57 results about "Computation history" patented technology

The invention provides a method and device for predicting telecom traffic. The method comprises the following steps: determining the prediction granulation of the telecom traffic; selecting a historical sample and a prediction sample; respectively calculating the growth rates and the initial values of the telecom traffics of the historical sample and the prediction sample by using a unary linear regression model; calculating the prediction traffics of the historical sample and the prediction sample; reading the actual traffic of the historical sample; and adjusting the second prediction traffic according to the deviation between the traffic of first prediction and the actual traffic to obtain the prediction traffic adjusted by a user. The behavior mode of the user is obtained through analyzing the deviation between the linear prediction value and the actual value in the historical sample, and accordingly, the linear prediction value of future telecom traffic data is adjusted to obtainmore accurate telecom network traffic, thereby providing decision support for the user to plan, organize and manage network more accurately.

The invention discloses an intelligent detection method for server exception of a hybrid strategy. The method comprises a historical data acquisition and analysis step and a real-time anomaly alarm step, wherein the historical data acquisition and analysis step comprises the steps of acquiring server historical data in a preset time period, preprocessing the historical data, screening out featuresneeding to be analyzed, and taking data corresponding to each feature as a time sequence, determining a normal value and an abnormal value corresponding to each time sequence in the historical data in combination with a time sequence decomposition algorithm and a Grubbs test algorithm; calculating a normal threshold value corresponding to each feature in the historical data, i.e., a maximum valuerange and a minimum value range of a normal value in each time sequence; and taking all time sequences in the historical data as training samples, establishing a plurality of abnormal data detectionmodels to predict and judge the to-be-detected sample, and outputting the probability that the to-be-detected sample is abnormal, so as to analyze the real-time data in the real-time alarm step.

Creation of new edges in a network may be used as an indication of a potential attack on the network. Historical data of a frequency with which nodes in a network create and receive new edges may be analyzed. Baseline models of behavior among the edges in the network may be established based on the analysis of the historical data. A new edge that deviates from a respective baseline model by more than a predetermined threshold during a time window may be detected. The new edge may be flagged as potentially anomalous when the deviation from the respective baseline model is detected. Probabilities for both new and existing edges may be obtained for all edges in a path or other subgraph. The probabilities may then be combined to obtain a score for the path or other subgraph. A threshold may be obtained by calculating an empirical distribution of the scores under historical conditions.

The embodiment of the invention provides a recommendation method, a recommendation device and a mobile terminal. The recommendation method comprises the steps that the item score data of users are acquired, wherein the item score data include historical score values and corresponding historical score time of the historical score values; the corresponding time weight of the historical score values is calculated according to the historical score time; a user similarity value is calculated by using the historical score values and the time weight; and the corresponding item is recommended to the target users based on the user similarity value. According to the recommendation method and the recommendation device, the item score data including the historical score values and the historical score time are acquired, the corresponding time weight of the historical score values is calculated according to the historical score time, and finally the user similarity value is calculated by using the historical score values and the time weight and the corresponding item is recommended to the target users based on the user similarity value so that the objective of recommending the item having high matching degree with the interest of the current user according to the real-time change of the user interest can be realized.

The invention is applicable to the technical field of computers, and provides an article recommendation method and a device. The method comprises the following steps of according to the similarity between a scored article and a to-be-scored article in historical scoring data, calculating the influence factor of the historical scoring data on the predicted score of the to-be-scored article; inputting the calculated influence factor into a pre-established restricted boltzmann machine model; calculating the predicted score of a user on each to-be-recommended article through the restricted boltzmann machine model; generating a recommendation list; and outputting recommended articles to the user according to the generated recommendation list. In this way, the cold start problem during the new article recommending process can be solved. The recommendation accuracy is improved.

The invention relates to a medical benefit fund actuarial prediction method and device. The method comprises the steps that historical basic medical insurance structure data, historical insured demographic data and historical medical expense data sent by a terminal are received; base year insured demographic data is acquired according to the historical insured demographic data, a historical insured demographic average growth rate is calculated, and a predicted year insured population is calculated according to the base year insured demographic data and the historical insured demographic average growth rate; predicted year medical benefit fund inflow is determined according to the predicted year insured population and the historical basic medical insurance structure data; predicted year medical benefit fund outflow is determined according to the predicted year insured population and the historical medical expense data; and when the predicted year medical benefit fund inflow is smaller than the predicted year medical benefit fund outflow, risk prompt information is sent to the terminal. Through the medical benefit fund actuarial prediction, medical benefit fund actuarial early warning analysis efficiency is greatly improved, and prediction precision is effectively guaranteed.

