43 results about "Feature dependency" patented technology

A computer-implemented method for detecting a set of inconsistent data records in a database including multiple records, comprises selecting a data quality rule representing a functional dependency for the database, transforming the data quality rule into at least one rule vector with hashed components, selecting a set of attributes of the database, transforming at least one record of the database selected on the basis of the selected attributes into a record vector with hashed components, computing a dot product of the rule and record vectors to generate a measure representing violation of the data quality rule by the record.

A convenient and natural way to draw polyline shapes on a stylus-based computing device is described. A tool or other drawing function relies on a timing mechanism to identify the various vertices in a polyline shape. For example, the endpoint of each line may be detected responsive to the user stopping (or substantially stopping, such as by remaining within a predefined maximum movement threshold) movement of the tip of the stylus relative to the touch-sensitive surface. In addition, feedback is provided indicating that a line and / or its endpoint is about to commit to the polyline shape being drawn. Such feedback may be displayed shortly before the line and / or endpoint actually commits. Upon further delay without substantial movement of the stylus, the line and / or endpoint may then actually commit. A “snap” function is also provided that assists the user in fully closing the polyline shape, if desired.

Metrology methods and respective software and module are provided, which identify and remove measurement inaccuracy which results from process variation leading to target asymmetries. The methods comprise identifying an inaccuracy contribution of process variation source(s) to a measured scatterometry signal (e.g., overlay) by measuring the signal across a range of measurement parameter(s) (e.g., wavelength, angle) and targets, and extracting a measurement variability over the range which is indicative of the inaccuracy contribution. The method may further assume certain functional dependencies of the resulting inaccuracy on the target asymmetry, estimate relative donations of different process variation sources and apply external calibration to further enhance the measurement accuracy.

An incremental greedy method to feature selection is described. This method results in a final classifier that performs optimally and depends on only a few features. Generally, a small number of features is desired because it is often the case that the complexity of a classification method depends on the number of features. It is very well known that a large number of features may lead to overfitting on the training set, which then leads to a poor generalization performance in new and unseen data. The incremental greedy method is based on feature selection of a limited subset of features from the feature space. By providing low feature dependency, the incremental greedy method 100 requires fewer computations as compared to a feature extraction approach, such as principal component analysis.

Systems and method are provided for analyzing target, process and metrology configuration sensitivities to a wide range of parameters, according to external requirements or inner development and verification needs. Systems comprise the following elements. An input module is arranged to receive parameters relating to targets, target metrology conditions and production processes, to generate target data. A metrology simulation unit is arranged to simulate metrology measurements of targets from the target data and to generate multiple metrics that quantify the simulated target measurements. A sensitivity analysis module is arranged to derive functional dependencies of the metrics on the parameters and to define required uncertainties of the parameters with respect to the derived functional dependencies. Finally, a target optimization module is arranged to rank targets and target metrology conditions with respect to the simulated target measurements.

Reusable JSP code snippets are defined and instrumented for Website testing and production. A user may define tile definitions in a markup language file for a number of JSP code snippets to be included in a store page. Upon receiving store inputs comprising tile definitions and jar files for the features, tables may be generated that map features to stores and features dependencies to features. Based on the tables, the store inputs may be validated. If no error occurs, run time files may be generated and JSP code snippets may be instrumented to the appropriate location for Website testing and production.

The invention belongs to the field of computer vision, and particularly relates to a method for detecting shielding of a road region monitoring video by leaves. Firstly, a Mask R-CNN-based deep learning platform is established for training a network model of a road region and a tree leaf target, and then a detected road region result is optimized to conform to recognition expectation in an urban area video monitoring scene; and secondly, by displaying and understanding a position relation, on a two-dimensional plane, of a three-dimensional space in video monitoring, a judgment result of whether the road region is shielded by the leaves or not is finally obtained according to shielding rules. The method can replace a traditional manual mode of checking a large number of browsing videos oneby one to confirm the shielding by the leaves; the feature dependency with subjective empirical disagreements based on tree leaf area sizes, distribution areas, distances and the like is got rid of; and the shielding perception application of the three-dimensional space is realized by simulation in a two-dimensional space.

