73results about "Machine learning" patented technology

The invention discloses a batch data processing method and device, computer equipment and a storage medium. The batch data processing method comprises the steps: selecting a target batch task from a non-real-time task queue, and creating a data processing queue based on the target batch task; determining target idle information from the target idle time queue, obtaining the number of to-be-processed data corresponding to the to-be-processed data, obtaining the target number based on the number of the to-be-processed data, the target idle duration and the estimated thread number, selecting theto-be-processed data corresponding to the target number from the data processing queue, and determining the to-be-processed data as segmentation processing data; when the original load of the system is smaller than a busy load threshold value, obtaining a target processing thread, adopting the target processing thread to perform data processing on the segmentation processing data in the target idle duration, and obtaining a data processing result; and based on the data processing result, updating the task state of each split processing data in the data processing queue. According to the batchdata processing method, system resources can be reasonably distributed, and the batch data processing efficiency is ensured.

A computer-implemented method, computer program product and computing system for defining a model that includes a plurality of variables. At least one variable of the plurality of variables is conditioned based, at least in part, upon a conditioning command received from a user, thus defining a conditioned variable. The model is inferenced based, at least in part, upon the conditioned variable.

The present invention discloses a regex matching method and system, and relates to the field of computer technologies. The method includes: sorting multiple regexes into several regex groups, where all regexes in one regex group include a common string, which is known as a generic string; compiling each regex group into a DFA, and setting up a correlation between the generic string of each regex group and the DFA; matching to-be-matched data streams with the generic string respectively, and using the matched generic string as a matched string; obtaining a DFA corresponding to the matched string; and performing regex matching for the to-be-matched data streams according to the DFA, and outputting a matching result. The embodiments of the present invention shorten the data loading process, decrease the time consumed by data loading, and improve the matching performance.

An incremental self-learning based dialogue apparatus for dialogue knowledge includes a dialogue processing unit configured to determine a intention of a user utterance by using a knowledge base and perform processing or a response suitable for the user intention, a dialogue establishment unit configured to automatically learn a user intention stored in a intention annotated learning corpus, store information about the learned user intention in the knowledge base, and edit and manage the knowledge base and the intention annotated learning corpus, and a self-knowledge augmentation unit configured to store a log of a dialogue performed by the dialogue processing unit, detect and classify an error in the stored dialogue log, automatically tag a user intention for the detected and classified error, and store the tagged user intention in the intention annotated learning corpus.

The invention discloses a telecommunication fraud prevention system and method based on big data and machine learning. The system comprises a mobile terminal, a big data analysis terminal and a fraud interdiction governance terminal, wherein the mobile terminal is used for performing fraud detection determination on current telecommunication data according to predetermined constraint rules when receiving a short message or an incoming call message, a machine learning algorithm is adopted to detect whether the telecommunication data is a telecommunication fraud, and if the detection result is that the telecommunication data is determined as a telecommunication fraud, fraud data information is uploaded to the big data analysis terminal; the big data analysis terminal is used for performing real-time statistics on the fraud data information uploaded and reported from the mobile terminal and sending fraud early-warning information to the fraud interdiction governance terminal according to a bank card account or/and a phone number with the number of received reports exceeding a certain threshold value; and the fraud interdiction governance terminal is used for taking corresponding measures in time to interdict occurrence of a telecommunication fraud event when receiving the fraud early-warning information. The system can unite the mobile terminal, an operator, a public security institution, a bank and other institutions, quick and effective prevention can be realized, and the telecommunication fraud can be cracked down in time.

An object of the invention is an automated analysis system for analyzing at least one of scientific, technological and business information. The automated analysis system comprises means 100 for processing a data amount to accomplish a collection data form, means for structuring the collection data form describing the content of a source document, means for automatically identifying input documents in data warehouses comprising similar structured data forms as said structured collection data form, means for preprocessing and tokenizing said input documents, and the automated analysis system comprises a classification engine using the tokenized and preprocessed documents as input classifying each item of the input documents, and said means for processing, means for structuring, means for automatically identifying, means for preprocessing and tokenizing and said classification engine being configured to perform their tasks in one or more modelling round(s).

