68 results about "Relevance prediction" patented technology

The embodiment of the invention discloses a method and a device for establishing a picture search correlation prediction model, and a picture search method and device. The method for establishing the picture search correlation prediction model comprises the following steps: using a training sample to train a pre-constructed original deep neural network, wherein the training sample comprises a query and picture data, and the original deep neural network comprises a representation vector generation network and a relevant computational network; and taking the original deep neural network which finishes training as the picture search correlation prediction model. The technical scheme of the invention optimizes the traditional picture search technology, and is better than the traditional technology and various fusion and variation capabilities on multiple aspects including the semantic matching of the query and a picture text, the semantic matching of the query and picture contents, click generalization and the like, and relevancy between a picture search result and the query input by the user can be greatly improved.

Enhanced meteorological measure-correlate-predict systems and methods. The systems and methods preferably consider publicly available, long-term data sets at each of a plurality of locations nearby a potential wind farm location. A test tower is preferably located at the potential location to collect a shorter-term data set, which, in combination with the long-term data set, is used to correlate and train embodiments of the systems and methods of the present invention using computational learning systems. Longer-term data can then be predicted for the potential wind farm location based on the correlation.

Enhanced meteorological measure-correlate-predict systems and methods. The systems and methods preferably consider publicly available, long-term data sets at each of a plurality of locations nearby a potential wind farm location. A test tower is preferably located at the potential location to collect a shorter-term data set, which, in combination with the long-term data set, is used to correlate and train embodiments of the systems and methods of the present invention using computational learning systems. Longer-term data can then be predicted for the potential wind farm location based on the correlation.

The invention discloses an improved miRNA-disease relevance prediction method based on collaborative filtering. A miRNA-disease prediction problem can be regarded as a recommendation repair problem. On the basis of a known miRNA-disease-related bipartite network, miRNAs are recommended to use according to known preferences of the miRNAs to related diseases and vice versa. Firstly, an importance matrix SIGd of one disease to another is defined, calculated and measured. When a disease d (i) is perceived to be more important than a disease d (j), the score of SIGd (d(i), d(j)) is higher. Similarly, SIGr is defined and calculated in order to measure the importance of two miRNAs. Secondly, a significant matrix and a similarity matrix are utilized as weight for calculating scores. The similarity matrix is defined to represent similarity between miRNAs or between diseases. The final score of miRNA-disease relevance is the sum of the scores of a miRNA and a disease and score of the miRNA scored by the disease. With the method, higher prediction accuracy is realized.

A correlation between develop inspect (DI) and final inspect (FI) profile parameters are established empirically with test wafers. During production, a wafer is measured at DI phase to obtain DI profile parameters and FI phase profile parameters are predicted according to the DI profile parameters and the established correlation. Each wafer is subsequently measured at FI phase to obtain actual FI profile parameters and the correlation is updated with actual DI and FI profile parameters.

An encoder (1600) and decoder (1700) for improving bit rate efficiency in a wavelet based codec includes an analysis filter bank (1601) for decorrelating the input data signal. A set of decimators (1701) are used to down sample the filtered input data signal and a predictor (1705) is used to extract cross subband dependence. The predictors (804, 904, 1104, 1204, 1304) are used in order to reduce the number of bytes of an encoded input data signal X(Z). The predictors exploit existing correlation amongst the subbands resulting from a multi-level analysis wavelet transformation or filter bank processing. Decimation required by the analysis filter bank is placed around the predictor on the basis of spatial location variance minimization to further facilitate subband prediction, and on computational complexity of the overall system.

Systems, apparatuses, and methods for optimizing a load-store dependency predictor (LSDP). When a younger load instruction is issued before an older store instruction and the younger load is dependent on the older store, the LSDP is trained on this ordering violation. A replay / flush indicator is stored in a corresponding entry in the LSDP to indicate whether the ordering violation resulted in a flush or replay. On subsequent executions, a dependency may be enforced for the load-store pair if a confidence counter is above a threshold, with the threshold varying based on the status of the replay / flush indicator. If a given load matches on multiple entries in the LSDP, and if at least one of the entries has a flush indicator, then the given load may be marked as a multimatch case and forced to wait to issue until all older stores have issued.

