current text data information belongs is determined, wherein the current text data information corresponds to the received current voice data information; a database corresponding to the target intention category is searched for response information corresponding to the current voice data information. According to the technical scheme, more accurate response information can be provided for a user, the searching time can be shortened, the search efficiency can be improved, and the user experience can be improved.