A system and method for the dynamic allocation of resources based on fairness, throughput, and user behavior measurement. A resource allocation decision can be made based on an index value computed by a selection index function, A fairness coefficient and a throughput coefficient, which represents the significance of fairness and throughput can be computed utilizing a reinforcement learning algorithm and the degree of fairness and throughput coefficient can be varied while allocating resources. A user behavior coefficient with respect to a user can be computed to determine the degree of cooperativeness of the user with other users and the value of user behavior coefficient can be updated each time it interacts with the system.

Techniques are presented for providing misinformation detection in online content. The described techniques can identify instances of misinformation in online content and pass a misinformation result to the user. A misinformation probability analysis can be performed by applying a syntactic analysis and a semantic analysis to detect misinformation with confidence by applying featurization to a URL, text of content referenced by the URL, and metadata associated with the URL using a feature set, the feature set comprising semantic-based features and syntactic-based features, wherein the semantic features and the syntactic features are selected from the group consisting of: sentiment amplifiers, sentiment continuity disruption features, lexical features, keywords, baseline features, speech act, sensicon features, emotion detection on the obtained text, exaggerated language, strong adjectives, heuristics, bag-of-words, objectivity, colloquial-ness score, and semantic difference.

The embodiment of the invention provides a feature processing method and device for a user classification model. The method comprises the steps that firstly, label data tables and first feature tablesare acquired, wherein each first feature table records a plurality of features of a user; and for each feature in each first feature table, a feature IV value is calculated, and first screening operation is performed on the features based on the IV value to obtain a corresponding second feature table. Then, the second feature tables and features in the second feature tables serve as first class nodes and second class nodes respectively, a bipartite graph is constructed, the minimum number of the first class nodes connected to all the second class nodes is determined in the bipartite graph, and then corresponding M second feature tables are obtained; then, the M second feature tables are combined to obtain a comprehensive feature table, and correlation coefficients between features are calculated based on the comprehensive feature table; and a second screening operation is performed on the features based on the correlation coefficient to obtain a plurality of selected features for training a user classification model.

A chip recognition system in which a chip is configured to at least partially have a specific color indicative of a value of the chip includes: a recording device that uses a camera and records a state of the chip as an image; an image analysis device that subjects the image so recorded to image analysis and recognizes at least the specific color and a reference color that is present in the image and differs from the specific color; and a recognition device at least including an artificial intelligence device that uses a result of the image analysis by the image analysis device and specifies the specific color of the chip, wherein the artificial intelligence device of the recognition device has been subjected to teaching using, as training data, a plurality of images of the chip and the reference color irradiated with different illumination intensities.

A contribution management system and process for facilitating the identification of individual users who have made contributions to an electronic content item, and the extent of that contribution. As an example, a reader may review the list of contributors with a representation of their relative degree or amount of contribution to the document, and also allow for access to other documents that have been developed by the identified contributor. These tools can provide collaborative document users the ability to more clearly distinguish casual contributors or non-contributing owners of the document from higher-level contributors whose authorship has significantly shaped the content. In addition, the listing or identification of top contributors can be configured to facilitate communication between an interested reader and the identified contributor, which increases the ease with which members of larger organizations can collaborate, seek mentorship, or develop useful relationships.

Disclosed are methods, circuits, devices, systems and functionally associated machine executable code for enhanced automated software code testing. A system for enhanced automated software code testing comprises a processing module for wrapping test script commands, of a software testing framework, with command execution monitoring or control code. The command execution monitoring or control code, is configured to collect and report test script execution parameters, resulting from test script executions. A failure root cause identification module automatically determines one or more root causes of a failure in the execution of a test script run, based on the analysis of test script execution parameters from prior test runs.