In a method and apparatus for predicting process results of objects being processed in a processing chamber of a processing apparatus, operation data and process result data obtained at the time of processing each of the objects are collected and a multivariate analysis is performed on the basis of the collected operation data and process result data to obtain a first correlation therebetween. Process results are predicted on the basis of the first correlation. Weighting coefficients for the respective operation data are set on the basis of the predicted process results. A second correlation between the weighted operation data and the process result data is obtained by a multivariate analysis. Process results are predicted by using operation data, obtained when objects other than the objects used to obtain the second correlation are processed, on the basis of the second correlation.

The invention belongs to the technical field of computer network data prediction, and discloses an academic field data correlation prediction method based on deep learning, and a computer. The methodcomprises steps of collecting public general data, and paper and patent data in the academic field; training an academic field word vector on the academic corpus by utilizing a word vector technologyof deep learning; for a given field, predicting other fields related to semantics according to the word vectors, and achieving prediction of related academic fields. The system comprises a data collection module used for collecting public data; the word vector training module is used for training academic field word vectors on the academic corpus by utilizing a word vector technology of deep learning; and the academic domain prediction module is used for predicting other semantically related domains according to the word vectors for a given domain to realize prediction of the related academicdomains. According to the method, the academic field word vector based on deep learning is constructed, and field-related rapid and accurate prediction is realized by means of the word vector.

The invention discloses a hierarchical dance movement posture estimation method based on sequence multi-scale depth feature fusion, and the method comprises the following steps: extracting a detectionframe of a dancer human body based on a YOLOv3 detector, inputting an RGB image into a YOLOv3 model to acquire the detection frame of the human body; extracting joint point features of the detectionframe of the obtained human body to obtain features integrated with multi-resolution multi-scale information, using a softmax function of the features integrated with the multi-resolution multi-scaleinformation to obtain a heatmap of joint points, and acquiring position information of all joints through estimation of the heatmap; and carrying out joint point geometrical relationship relevance prediction on the estimated human skeleton joint points, constructing a hierarchical attitude estimation model based on the joint point geometrical relationship by analyzing the geometrical relationshipbetween the joint points, and carrying out multi-level joint point estimation. According to the invention, the accurate estimation of the dancer joint point position can be improved, and the dancing action posture estimation effect is improved.

Embodiments of the present invention provide a method, an apparatus, an electronic device, and a storage medium for generating a recommendation word. The method comprises the following steps: acquiring keywords input by a user, and generating a plurality of related words according to the related relation of the keywords by querying the related relation database as recommendation words; The recommendation words are sorted according to the correlation prediction score of the recommendation words with respect to the keywords, and the recommendation words are displayed in the sorted order. The embodiment of the invention makes the recommendation word ranking which is more relevant to the user search keywords more forward, improves the user's use experience, and remarkably improves the ratio ofthe user clicking on the web page.

The invention discloses a page-level FTL solid state disk performance optimization method based on correlation perception. In an SSD based on a flash memory, an FTL is adopted to redirect write content to an idle physical address, and a mapping table from a logic address to the physical address is managed, which seriously affects the performance of the SSD based on the flash memory. In order to improve the performance based on the flash SSD, I / O correlation in workloads is utilized, a related perception page level FTL technology is provided, and a related perception mapping table is designed and a correlation prediction table is constructed based on correlation of reading operation so as to support rapid searching of mapping entries in the related perception mapping table. Moreover, the read-write cache is split, and a skew-perceived dirty item index is constructed, so that the cache hit rate is increased, and the garbage collection overhead is reduced. According to the method, the page mapping efficiency can be remarkably improved, the read-write performance is improved, and the junk collection overhead is reduced by using the semantic link related perception page level FTL method.

