33 results about "Negatively associated" patented technology

A data analysis computer system is provided that receives a timeseries dataset and generates implied data from the dataset. The dataset is further vectorized to reduce the dimensionality of the data. Users provide input to identify windows of data that either positively or negatively correlate to instances of a given type of occurrence within the data. The user defined windows are converted to fixed sized windows and a machine learning algorithm constructs a model from the data. The model is used to predict instances of the given type of occurrence in newly received data. Validation of the predications may be performed.

The present invention relates to the field of machine learning, in particular to a multi-label classification method based on a gravitational model, comprising: obtaining a labeled sample set as a training sample set; calculating and sorting the distance between the training sample and other training samples to obtain the training sample Neighbor set; in the neighbor set, the positive correlation matrix is ​​constructed with the positive correlation between the labels, and the negative correlation matrix is ​​constructed with the negative correlation between the labels; the neighbor set of the sample to be tested is calculated, and the positive correlation matrix to be tested is constructed according to the neighbor set. The correlation matrix and the negative correlation matrix to be tested; the positive correlation matrix and the negative correlation matrix to be tested are used to obtain positively correlated data particles and negatively correlated data particles; a gravity model is constructed, and the positively correlated data particles and negatively correlated data are obtained through the sample to be tested The gravitational relationship between particles is classified; the invention introduces the consideration of the negative correlation between labels, fully utilizes the correlation between labels, and discovers the correlation in the neighbor set, avoiding global calculation and reducing the complexity.

The invention discloses a translation-based multimodal modeling method and its application in commodity retrieval, including: constructing positive correlation data sets and negative correlation data sets for all commodities; Space obtains the implicit representation of the final visual features of each commodity and the implicit representation of textual features; fuses the implicit representation of the final visual features of each commodity with the implicit representation of textual features to obtain the fused The implicit representation of ; based on the transformation matrix of the translation model, the implicit representation is mapped to the latent space, and the hidden representation vector of each commodity after mapping is obtained; the new query text is mapped to the latent space using the transformation matrix based on the translation model, Calculate the distance between the hidden representation vector of the product corresponding to the new query text and the hidden representation vector of each product; sort the distances from small to large, and output the top-ranked set products as product retrieval results.

The invention discloses a data table association method and device, a storage medium and electronic equipment, according to a key value of each to-be-associated data table, an association rate of each data table corresponding to each other data table is determined, and according to attribute information of each data table and the corresponding association rate between every two data tables, the association rate of each data table corresponding to each other data table is determined. And determining an association weight between every two data tables, determining an association sequence of the to-be-associated data tables based on the association weights between the data tables, and performing association according to the association sequence. Wherein the association weight represents the data size of the virtual intermediate table after the two data tables are associated, and the association weight between the two data tables is negatively correlated with the association sequence. According to the method, the association sequence of the data tables is determined on the basis of the data volume of the virtual intermediate table after every two data tables are associated in the to-be-associated data tables, so that the frequency of loading the frequency of the data tables with large data volume in the association process is reduced, a large amount of computing resources are saved, and the multi-table association duration is shortened.

The present invention provides a weakly supervised fine-grained image classification method based on hierarchical feature transformation. By analyzing a data set, only a pre-trained convolutional neural network model is used to co-locate a target in an image. At the same time, the most discriminative components are obtained through negative correlation information analysis. Use images, images and objects, images and parts to train corresponding three-level convolutional neural network classifiers: image-level classifiers, target object-level classifiers, and target component-level classifiers. Granular images for classification. The invention performs feature transformation on the convolution feature, accurately locates the target and the most discriminative component in the fine-grained image, and solves the weakly supervised fine-grained image classification problem with a new idea.

The present application discloses a method, kit and application for screening target regions detected by methylation PCR. The method of the present application includes (1) obtaining the methylation chip of the tumor to be analyzed and the corresponding transcriptome sequencing data from the database; (2) counting the methylation degree values ​​of the normal group and the cancer group, and screening for significant differences among different groups (3) Combined transcriptome sequencing expression profile analysis, statistical correlation coefficient, and screening of negatively correlated sites; (4) Correlation between the methylation candidate sites obtained in step (3) and the disclosed Literature association, screen to obtain sites with many literature support reports, large methylation differences between groups, and negatively correlated expression levels; use regression algorithm to obtain the best sensitivity and specificity set of sites, that is, the target area. The method of the present application comprehensively analyzes databases, transcriptome sequencing and literature, and combines multiple data filtering and regression algorithms to obtain sensitive and specific methylation PCR detection target regions.