The invention provides a method and a device for determining a regression testing range. The method comprises the steps of S1, determining function units in to-be-tested software in advance and establishing a function dependence tree according to a dependence relationship between the determined function units; and S2, searching function units depending on a modified function unit by beginning from the modified function unit in the function dependence tree, searching function units depending on each searched function unit when a regression test is carried out, and using the modified function unit and all searched function units as the regression testing range of the to-be-tested software. According to the method and the device, which are provided by the invention, the accuracy of the regression test can be ensured while the testing resource and the testing time are saved.

A recommender system for use on a mobile platform or other computing device provides a user with a recommended list of products with default product quantities that are based on prior purchase behavior of the user or customer. This feature relies on analyzing a customer's past purchases and associating a score with each product. The scoring system indicates the statistical likelihood that the customer would want or need to purchase that item on their next order. The customer analysis takes into account, e.g., the number of times a customer has ordered a product, how frequently the product is ordered by the customer, how many days since the respective customers last order and the customers average order size in providing the recommended list of products.

The present invention is a method for determining during run-time an adaptation to a software architecture comprising a plurality of components. The method includes the steps of—evaluating constraints for the adaptation relating to a functional dependence of components of the software architecture,—determining during run-time a context for the software architecture taking into account information on the state of the architecture,—selecting among a plurality of context models a context model corresponding to the determined context,—evaluating constraints for the adaptation imposed by the selected context model,—taking a decision on performing the adaptation based on the outcome of the evaluation steps.

Techniques and mechanisms described herein facilitate adaptive fuzzy fallback stratified sampling. According to various embodiments, an actual or estimated minimum vertex cover of a feature dependency graph representing a dataset may be determined. The dataset may include a plurality of feature vectors and a plurality of features. Each feature vector may include a plurality of feature values that correspond with the features. The feature dependency graph may represent a plurality of conditional dependencies between the features. The minimum vertex cover may designate a subset of the features for strata selection. The feature vectors may be partitioned into a plurality of strata based on the designated subset of features. Each stratum may include one or more of the feature vectors. Each feature vector may be assigned to a corresponding stratum based on the values of the designated subset of features for the feature vector.

The embodiment of the invention provides an aspect-level sentiment analysis method and device, electronic equipment and a storage medium. The method comprises the steps of obtaining a target sentence to be subjected to aspect-level sentiment analysis and aspect words in the target sentence; based on the similar structure data set, performing aggregation feature processing on the target sentence and the aspect words to obtain fusion feature information corresponding to the target sentence; performing feature conversion on the fused feature information by using a bidirectional long-short-term memory network to obtain hidden state features corresponding to the target sentence; preprocessing the target sentence and the aspect words to obtain a dependency graph corresponding to the target sentence and position coding features corresponding to the words in the target sentence; and inputting the hidden state features, the dependency graph and the position coding features into a pre-trained graph convolutional neural network to obtain an emotion analysis result corresponding to the aspect words. According to the embodiment of the invention, the accuracy of the sentiment analysis result can be improved.

The invention relates to a cloud service API adaptation method based on execution path similarity. The problem that multi-cloud deployment needs to face diversity and complexity of different cloud computing platforms is solved. According to the method, the function dependency graph is constructed by using the execution track with the function as the unit, the function dependency graph is analyzedthrough methods such as formula calculation and the maximum public subgraph, the similarity between codes of different cloud service providers is obtained, and therefore the application range of a management mode is defined. Meanwhile, an execution track graph is constructed by using the execution track taking the module as a unit, and a standardized execution track is analyzed based on a rule, sothat a multi-cloud management mode is constructed.

The invention discloses a document-level sentiment classification method based on an attention combination neural network. The method is divided into two stages to obtain the characteristics of the document-level comments for sentiment classification: in the first stage, a two-layer bidirectional gating recurrent neural network is used for obtaining document matrix representation with compositionsemantics, and a double-layer attention mechanism is used for distinguishing the importance of different words and sentences; in the second stage, a two-dimensional convolutional neural network is used to sample more significant feature dependencies in the matrix representation obtained by the first stage and generate a high-dimensional representation for emotion classification, and in the secondstage, a convolutional attention module is used to focus on important features and suppress unnecessary features. According to the method, the problems that long text modeling is difficult to performand dependence among features is difficult to capture in an existing sentiment classification method are solved, and important information can appear in any position and composition part of a documentin a long text.