The invention provides a method for measuring collapse influence factors of a punched bored concrete pile hole wall. The method for measuring collapse influence factors of the punched bored concrete pile hole wall comprises the following steps that collapse hole testing study area is confirmed; punched hole geometry factors of the collapse hole testing study area are collected and a theory volume Vi and an actual volume ( please refer to the formula )of a punched hole are calculated; qualitative factors and quantitative factors of the punched hole are confirmed and the qualitative factors and the quantitative factors are respectively defined; a collapse response matrix is confirmed; a collapse rate of the drill hole and a collapse hole reference variable are confirmed; a correlation evaluation formula of collapse hole collapse factors is established; a precision evaluation is conducted on the correlation evaluation formula of the collapse hole collapse factors; an analysis and an evaluation are conducted on an evaluation precision of a correlation prediction formula of the collapse hole collapse factors by applying a complex phase correlation figure of the correlation evaluation formula; an effect degree and size of the collapse factors of all collapse holes are analyzed and evaluated, and major collapse factors and minor collapse factors of the collapse holes are confirmed. According to the method for measuring collapse influence factors of the punched bored concrete pile hole wall, due to the fact that the qualitative variables are divided based on the values, the important practical value towards the analysis and evaluation of influence factors of a hole wall of a bored concrete pile collapse hole is achieved.

The invention discloses a parallel drug-target correlation prediction method based on sorting learning, and belongs to the field of bioinformatics. According to the method, multiple types of similarity, correlation characteristics, chemical spatial characteristics and gene spatial characteristics are extracted through multiple characteristic extraction methods; then, due to the fact that a characteristic set with high dimensionality can be obtained through multi-angle characteristic extraction and samples do not have conventional positive and negative example labels, dimensionality reduction processing is conducted through a principal component analysis method, then, the dimensionality reduced characteristic set is input into a sorting learning algorithm, and finally the drug-target correlation degree under each query can be predicted and output. By utilizing sorting learning, the drug-target correlation is no longer simply divided into correlation or uncorrelation, and sorting is performed according to the correlation degree of the drug and the target, so that research and development of a new drug are facilitated, and redirection of the drug is also facilitated.

The embodiment of the invention provides a synonym recognition model training method, a synonym determination method and synonym determination equipment, and relates to the technical field of machinelearning and computers. The method comprises the steps of obtaining a plurality of words; obtaining multi-source feature information of the words, wherein the multi-source feature information comprises semantic feature information and character feature information; determining a plurality of training samples based on the plurality of words; determining a synonym prediction result and a correlationprediction result of the training sample based on the multi-source feature information of the two words in the training sample through a synonym recognition model, the correlation prediction result being a prediction result of correlation between the two words in the training sample; calculating a loss function value of the synonym recognition model based on the synonym prediction result and thecorrelation prediction result of the training sample; and training the synonym recognition model according to the loss function value. According to the technical scheme provided by the embodiment of the invention, the synonym identification accuracy can be improved.

The invention discloses a 3D target tracking method and system based on point cloud sequence data, and belongs to the field of digital image recognition. The 3D target tracking method comprises the following steps: respectively extracting and standardizing a search point cloud and a template point cloud containing a target frame from a current frame and a previous frame, and predicting the position and posture of the target frame in the current frame by utilizing a 3D target tracking model so as to determine the position of a 3D target in the current frame, wherein in the 3D target tracking model, the feature extraction network is used for extracting template point cloud features and searching point cloud features, and the correlation prediction network is used for predicting a target score of each feature point in the search point cloud, and the integrated regression network is used for carrying out point-by-point regression after the two features are fused, and the position prediction network is used for carrying out weighted multiplication on the distance and posture between each feature point in the fused features and the center of a target box according to the target score ofeach feature point of the search point cloud. According to the 3D target tracking method, the three-dimensional attribute of the object can be fully utilized, and the calculation efficiency and stability of 3D target tracking are improved.