The invention discloses a humor achievement method and system based on correlating analyses of big data and small data. The method comprises the following steps that data of a user is obtained; part of the data of the user is extracted as first small data, and another part of the data is extracted as second small data; big data is obtained; data correlating to the first small data and the second small data are obtained from the big data respectively and as a first correlated data set and a second correlated data set respectively after cleaning; positively correlated sentences excavated from the first correlated data set and the degree of positive correlation of the positively correlated sentences are logged into a positively correlated sentence set; negatively correlated sentence excavated from the second correlated data set and the degree of negative correlation of the negatively correlated sentences are logged into a negatively correlated sentence set; a humor sentence set is obtained according to the positively correlated sentence set and the negatively correlated sentence set. According to the method and system, a robot can obtain sentences which can bring out the humorous meaning from the big data, and the degree of humor and the level of humor of the robot can be enhanced.

The technical schemes of the invention include a shunt reactor current characteristic based adaptive three-phase reclosing method and system. The method and system are used for realizing the followingsteps: obtaining a current-mode component and voltage-mode component corresponding to three-phase current data according to the three-phase current data and voltage characteristic data through calculation; analyzing the consistency and correlation of the three-phase current data, the current-mode component and the voltage-mode component, and obtaining consistency and correlation determination fault types through analysis; and automatically executing corresponding three-phase reclosing commands according to the determined fault types and time domains where the faults are located, the time domains including before and after arc extinction. The beneficial effects of the method and system are that negative correlation characteristics under permanent failures can be detected by utilizing correlation coefficient values so that fault properties can be identified without using spectrum analysis and frequency measurement algorithms to calculate the correlation parameters of related signals andbeing affected by fault location and transition resistance; and the method is small in calculated amount and does not need to modify criterions according to line parameters, so that the method can have adaptivity.

An embodiment of the present invention provides a method for determining image similarity, a video scene switching recognition method and device, wherein the method for determining image similarity includes: the first feature point of the current frame image in the acquired video data Set, and the feature points that match the second feature point set of the comparison frame image, as the associated feature point pair set; obtain the preset similarity of the feature points that do not match between the first feature point set and the second feature point set degree; based on the total number of associated feature point pairs and the Euclidean distance of associated feature point pairs, calculate the similarity of associated feature point pairs; the similarity of associated feature point pairs is positively correlated with the number of associated feature point pairs, It is negatively correlated with the Euclidean distance of the associated feature point pair set; the weighted average of the preset similarity and the similarity of the associated feature point pair set is determined as the global feature similarity between the current frame image and the comparison frame image.

The invention provides a method, a device and a system for map quality inspection, and a storage medium. The method comprises: obtaining thermodynamic values of all regions on the map; wherein the thermodynamic value of each area is used for representing the reliability of the corresponding area; wherein the thermodynamic value and a reliable value of an area corresponding to the thermodynamic value are in a negative correlation mapping relation; determining an area of which the thermal value is greater than a preset threshold as a target area; and displaying the target area on the map. Therefore, map quality inspection personnel can quickly determine a problematic area, the area is corrected, the manual detection time is shortened, and the labor cost is reduced. The map production time isshortened, and the map production efficiency is improved.

The present disclosure provides systems and methods for enhancing audience measurement data. Offline and online audience measurement data can be compared and correlated to improve the quality of each data and source collection. A positive correlation between offline and online datasets may indicate that demographic characteristics are likely to be true, such that outliers are removed from the set or considered with reduced weight. A negative correlation may indicate that demographic information, such as online measurement data, within a set of sources may be false or untrustworthy.