The invention provides an online prediction method for future tool wearing capacity. The tool wearing capacity for some time to come is predicted by taking tool wearing capacity data in some time agoas input. The online prediction method for the future tool wearing capacity comprises the steps that firstly, influences of the historical wearing capacity on the future wearing capacity are calculated through a long-short-term memory unit encoder, and a state tensor is generated; secondly, the state tensor is taken as input of a long-short-term memory unit decoder, and the wearing capacity for some time to come is generated through the decoder; and in the encoding and decoding process, the encoder, the decoder and the state tensor form a recurrent neural network for predicting future wearingcapacity changes, internal parameters of the long-short-term memory unit encoder and the long-short-term memory unit decoder are automatically obtained through an adam algorithm, and influence factorsof the historical wearing capacity are adjusted. According to the online prediction method for the future tool wearing capacity, the tool wearing capacity evolution trend prediction problem is solved, and the online prediction method for the future tool wearing capacity has the characteristics that the process is easy and convenient to operate, the processing speed is high, prediction is accurate, and the generalization performance is good and can be suitable for the cutting process under different working conditions.

Embodiments are described for a system and method for adapting an online user input form that provides user input to a web-based application. A method comprises defining one or more attributes related to one or more fields of the online user input form, obtaining historical data comprising past user input to related user input forms, calculating historical most common values for each attribute of the one or more attributes; and modifying the one or more fields of the online user input form in accordance with the calculated historical average values. The attributes may include at least one of the size, location, and datatype of each field of the form. The method may be implemented through an application programming interface for implementation through server-side and client-side processes that are written in a web-browser compatible markup language.

The embodiment of the invention provides a data uploading method and device based on a consensus mechanism and a readable storage medium. Under the condition that a center node generates a new block,broadcast information corresponding to the new block is broadcast to all consensus nodes, after receiving the broadcast information, each consensus node feeds back read-write performance information for the block data corresponding to the historical block number and calculation performance information obtained by calculating the block data corresponding to the historical block number by adopting the random data and the calculation round number; the central node feeds back read-write performance information and / or calculation performance information based on each consensus node, determines network performance information of a consensus node; based on a consensus mechanism, read-write performance information, calculation performance information and network performance information based on each consensus node are comprehensively considered, candidate nodes for packaging and chaining new blocks are determined more reasonably, finally, the number of historical blocks generated by each candidate node is balanced and considered, and a target node is determined from the candidate nodes more fairly.

The invention discloses a trust management method based on a nested game in a center based cognitive wireless network. The method comprises the steps of establishing a nested game model, perceiving a spectral state, making a secondary user select a perception stage strategy and upload the perception data, making a data center fuse the perception data, making the secondary user select a transmission stage strategy, selecting a sliding window value, calculating a historical credit value and the credit value of this time based on the strategy, calculating utility functions of the first and second stages, optimizing the utility functions based on the game theory to solve the optimal strategy, updating trust function values, and distributing the spectrum based on the ranking of the trust values. The method aims at the buildup of the whole cognitive cycle, and uses the nested game theory and the marginal utility theory to be capable of effectively resisting malicious attacks. The cognitive process is classified into the perception stage and the data transmission stage. A secondary user can assess the credit value in the strategies in different periods of time. Secondary users game each other to acquire the spectrum, eliminate the malicious users and make the whole system tend to be better.

Embodiments for efficiently computing complex statistics from historical time series data are provided. A hierarchical summarization method includes receiving at least one stream of data and creating data blocks from the at least one stream of data. In another embodiment, a method for computing statistics for historical data includes accessing at least one online stream of historical data, the online stream of historical data including metadata, and creating data blocks from the at least one online stream of historical data. Each data block includes a pair of timestamps indicating a sampling start time and a sampling end time, a number of data samples spanned by the data block, a SUM(X) statistic, a SUM(XX) statistic, and a SUM(XY) statistic computed for the data samples spanned by the data block. Other methods are also presented, such as methods for efficiently and accurately calculating statistical queries regarding historical data for arbitrary time ranges, among others.

The invention provides a project case screening method and device. The project case screening method comprises the steps: obtaining age information and occupational information recorded in case details of a historical audit case where audit group members participate in audit; calculating historical audit age distribution and historical audit occupational distribution according to the number of cases of the historical audit cases; determining an audit recommendation age interval and an audit recommendation vocational set; retrieving a to-be-audited case with age information recorded in the case details in an audit recommendation age interval and recorded occupational information in an audit recommendation occupational set in the project case library; and screening to-be-audited cases of which the matching degrees with the age information and the occupational information of the members of the auditing group meet a preset matching degree threshold value from the retrieved to-be-audited cases, so that, the effectiveness of screening the to-be-audited cases in the project is improved, and meanwhile, the accuracy of pushing, auditing and voting the to-be-audited cases on the basis of the screened to-be-audited cases is higher.