A unit dosage form, such as a capsule or the like, for delivering drugs into the body in a circadian release fashion comprising one or more populations of drug-containing particles (beads, pellets, granules, etc.) is disclosed. Each bead population exhibits a pre-designed rapid or sustained release profile with or without a predetermined lag time of 3 to 5 hours. Such a circadian rhythm release drug delivery system is designed to provide a plasma concentration-time profile, which varies according to physiological need at different times during the dosing period, i.e., mimicking the circadian rhythm and severity / manifestation of gastric acid secretion (and / or midnight gerd), predicted based on pharmaco-kinetic and pharmaco-dynamic considerations and in vitro / in vivo correlations.

A method for predicting the joint probability density of output power of multi-wind farms includes such steps as establishing a prediction model of a sparse Bayesian learn machine, predicting the probability density of output power of wind farms in multiple independent time periods in the future, predicting the probability density of output power of multi-wind farm, predicting the probability density of output power of sparse Bayesian learning machine, predicting the probability density of output power of wind farm in multiple independent time periods in the future, predicting the probabilitydensity of output power of sparse Bayesian learning machine, predicting the probability density of output power of multi-wind farm. The sparse Bayesian learning machine is used to get the prediction error samples, and then the correlation coefficient matrix between prediction errors is obtained according to the prediction error samples. The sparse Bayesian learning machine is used to forecast themean and variance of wind farm output power, and the covariance matrix is obtained by combining the mean and variance predicted with correlation coefficient matrix, and the joint probability density prediction is completed. The method improves the accuracy and effectiveness of wind farm output power prediction by forecasting the output power of each period of wind farm and the correlation betweenthe output power of each period of wind farm, makes the prediction more close to the actual situation of the real wind farm, and provides more abundant and accurate information for the dispatching decision-making of the power system with wind farms.

The invention discloses a convolutional neural network-based bug positioning method combining source code semantics and grammatical features. The method is characterized by providing the method for positioning a source code file for generating a bug according to a bug report submitted by a user; according to the method, a CNN is used for carrying out feature extraction on a bug report, source codesemantics and source code syntax, then the features are fused, unified features are extracted, finally, the CNN is used for carrying out correlation prediction on the bug report and source codes, andTopK source code files related to the bug report are obtained. Therefore, when a user submits a new bug report, a maintainer can perform positioning of the related source code file in time and notifya developer of repairing, and the bug repairing efficiency and the project maintenance efficiency are improved. The whole process of the method is shown in the attached drawing of the abstract.

The invention provides a semantic map construction method based on a convolutional neural network and a computer storage medium, and the method comprises the following steps: S1, receiving a 2D image,transmitting the 2D image to a convolutional neural network model, and outputting neurons of dense pixel-level semantic probability map points; S2, tracking classification probability distribution ofeach curved surface by adopting a Bayesian updating model; S3, providing data by adopting an ElasticFusion method to carry out relevance prediction, and updating probability distribution; and S4, improving semantic detection through the scale information of the map by utilizing a conditional random field regularization model. According to the semantic map construction method based on the convolutional neural network, semantic segmentation can be carried out based on the convolutional neural network, the semantic map is generated, and the robustness of the semantic map under few weak texturesis enhanced.

The invention discloses a network model capable of jointly realizing semantic segmentation and depth-of-field estimation, and the network model comprises a feature sharing module and a multi-task sub-network; the multi-task sub-network comprises a plurality of task sub-networks with the same structure for processing different task targets, and comprises a feature screening module, an attention concentration module and a prediction module; the feature screening module screens out features related to the task from the shared features; the attention concentration module is used for improving thecorrelation between the screening features and the task target; and the prediction module is configured to output a processing result of each task target after convolution of the concentrated attention features. The invention further discloses a training method of the model. Back propagation iterative training is carried out on semantic segmentation and depth-of-field estimation. The model provided by the invention is high in accuracy, strong in robustness and light in weight.

