32 results about "Probability representation" patented technology

A system and method for decentralized cooperative control of a team of agents for geographic and other search tasks. The approach is behavior-based and uses probability particle approach to the search problem. Agents are attracted to probability distributions in the form of virtual particles of probability that represent hypotheses about the existence of objects of interest in a geographic area or a data-space. Reliance on dependable, high-bandwidth communication is reduced by modeling the movements of other team members and the objects of interest between periodic update messages.

In one illustrative embodiment, a computer-implemented method for identifying program components, generates, by a processor unit, associations between the program components stored in a computer usable storage medium to create a first set of relationships, and assigns, by the processor unit, a probability to each relationship in the first set of relationships to create a second set of relationships, wherein the probability indicates a strength of the associations between the program components.

A probabilistic clustering method of trans-type data based on keyword entries belongs to the database field and comprises the following steps: (1) defining the type of the keyword entry; and dividing the trans-type data into a keyword correlation entry, a keyword half-correlation entry and a keyword non-correlation entry; (2) allocating probability for each entry; (3) expressing data keywords by the probability; (4) constructing a data keyword entry probabilistic similarity matrix M; for any two data of the trans-type data dx and dy in the step (3), computing similarity of any two descriptive forms of the dx and the dy, summing the probabilities of the similarity which is greater than a certain threshold, and storing the direct correlation probabilities of the any two data in the matrix M; (5) constructing a clustering model M<c> based on the matrix M; and (6) obtaining the clustering method based on the clustering model M<c>. The method clusters the trans-type data by utilizing the similarity of the entry related to the keywords, which improves the data clustering precision and reduces the clustering time.

The inventor provides a vehicle motion planning method and an unmanned vehicle, and relates to the field of unmanned vehicles, and especially relates to an unmanned vehicle motion planning. The method comprises: according to n travelling paths, updating probability distribution of an experienced region, in a vehicle motion process, according to the updated probability distribution of the experienced region, combined with probability of vehicle state conversion, calculating probability to transfer to an adjacent grid; according to the probability to transfer to the adjacent grid, expanding a random tree, and doing collision detection, if the vehicle passes the collision detection, using the corresponding grid to expand the random tree, and updating a current state; until all grids have been tried and the vehicle cannot pass the collision detection, or the vehicle has reached an objective grid. In the method, n actual paths are converted to probability to represent, and are applied in the path planning method, and according to the state information of the vehicle, the probability to transfer to the corresponding grid is calculated, so a motion planning module makes full use of the experiential knowledge to improve search efficiency.

The invention provides an abnormal data detection method, an abnormal data detection device, a storage medium and a program product. The method comprises the steps of conducting at least two structural treatments on to-be-detected target data so as to obtain at least two structured data; carrying out feature extraction on each structured data, and acquiring the feature data of each structured data; performing feature fusion on each feature data so as to obtain target feature data; carrying out machine learning on the target feature data, and obtaining the recognition probability of the targetdata, wherein the recognition probability represents the probability that the target data is recognized as normal data. By means of the method, rich feature data can be extracted when no or few abnormal data are available. The accuracy of data anomaly detection is improved. The dependence on data distribution, variable independence or data volume balance is relieved. The technical problems that inthe prior art, abnormal detection is difficult to achieve and the accuracy is low are solved.

A method for generating mathematical analysis of a communication protocol in a vehicular communications network. The method defines features of a vehicular network, which may include a graph of a street map within a geographic area. A random geometric graph with a plurality of parameters is generated. A plurality of communications protocols on the vehicular network are defined. A communication protocol over the random geometric graph is redefined. A communication protocol's basic properties and associated features on the random geometric graph are analyzed. Results of the analysis are generated. The results of the analysis based on the random geometric graph's parameters are translated into results based on the vehicular network features. The random geometric graph with the parameters are displayed. The parameters may include: a number of graph nodes; and a probability that any two nodes are communicably connected being expressed as a function of the vehicular network features.

An automatic speech recognition (ASR) apparatus for an embedded device application is described. A speech decoder receives an input sequence of speech feature vectors in a first language and outputs an acoustic segment lattice representing a probabilistic combination of basic linguistic units in a second language. A vocabulary matching module compares the acoustic segment lattice to vocabulary models in the first language to determine an output set of probability-ranked recognition hypotheses. A detailed matching module compares the set of probability-ranked recognition hypotheses to detailed match models in the first language to determine a recognition output representing a vocabulary word most likely to correspond to the input sequence of speech feature vectors.

The invention discloses a recommendation information determination method and a server. By combining the attribute information of the user and the interaction information between the users, the walking probability of random walking on the social network is set, the influence of the attribute information and the interaction information on the walking probability is fully considered, the recommendation node is constructed based on more comprehensive information, and a more accurate recommendation result can be obtained. The method comprises the steps of obtaining a relation structure of a socialnetwork; determining node characteristic values and edge characteristic values according to the relation structure of the social network; determining a walking probability according to the node characteristic values and the edge characteristic values, the walking probability representing the walking probability of each node in the relation structure of the social network; acquiring a walking pathin a relation structure of the social network according to the walking probability; and determining a first recommendation node set according to the walking path, wherein the recommendation node setcomprises at least one node corresponding to the walking path.

