44 results about "Target weight" patented technology

The invention discloses a network risk value determination method and device, equipment, a medium and a program product, and the method comprises the steps: obtaining the weight adjustment amount of a target network corresponding to each risk dimension and the influence degree of the weight under different risk dimensions on the network risk value of the target network under the condition that a weight adjustment trigger event is detected; according to each weight adjustment amount and each influence degree, an adjustment loss value is determined, and the adjustment loss value is in positive correlation with the absolute value of the weight adjustment amount and the influence degree; in the plurality of adjustment loss values, determining a weight corresponding to the minimum adjustment loss value as a target weight; and obtaining a data score value corresponding to the target network under each risk dimension, and determining a target network risk value according to the data score value and the corresponding target weight. According to the embodiment of the invention, the network risk assessment result can respond to the change of the network threat situation in real time, and the accuracy of network security risk assessment is improved.

The invention discloses a sample labeling resource allocation method and electronic equipment, and the method comprises the steps: obtaining a comprehensive value score of each sample in a to-be-labeled sample set, the comprehensive value score being a weighted sum of an uncertainty score, a diversity score, a time correlation score and a domain migration cost score; based on data distribution of the to-be-labeled sample set, adjusting weighting coefficients of the uncertainty score, the diversity score, the time correlation score and the domain migration cost score to obtain a target weight parameter group; updating the comprehensive value score of each sample according to the target weight parameter group, and determining a to-be-labeled sample set with the maximum comprehensive score sum under the labeling budget constraint as a target to-be-labeled sample set based on a Lagrange multiplier method; the sample set with the maximum comprehensive value sum is optimally selected under the constraint of the marking budget, accurate and efficient allocation of marking resources is achieved, and the selected sample can better meet the requirement of model training for high-value data.

The invention discloses a line icing thickness prediction method and device and electronic equipment. The method comprises the following steps: receiving an icing thickness prediction request; in response to the icing thickness prediction request, determining a plurality of line parameters corresponding to the target line according to the line identifier; inputting the plurality of line parameters into a first network layer of the prediction network model, obtaining corresponding output parameters according to a target weight matrix corresponding to the first network layer, and inputting the corresponding output parameters and the plurality of line parameters into a next network layer until a plurality of output parameters are obtained; according to the plurality of output parameters, determining a target icing stage and an icing change rate corresponding to the target line in the preset time; and according to the initial icing thickness, the icing change rate and the target icing stage of the target line, determining the target icing thickness of the target line at the preset time. According to the invention, the technical problems of complex icing generation and disaggregation change process and low accuracy of the predicted line icing thickness in the prior art are solved.

Embodiments of the present application provide a virtual object interaction starting method and device, electronic equipment and storage medium. The virtual object interaction starting method comprises: first, obtaining a target weight corresponding to a user initiating an interaction request, the target weight being determined based on a plurality of user information corresponding to the user; then, if the target weight belongs to a preset weight range, starting the interaction for the user. Through the above method, the target weight corresponding to the user initiating the interaction request is obtained through a plurality of user information, and then the obtained target weight is compared with the preset weight range. If the obtained target weight belongs to the preset weight range, the virtual object starts the interaction, so that the interaction starting mode is more diversified, and the user experience is improved.

The embodiment of the invention provides a data processing model training method and device, and the method comprises the steps: determining a reference data processing model and an initial data processing model, and enabling the initial data processing model to be obtained through the parameter adjustment of the reference data processing model; inputting the target sample data into the reference data processing model to obtain a target reference result, and inputting the target sample data into the initial data processing model to obtain a target prediction result; obtaining a target weight according to the target reference result and the target prediction result, wherein the target weight is used for quantifying the difficulty level of the target sample data; according to the target weight, the target reference result and the target prediction result, training the initial data processing model to obtain a target data processing model; the target weight is introduced in a model training process for an initial data processing model, so that the model can automatically identify and focus on sample data rich in learning signals, the training efficiency is improved, and the distinguishing and learning ability for complex samples is enhanced.

The invention provides a multi-path recall method which is applied to the field of artificial intelligence. The multi-path recall method comprises the following steps that a processing device obtains a word to be retrieved; the processing device obtains M similarities between the to-be-searched word and the M training words; the processing equipment obtains a target weight group in a first weight set according to the training word corresponding to the maximum similarity in the M similarities, the first weight set comprises M weight groups, and each weight group in the M weight groups comprises N weights; the processing equipment inputs the to-be-retrieved word into N single-path recall models to obtain N first recall results; and the processing device fuses the N first recall results according to the target weight group to obtain a second recall result. In the technical scheme provided by the invention, through similarity matching, corresponding weights can be adaptively configured for different single-path recall models, and the recall effect is improved, so that the user experience is improved.