The invention discloses a grounding resistor intelligent monitoring method and device, a storage medium and computer equipment. The grounding resistor intelligent monitoring method is applied to an intelligent monitoring system. The intelligent monitoring system comprises a server. The intelligent monitoring method comprises the steps of according to a preset mode, acquiring a first resistance value of a grounding resistor, wherein the first resistance value is an initial resistance value of the grounding resistor; acquiring historical record values of the grounding resistor in a preset time period from a server; calculating an average value of the historical record values according to the historical record values, and generating a threshold interval according to the average value; judgingwhether the first resistance value is in the resistance value range of the threshold interval; and if yes, judging that the first resistance value is a normal value, and sending the first resistancevalue to the server. Compared with the prior art, the scheme has the advantage that the reliability of the test result of the grounding resistor is improved.

The invention discloses a collaborative recall method based on user clicking and conversion duration feedback. The collaborative recall method comprises the following steps of acquiring a historical behavior log of a user; storing the filtered data of the historical behavior log in a first database; calculating the historical average conversion duration of each click video in the historical behavior log and storing the historical average conversion duration in a second database; performing interval division on each click video in the historical behavior log, calculating a preference score of the user for each time interval, and storing the preference score in a third database; recalling the candidate video set pushed by the recommendation system, and calculating a sorting score of each candidate video; and recommending the first N candidate videos to the user according to the ranking score. According to the method, the problem that only the feedback click rate is considered in an existing video recommendation technology, and other duration factors are not considered, so that the overall duration of the system is shortened, is effectively solved, and a final recommendation result ismore accurate.

The invention discloses a resource channel switching method and device, equipment and a storage medium. The method is to the field of computer programs, the method comprises the following steps: historical resource transfer information is obtained from the terminal through the asynchronous data burying point; storing the historical resource transfer information into a database; acquiring historical resource transfer information of the target object in a recent time period from a database, calculating a transfer success rate of resource transfer through the first resource channel in the historical resource transfer information, determining a second resource channel different from the first resource channel in response to the transfer success rate being smaller than a preset threshold, and switching the first resource channel used by the target object to the second resource channel before channel switching; wherein the latest time period is a minute-level time period, monitoring and switching of the resource channels are carried out in the minute-level time period, and switching of the resource channels can be carried out in a shorter time dimension under the condition that the resource transfer requests are highly concurrent.

The invention relates to the technical field of a power frequency parameter calculation method of a high voltage transmission line, belonging to the application of computer technology in estimation of power frequency parameters of the high voltage transmission line. The power frequency parameter calculation method of the high voltage transmission line comprises three modules including a parameter calculation module, a historical data query module and a parameter setting module. By the application of the calculation method, a fine human-computer interface is provided; the calculation method can be used on various types of poles and towers, thus the characteristic of wide application is realized; simultaneously the accuracy rate of line parameter calculation is improved; the storing function of test data, the output of a testing program and the print of a test report can be realized, and the work efficiency of staff is enhanced greatly; and a parameter upgrade window is reserved to facilitate future increase of physical data. The calculation method achieves the aims of use convenience for testers, and accurate and reliable calculation results, and the technical problems in actual production are solved.

A method for assessing underwriting and distribution risks associated with a portfolio of subordinate debt is provided. The method is performed using a computer system coupled to a database. The method includes storing in the database historical bond issue data for a period of time preceding and proceeding at least one historical liquidity event and generating a plurality of simulated subordinate debt warehouses using the computer and the historical bond issue data stored in the database. The method also includes calculating a historical loss distribution based on the plurality of simulated subordinate debt warehouses generated. The method also includes determining a value at risk for a portfolio of subordinate debt resulting from a potential liquidity event by applying the historical loss distribution to the portfolio of subordinate debt.

The invention discloses a big data access method based on artificial intelligence, and the method comprises the steps that a cloud server obtains massive multi-source heterogeneous data, and carries out the hierarchical storage of the massive multi-source heterogeneous data; hierarchical snapshot is performed on the massive multi-source heterogeneous data stored hierarchically to generate a plurality of snapshots of the massive multi-source heterogeneous data; read data pressure values at the first N moments and storage capacity saturation values at the first N moments are acquired, and historical data read intensity is calculated; a historical data reading intensity low-frequency component and a high-frequency component are obtained; the low-frequency component is predicted, and first data reading intensity at the (N + 1) th moment is predicted; the high-frequency component is predicted, and second data reading intensity at the (N + 1) th moment is predicted; the data reading comprehensive intensity at the (N + 1) th moment is calculated, and the number X of snapshots needing to be increased is calculated based on the data reading comprehensive intensity; X snapshots are sequentially generated, X * Y clones are created through the X snapshots, and the read request is responded based on the X * Y clones.

A method of learning a structure of a Bayesian network includes computing an ordering of the random variables of the Bayesian network; wherein computing the ordering of the random variables of the Bayesian network is performed by computing an approximate solution to the history dependent traveling salesman problem.