The invention discloses a product quality control method based on a process network model and a machine learning algorithm. According to the method, product quality control is realized in a mode of fusing a product process complex network and a machine learning algorithm, namely, a quality transmission complex network is established based on a product process mechanism, a machine learning XGBoost model is established based on the quality transmission network and sample data and trained, and an SHAP algorithm model is established to analyze an XGBoost training result. Key process parameters influencing quality are quantitatively mined, and the linkage effect among the process parameters is accurately calculated. The method has the advantages that a product process mechanism and a big data analysis method are effectively combined, so that the problems of high dimension, strong nonlinearity and non-uniform sample distribution of industrial actual production data, which cannot be solved by a common statistical analysis method, are effectively solved, and the defect that a traditional product quality control method carries out qualitative description or only carries out product quality control for a single independent factor is overcome; and the method can comprehensively and quantitatively analyze the influence of complex process factors in the whole process of product processing on the final product quality and accurately calculate the linkage influence among the process factors to form a quality identification model based on process mechanism and data dual drive, and is large in processing data volume, high in speed and accurate in evaluation result.

The invention discloses a strip steel exit thickness prediction method by a symmetric extreme learning machine (Sym-ELM) optimized by a random frog leaping algorithm. Symmetry prior information is firstly added to the traditional extreme learning machine, the improved extreme learning machine can process a class of data containing symmetric properties, a symmetric extreme learning machine is put forward, the improved Sym-ELM algorithm is further optimized, the random frog leaping algorithm is used to select the input weight and the hidden layer bias value of the Sym-ELM algorithm, the optimized extreme learning machine is used for strip steel exit thickness prediction, and optimization on the symmetric extreme learning machine algorithm not only minimizes a root mean square error value but also considers a normal condition number of the output matrix of the hidden layer. The experiment shows that in comparison with the traditional extreme learning, the prediction performance of the traditional extreme learning machine to process the strip steel data containing the symmetric properties is improved, prediction errors of the model are reduced, and the generalization performance of the model is improved.

A method, system, and computer program product are provided for displaying tradeoff options for a plurality of items (e.g., treatment recommendations) and options (e.g., treatment outcomes) by analyzing the options based on user-specified preference data and weighting factors to identify a list of options for a comparative list of items, and then employing a user interface to display a comparison of tradeoff options for the comparative list of items as a treatment matrix which includes visual tradeoff indication for each option shared in common with the comparative list of items to assist with the comparison of tradeoff options for the comparative list of items, where the user-specified preference data and/or weighting factors may be dynamically modified through interaction with a user interface to specify importance and impact weighting factors.

The invention discloses a software defect prediction method and device and electronic equipment, and the method comprises the steps: obtaining a feature vector of to-be-predicted software; and determining a software defect prediction result of the to-be-tested software based on the feature vector and a pre-trained prediction model used for software defect prediction with the prediction model beingobtained based on a gradient lifting algorithm and random forest learning machine training. By adopting the method provided by the invention, the gradient lifting algorithm and the random forest learning machine are used for training to obtain the training model for software defect prediction, so that the software defect prediction result output by the prediction model obtained by training is relatively high in accuracy, and meanwhile, relatively large influence on the calculation complexity is avoided.

The method of the present invention makes it possible to recognize partial discharges acquired by means of sensors in electrical networks, comprising a series of steps, among which are a post-processing step (13) of the acquired signals and a recognition step (17) of said signals by means of a convolutional neural network (CNN). The method also includes adaptation (15) and training (16) steps of the neural network, as well as a step to build a library (14) of partial discharge signals from known sources that serve as training of the convolutional neural network (CNN).

The present invention discloses an intelligent toilet seat ring control method, an electronic device and a storage medium. The method comprises: detecting a received wireless signal, wherein the wireless signal is sent by a signal transmitting device, and is reflected and received; when detecting that a reflection object is within a set distance threshold, performing identification; and after detecting that the reflection object is near the toilet movement, if a standing movement is identified, that the reflection object has a human chest movement is identified, and/or that the reflection object has a human standing pee posture is identified, controlling the toilet seat ring to be opened. According to the technical scheme of the present invention, by identifying the characteristic movementof the male urination behavior, the toilet seat ring and cover can be automatically flipped, so that the sanitary problems caused by manually flipping the toilet seat cover and urination liquid in the toilet seat ring caused by male child's lack of flipping movement can be avoided.