The application provides a large model-based weight recognition method and device and medium, and relates to the technical field of data processing. The method comprises the following steps: obtaining a maximum prediction value and a minimum prediction value of dish weight based on a shooting image of a target area obtained by a camera device; when a measured weight given by the target area is not between the maximum prediction value and the minimum prediction value, an early warning prompt is given; otherwise, a plurality of target sub-areas of the target area where a base plate is located are determined based on the shooting image, dish effective measured values of each target sub-area are obtained, a sum value of the dish effective measured values of all target sub-areas is obtained, and the sum value of the dish effective measured values of all target sub-areas is subtracted by an intermediate value to obtain a dish net weight as a target weight, so that the weight of the dish is accurately recognized.

The invention discloses a weight matrix processing method and device, equipment, a storage medium and a product. The method comprises the steps that a target weight matrix is loaded; wherein the target weight matrix is a third-precision weight matrix which is arranged in a preset format and is formed by packaging a first-precision weight matrix; carrying out unpacking and inverse quantization processing on the target weight matrix to obtain a weight matrix with second precision; wherein the third precision is higher than the second precision, and the second precision is higher than the first precision; and storing the weight matrix with the second precision to an on-chip storage unit so as to be loaded to a tensor core to execute matrix multiplication operation. According to the embodiment of the invention, the weight matrix can be compatible with the operation logic of tcore, so that the matrix multiplication operation can be smoothly completed in the tcore.

The invention relates to a method for detecting micronutrients of weight-losing meal replacement powder. The method comprises the following steps: analyzing the content of each component in a plurality of samples and the absorbance of each wave band in spectral data to obtain a characterization intensity value of a target component in each wave band; determining an aliasing influence coefficient according to a plurality of maximum value points in the characterization intensity vector of the target component and characterization intensity values of the other components in wavebands corresponding to the maximum value points; determining a weight coefficient of the target component according to the content of the target component in the plurality of samples, the content of each other component in the plurality of samples and an aliasing influence coefficient of the other components on the target component; decomposing the initial base spectrum matrix according to a target function to obtain a weight matrix and a base spectrum matrix; and inputting the base spectrum matrix and the spectrum vector of the target sample into the target function to obtain a target weight matrix, and inputting the target weight matrix into the regression model to obtain the content of each component in the target sample. According to the method, content results of various components can be accurately obtained.

The application provides a method and device for determining a sensor weight model of an unmanned device, equipment and a medium. Actual pose information and sensor collection information of the unmanned device at a historical time are obtained. Sensor collection residuals of the sensor at a to-be-processed time after a processable time of the unmanned device are determined according to a surrounding map and the sensor collection information at the historical time. At least one initial weight coefficient of the sensor is obtained, and the initial weight coefficient is iterated. The inference pose information of the unmanned device at the to-be-processed time is determined by using the currently iterated initial weight coefficient and the sensor collection residuals. The inference error of the currently iterated initial weight coefficient is determined according to the inference pose information and the actual pose information. The target weight coefficient is determined according to the inference error of the initial weight coefficient. The sensor weight model is obtained by training the neural network model using the sensor collection information and the target weight coefficient, and the sensor weight of the unmanned device in different environments is generated.

The application provides a training method of an industrial defect detection model and an industrial defect detection method. The weight and bias of the model are optimized through the energy value of the model. When the energy value reaches a dynamic balance state, the target weight is obtained by updating the candidate weight according to the updated temperature value, and the target bias is obtained by updating the candidate bias according to the updated temperature value. The trained industrial defect detection model is obtained according to the target weight and the target bias. The problem of gradient vanishing or explosion is avoided by optimizing the weight and bias of the network, and the stability of training and the performance of the model are improved.

An exercise recommendation system determines recommended weights for users to perform exercises with. The exercise recommendation system accesses a plurality of exercise pairs, each labeled with performance statistics of users who performed the exercises. Each exercise in an exercise pair is associated with a weight. The exercise recommendation system trains a machine learning model on the plurality of exercise pairs to determine a weight to recommend to a user for a first exercise based on performance statistics of the user associated with one or more second exercises, which are each in an exercise pair with the first exercise. The exercise recommendation system retrieves performance statistics of a target user including weights for exercises previously performed by the target user. The exercise recommendation system applies the machine learning model to the performance statistics to determine a target weight to recommend and modifies a user interface to include the target weight.