A method is provided for decision support in regulating an adhesive coating applied to Yankee dryers. Online sensors are configured to continuously measure stock characteristics, and additional sensors provide actual stock flow rate and machine speed. A controller predicts potential natural coating application from a fibrous sheet generated from the stock to the Yankee dryer surface, substantially in real time, based on the measured characteristics and sensed actual machine values. An output signal may be provided to a display unit, wherein an optimal adhesive coating feed rate may be determined and displayed for operator decision support. The controller may in an automatic mode be configured to regulate the adhesive coating feed rate based on a comparison of one or more determined optimal values associated with respective actual values. The method may include identifying fiber source changes in real time, and predicting a natural coating potential based partly on predetermined correlations.

The invention provides a method for increasing credibility of question prerequisite in a visual question answering scene. The method comprises following steps: prerequisite information extraction, question correlation prediction database, question correlation detection, visual question answering data expansion. First, the prerequisite information in the problem is extracted; the problem correlation prediction and explanation database are constructed; and binary classification of problem image pairs (Ii, Qi) is performed; whether the image Ii has the prerequisite information in question Qi is identified; then on the basis of one-hot coding, the image Ii and the problem Qi are encoded using the VGG network and the short and long term memory network, respectively, and input the data to the multi-layer sensor for prediction. The method of the invention can handle a plurality of target objects and their relations in different scenes, provide an encoding method to calculate the image matching distance, and improve the credibility of the question prerequisite information.

The invention provides a method for sensing a frequency spectrum by using the correlation among channels, which can be used for shortening the sensing time and improving the throughput under the condition of meeting the precision requirement. The method in the embodiment of the invention comprises the following steps of: acquiring correlation information among target channels; acquiring a frequency spectrum sensing result of at least one the target channel; and determining sensing results of the other target channels according to the frequency spectrum sensing result of the at least one target channel and the correlation information among the target channels. According to the embodiment of the invention, part of channels is required to carry out frequency spectrum sensing, and thus the time of a sensing period is shortened; and sensing results of other non-sensed channels are predicted and acquired by using the correlation among the channels, so that the expenditure of frequency spectrum sensing is reduced and the throughput of the system is improved.

The embodiment of the invention discloses a public opinion analysis method and device, electronic equipment and a readable storage medium, and relates to the technical field of big data. The specificimplementation scheme is as follows: receiving a public opinion analysis request from terminal equipment, wherein the public opinion analysis request comprises an analysis keyword set by a user; and obtaining at least one document related to the analysis keyword according to the correlation result of the analysis keyword and each document in a target document set, the correlation result being obtained by a correlation prediction model through prediction by using correlation features, and the correlation features being obtained in advance based on the analysis keyword and each document; performing public opinion analysis on the at least one document to obtain a public opinion analysis result for the analysis keyword; and sending the public opinion analysis result for the analysis keyword tothe terminal equipment. According to the method, the accuracy and efficiency of correlation calculation can be ensured, and the actual public opinion analysis requirement is met.

The invention discloses an RFID spatio-temporal data traffic flow characteristic parameter prediction method which comprises the following steps: S1, acquiring traffic data of an RFID acquisition target road section, and performing spatio-temporal correlation analysis on the traffic data; S2, obtaining correlation between traffic flow characteristic parameters influencing the traffic state of thetarget road section and traffic flow characteristic parameters capable of reflecting the traffic state of the target road section; S3, predicting traffic flow characteristic parameters of the target road section in a traffic flow stable state and a traffic flow unstable state; and S4, carrying out weighted combination on the traffic flow characteristic parameters in the two states. The method solves the problems of large calculation amount, poor real-time performance and anti-interference capability, low prediction precision, low prediction efficiency and the like of an existing prediction method, and has the advantages that accurate, comprehensive and reliable traffic flow characteristic parameter prediction can be realized, and a new thought is provided for improving the traffic jam problem.