The invention discloses a security federated learning logistic regression algorithm. The algorithm comprises the following steps that a host calculates E (WBXB) and sends the E (WBXB) to a gain; a gain calculation calculation calculation calculation E (delta y) = E (yy ') is sent to a host, and then a local model parameter W' A is updated by calculating a gradient value LA; if isstop is not equal to 0, a gradient value LB is calculated, a model parameter E (W 'B) is updated and then the step 1 is repeated: if isstop is equal to 0, the next step enters; guest returns to WA, the host selects a random vector and the W 'B is sent to which disturbance is added to the guest; the guest helps decrypt the W 'B and sends the W' B to the host; and 6, the host returns the model parameter WB. The system architecture provided by the invention can also be easily expanded to support multi-party model training, a joint model can be trained on a large corpus of dispersed data owned by different parties, meanwhile, the privacy of the data is kept, and the precision of the model is ensured.

The invention provides a link library function name recognition method and device for a computer binary program. The method comprises the steps of archiving and sorting a static link library is collected; the collected static link libraries; writing the trained machine learning model into an IDA plug-in, and loading and using the IDA plug-in in IDA; and enabling the IDA to receive a suspicious malicious binary program input by a user, call the IDA plug-in compiled by the trained machine learning model to perform automatic analysis and detection, and displays a detection result in an IDA interface. According to the scheme, reverse analysis personnel can conveniently analyze unknown malicious software performance functions in a targeted mode. under the condition that the static link libraryis insufficient, the trained machine learning model can actively distinguish the static link library function from the performance function of the malicious program, the accuracy of static link library function recognition can be improved, and the situation that an attacker maliciously constructs the malicious function consistent with the static link library function signature and cannot recognizethe malicious program is avoided.

A method, system and apparatus for source code auto-suggestion, including receiving and processing source code from a source code repository, extracting one or more features from the source code received from the source code repository, extracting one or more features from the source code within a development environment, comparing the one or more features from the source code received from the source code repository with the one or more features extracted from the source code within the development environment, and providing a segment of source code from the received and processed source code repository according to a similarity to the source code within the development environment.

The invention relates to the field of big data, and provides a federated learning method, which comprises the following steps: sending a plurality of corresponding ID intersection requests to a plurality of data providers; receiving corresponding sample data returned by each data provider according to the corresponding ID intersection request to obtain a plurality of sample data, and uploading theplurality of sample data to the block chain; judging whether each piece of sample data has a corresponding federation model or not; if the corresponding federation model does not exist in the sampledata, sending the sample data to a target federation model for training; and if the corresponding federation model exists in the sample data, sending the sample data to the corresponding federation model for training. According to the invention, the precision and the business effect of the federated learning model are effectively improved.

A method for operating a real-time time series prediction model includes obtaining a first sequence corresponding to a target time series; obtaining a second sequence corresponding to a peer time series of the target time series; determining whether a predetermined number of time steps have elapsed since a last retraining of the real-time time series prediction model; and if the predetermined number of time steps have elapsed since the last retraining of the time series prediction model, retraining the real-time time series prediction model using a first portion of the first and second sequences; and applying a remaining portion of the first and second sequences to the real-time time series prediction model to predict a next value of the target time series.

The invention discloses an active sampling method based on meta-learning, and the method comprises the steps: employing a regression model M to learn an index of a query strategy through the previously obtained experience (MetaData) of an active learning task by referring to the idea of meta-learning; wherein the input of the regression model M is based on some designed underlying features about acertain unmarked sample x and a current classification model C, the designed Meta features are irrelevant to the feature space of the data set and the form of the classification model, and the outputvalue is used for measuring the effect of x on improving the classification model C; moreover, the Meta features are independent of the feature space of the data set and the form of the classification model, so that the method has better generalization ability.

The embodiment of the invention discloses a target vision model training method and device, equipment and a storage medium. The target vision model training method comprises the following steps: acquiring a first sample image and a background image; obtaining a target image according to the first sample image, and placing the target image in the background image to obtain a second sample image; inputting each sample image into a to-be-trained model, so that the to-be-trained model predicts a corresponding prediction category according to the sample image; and determining a first loss value according to the target category and the prediction category, adjusting model parameters of the to-be-trained model according to the first loss value until a preset training stopping condition is met, and taking the trained to-be-trained model as a target visual model. According to the embodiment of the invention, the problem of low model performance caused by few sample data, difficulty in sample data acquisition and the like can be solved.