The application discloses a target detection method, a processing chip, an electronic device and a storage medium, and belongs to the computer field. The method comprises the following steps: acquiring environment data of a vehicle-mounted millimeter wave radar; determining a target index value of a to-be-detected unit in a range-Doppler matrix based on the environment data; determining a target weight value of a reference unit in the range-Doppler matrix based on the target index value, the reference unit being located around the to-be-detected unit, and the target weight value changing with the change of the target index value; and performing target detection by using a target constant false alarm rate detection algorithm based on the target weight value.

The application discloses a complex system performance evaluation method and device considering missing data, and relates to the technical field of performance evaluation. The method comprises the following steps: calculating an estimated value, data integrity, target weight and reliability of a test index according to observation data of the test index of a complex system; constructing a reference value of an evaluation grade space, and calculating an output utility of the complex system when containing missing data in combination with the estimated value, the target weight and the reliability; and calculating the sensitivity of the complex system according to the output utility and the data integrity, and evaluating the output utility of the complex system by using the sensitivity when the data integrity changes. The application introduces comprehensive indexes such as data integrity, target weight and reliability, can adapt to non-equal-interval characteristics caused by high-frequency and non-uniform-interval testing, reduces the interference of missing key characteristics on system performance evaluation, and dynamically captures the change trend of system output utility through correlation analysis of data integrity and sensitivity.

The invention provides a multistage precision control method of a powder liquid weighing system, and relates to the technical field of weighing control, and the method comprises the following steps: collecting characteristic parameters of a powder liquid material, and setting an initial precision grade according to the characteristic parameters of the material; based on the initial precision grade, three-stage segmented control is adopted for weighing, and the three-stage segmented control comprises a coarse feeding stage, a middle feeding stage and a fine feeding stage; in the weighing process, self-adaptive residual quantity compensation is carried out according to powder flying dust and liquid adhesion conditions; judging whether the target weight is reached or not according to the compensated weighing result, if not, continuing feeding, and if yes, stopping feeding; according to the invention, by constructing a material characteristic parameter acquisition and hierarchical control mechanism, accurate adaptation to characteristics such as powder flying dust and liquid adhesion is realized, a control strategy can be dynamically adjusted according to physical attributes of different materials, the problem of poor compatibility of traditional weighing equipment to multiple materials is effectively solved, and the weighing efficiency is improved. And the adaptation capability of a complex mixing process is obviously improved.

The invention discloses a dataset distillation method based on adversarial training trajectory matching, which comprises the following steps: selecting a real dataset, and randomly extracting a small part of data from the real dataset to form an initialized synthetic dataset; training the model by adopting a real data set, recording current parameters of the model at the end of each training period, forming an expert training track by the recorded model parameters, and retaining the weight at the end of each training period; training the model by adopting the synthetic data set, for each iteration, randomly selecting an expert training track, randomly selecting a weight at the end of a certain training period as an initial weight, and taking a weight after a plurality of rounds as a target weight; in the training process, the model performs multiple rounds of gradient descent on the synthetic data set based on the initial weight to obtain a new weight, and the matching loss between the new weight and the target weight is calculated; and according to the matching loss, a back propagation algorithm is adopted to optimize the synthetic data set.

The application discloses a model training method and device, a computer readable storage medium and computer equipment. The method comprises the following steps: obtaining an initial task model, wherein the initial task model comprises a pre-training model and a group of prefix parameters, the prefix parameters are hyperparameters of the initial task model, and the prefix parameters are used for adapting the initial task model to different tasks; training the initial task model based on training sample data to obtain target weights corresponding to the prefix parameters in the initial task model, wherein the training sample data correspond to a target task; and determining a target task model based on the target weights and the initial task model. The application solves the technical problem that the pre-training model with the prefix is prone to overfitting or underfitting due to the length of the prefix being too long or too short.

The present invention relates to a method for gathering a number of articles with a target total weight, the method comprising supplying articles to a plurality of weighing units in order to obtain measured weight values in articles, performing combinatorial calculations on the basis of the measured weight values to select a combination of these measured weight values the sum of which coincides with the target weight; and gathering the articles of the weighing units the measured weight values of which are part of the combination, wherein the weighing units are conveyor elements that are movable by levitation with respect to a work surface, capable of following different trajectories between at least two areas of the work surface of which one is a loading area, in which the articles are supplied to the weighing units; and another is an unloading area, in which the weighing units that are part of the combination dump their content to at least one article collector.