Methods and systems are provided for condition-based maintenance of a structural component exhibiting a physical defect. One exemplary system comprises a measurement system to obtain a current measurement of a defect in a structure, a monitoring system, and a maintenance system. The monitoring system determines a probabilistic representation of degradation development data for the defect using the current measurement, determines a reference remaining usage life metric for the structure based on the probabilistic representation and a target probability value, and determines one or more remaining usage life bounds based at least in part on the reference remaining usage life metric. The maintenance system determines maintenance schedule or other remedial action(s) for the structure in a manner that is influenced by the reference remaining usage life metric and the one or more remaining usage life bounds.

The invention discloses a text question and answer matching method inspired by quantum interference; the method comprises the steps: extracting global and local question and answer features respectively to represent and optimize the matching accuracy of a question and answer system, the system comprising a question-answer composite system module, a word weight calculation module, an answer probability distribution module, a joint probability representation module and a dual feature fusion and score calculation module; respectively extracting local features of a question density matrix and a reduced density matrix of an answer sentence by constructing the question density matrix and the reduced density matrix of the answer sentence, wherein the local features of interference terms among words contained in the answer sentence are captured based on the reduced density matrix of the composite system under the enlightenment of quantum interference; fusing the feature information of the question sentence and the feature information of the answer sentence based on a joint probability representation module, and then extracting effective feature information through convolution; and finally, splicing global and local features and obtaining a final prediction matching result through a full connection layer. Based on the end-to-end quantum-like language model, the question and answer matching method for capturing global-local feature representation is provided, the problem that local interaction information is lost in the quantum language model is solved, and the question and answer matching effect is improved.

The invention provides an image classification method based on a self-attention mechanism, and the method comprises the steps of constructing a Transform model containing the self-attention mechanism, and adding a classifier unit for an image classification task; processing the public data set ImageNet, and adjusting the original picture to a proper size; dividing the adjusted picture into sub-pictures with fixed sizes, connecting the sub-pictures, and performing dimension adjustment to obtain a picture embedding vector; performing two-dimensional position coding to obtain a two-dimensional position coding vector, and connecting the two-dimensional position coding vector with the picture embedding vector to serve as model input; and sending the connected vectors to a Transform model, extracting picture features, converting the vectors output by the model into probability representation through a classifier unit during final decoding, and completing image classification. By using the self-attention mechanism, the global information, namely the picture features extracted by the traditional convolutional neural network, can be effectively extracted from the picture, and the picture classification can be effectively completed based on the extracted features.

The invention discloses a sparse nonparametric body area channel probability representation method. The method comprises the following steps: S1, adopting an electromagnetic wave transceiver in a body area network to collect data, S2, establishing an empirical distribution function and solving the empirical distribution function through a step function, S3, adopting an empirical distribution function approximate density function and establishing a regression function, S4, solving the regression function through linear programming, S5, obtaining a corresponding probability function through derivation of the regression function, and S6, evaluating and analyzing an obtained sparse nonparametric probability model. A representation model proposed in the method is not restricted by specific propagation circumstances, and is more applicable to wireless communication in a body area network because of nonparametric property. The method overcomes the problem that the requirements on simple capacity of traditional models are rigorous, ensures that coefficients of a large amount of linear combinations are zero by controlling the quantity of support vectors during the process of regression, and realizes the sparsification of regression coefficients.

The invention discloses a sleep quality evaluation method and device for portable intelligent wearable equipment. The method comprises the following steps: an acquisition step: acquiring individual attribute information data, sleep information data and sleep quality feedback results of a user; an interactive calculation step: interacting the acquired individual attribute information data and sleepinformation data of the user with one another to generate interactive sleep characteristic data; a construction step: performing two-time nonlinear transformation on the sleep characteristic data, and constructing a sleep quality probability model to obtain probability representation data of sleep quality; and an evaluation step: calculating a sleep quality score by using the probability representation data of the sleep quality. According to the sleep quality evaluation method and device disclosed by the embodiment of the invention, the intelligent degree of the portable wearable device for sleep quality evaluation is effectively improved in combination with the accuracy of the wearable device for monitoring the sleep information from the viewpoint of individual attributes of the user.

The embodiment of the invention provides a human factor reliability assessment method and device and information processing equipment, and the method comprises the steps: generating a topological structure based on a Bayesian network according to a plurality of factors which can cause human errors and the causal relationship among the plurality of factors, wherein each factor in the plurality of factors is expressed as a node in the topological structure, the plurality of factors are obtained by analyzing multi-source data, and the multi-source data at least comprises historical data; adding probability data corresponding to the node according to the type of the multi-source data, the probability data corresponding to the node being expressed by a deterministic probability and/or an interval probability; and according to the topological structure and the probability data, obtaining at least the probability of human error occurrence. Accurate evaluation of human factor reliability is feasible by using limited statistical data.