The invention discloses a power load prediction method based on phase space reconstruction and data driving, and the method comprises the steps: firstly calculating the delay time and embedded dimension of historical load data, dividing a data set into a training set and a test set, carrying out the phase space reconstruction of the training set and the test set according to the delay time and embedded dimension, and carrying out the normalization processing; secondly, comparing the prediction precision of the twelve data driving models under the same data set, wherein the obtained XGBoost isthe optimal model in the statistical learning model, tand he LSTM and the extreme learning machine are the optimal models in the neural network; carrying out model, day, week, half month and month load prediction on the three models respectively, and keeping high prediction precision; and finally, performing unequal weight combination on the XGBoost, the LSTM and the extreme learning machine by using a grey relational degree method, and constructing a combined data driving model. The method can improve the prediction precision.

The invention discloses an equipment health state assessment method. The method comprises the following steps: firstly, selecting an absolute mean value and a root-mean-square value as features for describing related parameters; calculating sphere centers of the historical vibration parameters and the real-time vibration parameters through a support vector description algorithm model, and constructing a health index calculation model through the sphere centers; and finally, constructing a curve of the running state of the equipment through a membership function, importing a calculation result of the health index calculation model into the curve of the running state of the equipment to calculate a corresponding health evaluation value, and judging the running state of the equipment through comparison of the health evaluation value and a standard value. According to the method, historical parameters and real-time parameters of equipment operation are selected as basic data, characterization indexes of the health state of the equipment spindle can be established without failure parameter data and degradation characteristics of the equipment, a characteristic threshold value when the spindle fails is deduced, and prediction of the failure state of the equipment is achieved; therefore, accurate evaluation of the operation state of the equipment is realized, and technical support is provided for fault diagnosis and predictive maintenance.

Associated with an instruction in a program is a computation history. The computation history represents all objects that affect the result of the instruction, such objects including (but not limited to) registers, memory locations, static values, and instruction program counters. The computation history may be used to make a prediction about a property of the instruction.

Disclosed is method and system for turning existing object-oriented programming languages into smart contract languages without introducing new syntactic features. The invented method and system provide a protocol that enables storing a history of computations on a decentralized computer network, such as a UTXO-based blockchain system, for any object-oriented computer language. The invented method and system further provide for storing and updating data on blockchains, where such blockchains may be used in cryptocurrency applications and for smart contracts.

The invention discloses a student personalized time interval perception attention mechanism knowledge tracking method, which comprises the following steps of: 1) collecting historical interaction data of student learning and timestamps for completing learning, and designing and realizing a student learning time interval relation matrix; 2) carrying out vector coding on the exercise sequence, the learning sequence, the absolute position and the personalized time interval; 3) establishing a time-aware attention mechanism to calculate the influence weight of the historical completed exercises on the subsequent new exercises, and summarizing the initial knowledge state of each node student based on the weight; and 4) fusing the preliminarily summarized historical knowledge states of the students by using a full connection layer, and tracking continuously changing knowledge states in the learning process of the students. Compared with other technologies, the method has the advantages that different time intervals for students to practice each question are effectively utilized, different knowledge mastering conditions of different students under the same learning sequence are mined, and the knowledge tracking accuracy is improved.

The invention relates to a training method and device of a cycle prediction model, a cycle prediction method and device, and relates to the technical field of machine learning. The method comprises the steps of calculating a historical order behavior sequence of a historical user and an actual re-purchase cycle of the historical articles included in a historical order according to the historical user data; constructing a data set according to the historical articles, the actual re-purchase cycle, the historical order behavior sequence and the historical user portraits in the historical user data; inputting the historical articles, the historical order behavior sequence and the historical user portrait included in the data set into a to-be-trained period prediction model to obtain a predicted re-purchase period of the historical article; and constructing a first loss function according to the actual re-purchase cycle and the predicted re-purchase cycle in the data set, and training the to-be-trained cycle prediction model by using the first loss function to obtain a trained cycle prediction model. The method improves the accuracy of the cycle prediction model.

Some embodiments of the present specification disclose a method and a system for synchronizing data. The method comprises the steps of performing calculation through a big data platform to obtain a current to-be-synchronized data table; obtaining a version number of the current to-be-synchronized data table according to the calculation history of the big data platform and the historical to-be-synchronized data table; splicing a main key of the current to-be-synchronized data table and the version number to obtain a row main key; and storing the row main keys and the corresponding numerical values into a key value pair database to facilitate query.

According to the high-performance computing resource scheduling fair sharing method provided by the invention, more fair resource sharing is realized by calculating the weight of historical resources used in the dynamic quota, and the user service quality of a high-performance computing system is improved. The leaves to which the tasks belong are searched through the hash table, and the sorting of the shared tree nodes is optimized after each task is successfully scheduled, so that the scheduling speed is improved, and the high throughput of the system is realized.