The invention provides an event importance judgment method and device, a storage medium and a program product, and the method comprises the steps: carrying out the event element extraction of a target event text cluster corresponding to a target event, and obtaining an event element corresponding to the target event; determining the propagation condition of the target event based on the media coverage and social media participation degree data corresponding to the target event; determining a target weight ratio and a target scoring rule corresponding to the event elements and the propagation conditions; scoring the event elements and the propagation condition based on a target scoring rule to obtain an event element score and a propagation condition score; and according to a target weight ratio, performing weighted summation on the event element score and the propagation condition score to obtain an importance degree score corresponding to the target event. The problem that the accuracy of the event importance degree judgment result is low possibly due to the fact that the importance degrees of events are judged and classified by means of keyword matching can be solved. The accuracy and the judgment efficiency of event importance judgment can be improved.

The embodiment of the invention provides a model operation method and device based on weight precision and computing equipment. The method comprises the steps of obtaining performance parameters of computing equipment; determining target weight precision according to the performance parameters; and based on the target weight precision, the target reasoning model is operated, and the precision of the weight parameter of the target reasoning model is the target weight precision. According to the embodiment of the invention, the target weight precision can be determined in real time based on the performance information of the computing equipment, and then the corresponding target reasoning model is operated, so that the performance bottleneck can be avoided, the waste of computing power resources can be avoided, and the balance between the computing performance and the utilization rate of the computing power resources is realized.

Embodiments of the present application disclose a data processing method and device, after obtaining a plurality of training samples to be labeled, the embodiments of the present application generate pseudo labels of each target object in the training samples, then determine a target weight set of the target object of each pseudo label category, and then train a target model according to the target sample carrying the pseudo label and the target weight set. Wherein, the target weight set includes the loss value weight of the target object of the corresponding pseudo label category when being predicted by the target model as belonging to different object categories. Thus, by training the target model using the target sample carrying the pseudo label and the target weight set, the embodiments can reduce the labeling cost of the training sample without affecting the accuracy of the trained model.

The invention provides a virtual character control method and device, a computer storage medium and electronic equipment, and relates to the technical field of computers. The method comprises the following steps: when a plurality of candidate drop point positions exist, determining target parameter values of a plurality of physical parameter factors corresponding to each candidate drop point position; based on the nonlinear mapping function of each physical parameter factor, determining an initial jump difficulty evaluation value which is in one-to-one mapping with the target parameter value of each physical parameter factor; obtaining an instantaneous jump speed, and adjusting the initial weight values of the plurality of physical parameter factors based on the instantaneous jump speed; and on the basis of the initial jump difficulty evaluation value of each physical parameter factor and the adjusted target weight value, determining a comprehensive jump difficulty evaluation value corresponding to each candidate drop point position, and determining a target drop point position from the plurality of candidate drop point positions so as to control the controlled virtual character to jump from the initial jump point position to the target drop point position. The method can improve the accuracy of jump difficulty assessment.

The present application relates to a model generation method and apparatus, and a computer device and a storage medium. The method comprises: acquiring weight parameter variations of models to be merged; selecting a target weight parameter variation from among the weight parameter variations, respectively grouping the weight parameter variations by using the target weight parameter variation as a grouping threshold, and determining a target grouping result on the basis of the similarity between grouping results; and acquiring a target group from the target grouping result, and merging weight parameters corresponding to the weight parameter variations in the target group, so as to obtain a target model.

The present disclosure provides methods, apparatuses and devices for coding and decoding. When determining to enable a weighted prediction for a current block, the method includes: for the current block, obtaining a weighted prediction angle of the current block; for each pixel position of the current block, determining a surrounding matching position pointed by the pixel position from surrounding positions outside the current block based on the weighted prediction angle of the current block; determining a target weight value of the pixel position based on a reference weight value associated with the surrounding matching position; determining an association weight value of the pixel position based on the target weight value of the pixel position; determining a first prediction value of the pixel position based on a first prediction mode of the current block; determining a second prediction value of the pixel position based on a second prediction mode of the current block; and based on the first prediction value, the target weight value, the second prediction value and the association weight value, determining a weighted prediction value of the pixel position; and determining weighted prediction values of the current block based on the weighted prediction values of all pixel positions of the current block.