The invention relates to the technical field of text abstracts, in particular to a Gibbs limited text abstract generation method utilizing a pre-training model. According to the method, a model is used for training and generating a text abstract, and the training comprises the following steps: 1) performing word vectorization on a text source sequence, and adding a relative position code to obtain Word Embedding; 2) extracting features by using an attention mechanism and Bi-LSTM, training a model, and finely tuning the model to obtain the output of an encoder; (3) adding a relative position code to obtain a target sequence Word Embedding; (4) enabling the parameters of the decoder end to be consistent with those of the Transformer; 5) inputting the Attention matrix into a full connection layer, and then calculating to obtain probability representation of the vocabulary; and 6) fusing an LDA model into a decoder end to carry out keyword extraction, and extracting and generating an abstract in combination with a Gibbs sampling algorithm. According to the method, the text abstract can be better generated.

A parameter estimation unit (16) estimates a set of parameters so as to optimize a likelihood function of a strength function expressing the event occurrence probability of a type m space-time event at a time t and a geospatial location s when the strength function is modelled with use of the occurrence probability of the type m space-time event at the time t and the geospatial location s, the function expressing the degree of influence of the event occurrence history, the value of the strength function representing the event occurrence probability in an observation section that includes the time t and the geospatial location s, and the relationship between the type m and the type of the event occurrence history included in the observation section, and here, the estimated parameters include the value of the strength function expressing the event occurrence probability in the observation sections, the relationship between types, and the function expressing the degree of influence of the event occurrence history.

An activity model is generated at a computer. The activity model may be generated by monitoring incoming and outgoing data in the computer. The collected data is analyzed to form a graph that describes and predicts what output is generated in response to received input. Later, a window of input and output data is collected from the computer. This collected window of data is used to query the activity model. The graph in the activity model is then used to give the probability that the collected window of data was collected from the computer used to generate the activity model. A high probability indicates that the computer is performing normally, while a low probability indicates that the computer may behaving erratically and there may be a problem with the computer.

The invention discloses an image generation method based on an improved graph convolutional network. The method comprises the following steps: 1, building an input layer, and enabling words in a sentence to be mapped into a low-dimensional and continuous word vector through a pre-trained word vector; 2, establishing a Bi-LSTM layer, and mixing semantic information through the Bi-LSTM layer; 3, constructing implicit layer representation of a target vector: firstly, syntactically mixing information of a target and other words in a sentence through a GCN layer, and then calculating context representation related to the target by using an attention mechanism; 4, constructing an MDGCN layer: constructing a multi-target dependency graph of the sentence according to the dependency syntax tree, and modeling a plurality of targets of the same sentence by using a graph convolution network according to the multi-target dependency graph; 5, establishing an output layer, converting a dimension by using a full connection layer, and converting the dimension into probability representation through a softmax function; and 6, training the model, and taking a cross entropy error function and L2 weight recession as a loss function together.

The invention provides a method for detecting chemical bonding of a connection paint/anti-fouling paint organic coating interface, and relates to a detection method, wherein the method comprises: preparing a modified isocyanate polymer solution, brushing the modified isocyanate polymer solution, ultrasonically cleaning, analyzing chemical products in the connection paint/anti-fouling paint organic coating interface, analyzing characteristic element representation of the chemical reaction products in the organic coating interface, and analyzing probability representation of the chemical bonding of the organic coating interface, to implement detection of the chemical bonding of the connection paint/anti-fouling paint organic coating interface. The method not only can detect probability of chemical bonding of a liquid/solid interface, but also prolongs service life of an antifouling coating system, has high economic benefit. The method provided by the invention has advantages of simple manner, rapidness, high efficiency and accurate evaluation, and has guidance meaning on theory research.

The invention relates to the technical field of computer vision, in particular to an image classification method based on a linear self-attention Transform. The method comprises the following steps: S1, sending a picture to an overlapping convolution coding module of a first stage, and coding the picture into a picture token by using roll operation; s2, sending the picture token into a Transform module in the stage, and extracting a picture feature vector; s3, the extracted picture feature vectors are sent to an overlapping convolution coding module of the next stage, and the feature vector dimensionality is increased while the number of the feature vectors is reduced; s4, repeating the step S2 and the step S3, and obtaining a final output vector from the Transform module of the last stage; and S5, converting the final output vector into probability representation through a classifier unit, and completing image classification. According to the method, the image features can be effectively extracted from the image, the calculation complexity of the Transform module is remarkably reduced, and the ability of the model to extract the image features is improved through the overlapping convolutional coding module and the convolutional feedforward neural network module.

The invention provides a power battery discharge interval judging method, which belongs to the field of power battery discharge interval judgment. A power battery discharge starting point and ending point probability model is firstly built; the discharge starting point and ending point probability of each selected sampling point is calculated one by one; and based on the calculated probability, the power battery discharge interval is obtained. Through building the probability model to judge the power battery discharge interval, the characteristics of a discharge starting point and an ending point are expressed by probabilities, selected sampling points are analyzed, the discharge interval can be accurately obtained, and the judgment accuracy is improved.