The present application provides a target detection method and system, which relate to the technical field of automatic driving. Among them, the method of target detection includes: obtaining the image data and radar data of the target in front of the vehicle; obtaining the environmental information of the environment in which the vehicle is located; detecting the target according to the confidence coefficient of the image data, radar data and image data, the confidence coefficient and the environmental information The indicated harshness of the environment is negatively correlated, and the product of the confidence coefficient and the confidence of the target detected based on the image data is the final confidence of the target in the image data. Determine the confidence coefficient of the image data through the environmental information of the vehicle, and then determine whether to use image data to detect targets based on the confidence coefficient of the image data, which can avoid determining whether to use image data for detection based on thresholds such as image brightness, color shift, or clarity. target, thereby effectively reducing the false detection rate of the target and improving the accuracy of target detection.

The invention puts forward a linear transformer table family multistage line loss analysis method and system based on same-period data. The method comprises the following steps that: according to obtained linear transformer table family basic equipment information and a power distribution network, collecting information data, setting a cut-through analysis rule for relevant data to which the same-period data relates to realize same-period cut-through association analysis, automatically carrying out association analysis in fixed time, outputting a linear transformer table cut-through abnormal data inventory, and forming an equipment set and a cut-through set; through line loss day monitoring, realizing double-high-loss analysis on a distribution line and an area line loss, and forming a monitoring set and an abnormal set; utilizing a system clustering analysis method to analyze a key factor which influences the line loss to form a clustering set; utilizing a positive and negative correlation analysis method to realize the influence analysis of a user for the distribution line and an area line loss rate, and forming a relevant set; and generating a linear transformer table family multistage analysis matrix graph. By use of the method, the detail inventory of each factor set can be provided so as to be favorable for quickly carrying out auxiliary analysis of abnormal problems, andtherefore, various auxiliary reference decisions are provided for the line loss governance of business personnel in time.

The invention provides a method for predicting section water quality parameter data, comprising the following steps: acquiring first information of a first monitoring station in real time; acquiring second information of a water gathering area formed by a plurality of first monitoring stations according to the first information; acquiring third information of the section in real time; calculatinga first correlation coefficient of each water quality parameter, a second correlation coefficient of each water quality parameter and each rain-shine parameter and a third correlation coefficient of each water quality parameter and each hydrological parameter according to the second information and the third information in a preset duration; fitting the first correlation coefficient, the second correlation coefficient and the third correlation coefficient according to a preset first model, determining parameters in which a positive correlation, a negative correlation and a non-correlation exist; and if in the third information of the current section, a target parameter is missing, determining a parameter which is non-correlated with the target parameter; and according to the parameter non-correlated with the target parameter and a preset second model, determining the numerical value of the target parameter. Therefore, the robustness of the prediction result is improved.

This application provides a method for training a recommendation probability prediction model, a recommendation probability prediction method and a device, which relate to the field of artificial intelligence technology. The sample data is input into the recommendation probability prediction model to obtain the predicted recommendation results corresponding to the sample data; the corresponding basic loss value is obtained according to the predicted recommendation results corresponding to the sample data and the actual recommendation results; based on the basic loss value and the first adjustment value, the corresponding The target loss value of ; wherein, the first adjustment value is used to represent the total difference degree of the output results of each two task networks obtained based on the corresponding sample data, and the first adjustment value is negatively correlated with the target loss value; according to the target The loss value is to adjust the parameters of the recommendation probability prediction model to improve the prediction performance of the trained recommendation probability prediction model, thereby improving the prediction accuracy of the prediction information to be predicted.

The invention discloses a method for evaluating the accuracy of CT dimension measurement by using the DVC deformation algorithm, specifically: determining a standard sample for dimension measurement accuracy evaluation according to the material and structure of the sample to be measured for CT dimension measurement; Under the CT test conditions, use CT equipment to scan the standard sample twice consecutively to obtain two sets of three-dimensional CT image data under the same test conditions; perform DVC calculation on the two sets of three-dimensional CT image data obtained to obtain three-dimensional full-field deformation data , the strains in three directions in the deformation data are denoted as ex, ey, ez; statistical analysis is carried out on the obtained full-field deformation data, and the mean value and standard deviation of each strain are obtained. The accuracy of CT size measurement is negatively correlated with the standard deviation, That is, the smaller the standard deviation, the higher the measurement accuracy; the larger the standard deviation, the lower the measurement accuracy. The method of the invention can quantify the influence of various influencing factors on the CT size measurement.