The invention discloses an interaction method and device in a game, equipment, a storage medium and a program product, and relates to the technical field of man-machine interaction, the method comprises the following steps: collecting face data, physiological data and sound data of a game user; determining a first target weight, a second target weight and a third target weight based on the acquisition results of the face data, the physiological data and the sound data; analyzing and determining the current emotional state of the game user according to the face data, the physiological data, the sound data, the first target weight, the second target weight and the third target weight; based on the matching relation between the current emotional state and the interaction state, determining an interaction strategy for the game user; and controlling the virtual object in the game to interact with the game user according to the interaction strategy. According to the technical scheme, the interaction strategy can be dynamically adapted based on the real-time emotional state of the user, and the emotional authenticity and immersion of man-machine interaction are improved.

The invention provides an unmanned equipment sensor weight model determination method and device, equipment and a medium, and the method comprises the steps: obtaining the actual pose information and sensor collection information of unmanned equipment at historical moments, and obtaining a weight model of an unmanned equipment sensor according to a surrounding map and sensor collection information of the unmanned equipment at a processable moment in the historical moments; determining a sensor acquisition residual error of the sensor at a to-be-processed moment after the processable moment; acquiring at least one initial weight coefficient of the sensor, traversing the initial weight coefficient, and determining reasoning pose information of the unmanned equipment at the to-be-processed moment by adopting the currently traversed initial weight coefficient and the sensor acquisition residual error; according to the reasoning pose information and the actual pose information, the reasoning error of the initial weight coefficient of the current traversal is determined; determining a target weight coefficient according to the inference error of the initial weight coefficient; and training the neural network model by adopting sensor acquisition information and a target weight coefficient to obtain a sensor weight model, and further generating sensor weights of the unmanned equipment in different environments.

The invention discloses a business environment assessment method and device, equipment, a storage medium and a program product, and the method comprises the steps: obtaining target index data, corresponding to a to-be-assessed time point, of a to-be-assessed object in a plurality of business environment assessment indexes to which the assessment dimensions belong for a plurality of assessment dimensions; determining a target weight of the business environment evaluation index at the to-be-evaluated time point according to the reference weight and the weight time sequence feature of the business environment evaluation index for the plurality of business environment evaluation indexes; and determining a business environment assessment result of the to-be-assessed object at the to-be-assessed time point according to the target index data of the plurality of business environment assessment indexes at the to-be-assessed time point and the corresponding target weights thereof. According to the technical scheme, the target index data and the target weight adapted to the to-be-assessed time point are combined, the assessment result capable of accurately representing the business environment change condition of the to-be-assessed object at the to-be-assessed time point is determined, and the precision of business environment assessment is improved.

The embodiment of the invention discloses a dynamic rail weighbridge weighing method and device, equipment and a storage medium. At least one weighing section is controlled to collect the weight of at least one carriage of a target train. And determining a corresponding weight characteristic value according to the weights of the plurality of carriages corresponding to each carriage. And for each carriage, abnormal value screening is carried out on the weights of the multiple carriages according to the weight characteristic values, and the corresponding target weight is calculated according to the weight of at least one carriage after screening. According to the embodiment of the invention, dynamic measurement is carried out on the train carriages through the plurality of weighing sections, and the optimal measurement value of the weight of the carriages is obtained through fitting after the measurement data is analyzed and processed by adopting a robust statistical technology, so that the weighing accuracy of the continuous-rail dynamic rail weighbridge is improved.

This application relates to a data processing method, apparatus, computer equipment, and storage medium. The method includes: acquiring a job and determining its target job type; querying a target database for the target job type's corresponding target weight, where the target weight is calculated based on the indicator value of the target performance event of a previous job of the same target job type; determining the target binding method corresponding to the target weight based on a preset correspondence between weights and binding methods; and binding the job based on the target binding method. By employing this method, the binding method of the job is determined based on the calculated weight of the current job. This allows for real-time adjustment of the current job's weight and its corresponding binding method based on the indicator values ​​of the performance events of preceding jobs, thereby ensuring the rationality of allocating computing resources to the job and improving job execution efficiency.

The present disclosure relates to an index selection method and device, readable medium and electronic equipment. The method comprises: obtaining medical data and a plurality of indexes of the medical data; classifying a plurality of indexes in the medical data to obtain first indexes and second indexes, the first indexes including indexes with an importance higher than a preset threshold, and the second indexes including indexes with an importance lower than the preset threshold; determining a first target weight coefficient of each index in the first indexes; performing fusion processing on the second indexes to obtain a fusion index and determine a second initial weight coefficient of the second indexes; adjusting the second initial weight coefficient according to the first indexes and the fusion index to obtain a second target weight coefficient; and determining a target index from the plurality of indexes according to the first target weight coefficient and the second target weight coefficient.