Reusable JSP code snippets are defined and instrumented for Website testing and production. A user may define tile definitions in a markup language file for a number of JSP code snippets to be included in a store page. Upon receiving store inputs comprising tile definitions and jar files for the features, tables may be generated that map features to stores and features dependencies to features. Based on the tables, the store inputs may be validated. If no error occurs, run time files may be generated and JSP code snippets may be instrumented to the appropriate location for Website testing and production.

A method and a controller for controlling the execution of a computer program having multitasking capability on a computing element, especially on a microprocessor, of a controller (20) for controlling and / or regulating a system. The system may take up various possible system states. At the beginning of the execution of a function, in order that the input variables required for carrying out the function of computer program are present with certainty, the following steps may be executed: subdividing the computer program into a plurality of functionally linked functionalities; defining possible operating states for the functionalities; defining the possible system states of the system, in that specifiable operating states are allocated to the functionalities for each system state; ascertaining the dependencies of the functionalities upon one another, a first functionality being dependent upon a second functionality if at least one input variable of the first functionality is ascertained in the second functionality; and centrally specifying the operating states required for a certain system state, taking into consideration the ascertained dependencies among the functionalities and further boundary conditions.

In business systems, one or more methods can be used to reduce an amount of redundant data. In one implementation, a method to reduce redundancy within a data model in a database, in which the data model is represented by at least one table, includes determining a number of distinct values of partial keys in a table. Each partial key represents at least one row in the table. The method includes reordering one or more columns of the table by cardinality of partial keys, in which the cardinality of a partial key represents a number of distinct values of the partial key. The method further includes determining whether pairs of partial keys are functionally dependent and eliminating one or more columns having functional dependencies from the table.

Techniques and mechanisms described herein facilitate adaptive fuzzy fallback stratified sampling. According to various embodiments, an actual or estimated minimum vertex cover of a feature dependency graph representing a dataset may be determined. The dataset may include a plurality of feature vectors and a plurality of features. Each feature vector may include a plurality of feature values that correspond with the features. The feature dependency graph may represent a plurality of conditional dependencies between the features. The minimum vertex cover may designate a subset of the features for strata selection. The feature vectors may be partitioned into a plurality of strata based on the designated subset of features. Each stratum may include one or more of the feature vectors. Each feature vector may be assigned to a corresponding stratum based on the values of the designated subset of features for the feature vector.

The disclosed embodiments provide a system for processing data. During operation, the system obtains a feature dependency graph of features for a machine learning model and an operator dependency graph comprising operators to be applied to the features. Next, the system generates feature values of the features according to an evaluation order associated with the operator dependency graph and feature dependencies from the feature dependency graph. During evaluation of an operator in the evaluation order, the system updates a list of calculated features with one or more features that have been calculated for use with the operator. During evaluation of a subsequent operator in the evaluation order, the system uses the list of calculated features to omit recalculation of the feature(s) for use with the subsequent operator.

A method, system and computer-readable medium for an enhanced dependency resolver are disclosed. The dependency resolver collects dependency information for features of a rich client platform from a plurality of update sites; stores the dependency information for the features in a database; constructs a plurality of dependency trees using the dependency information of the database, wherein each of the plurality of dependency trees is rooted by a node representing an installable feature; selects an installable feature; identifies all nodes of a dependency tree rooted by the installable feature; determines which of the nodes require an adjacent node. If the node requires an adjacent node, the dependency resolver climbs the include tree of the adjacent node until a node of a second installable feature is found and records in the database that the second installable feature is required for the installation of the other installable feature.

The invention relates to the field of financial technology, and discloses a feature processing method and device suitable for machine learning. The method comprises the following steps: after a feature processing request is obtained, constructing a feature pool according to each feature in each feature table, wherein the feature table is at least composed of a feature list, a feature library, a dependency feature table, a service and feature processing logic, the feature list comprises at least one feature, the dependency feature table is used for recording other feature tables having dependency relationships with the feature tables, and the feature processing request comprises features needing to be processed; determining a feature dependency relationship according to the to-be-processed feature and each feature in the feature pool, and determining a feature processing path according to the feature dependency relationship; and performing feature processing according to the feature processing path. The features in the database are stored in the form of a feature table, the feature table comprises a plurality of features and processing logics of the features, and feature processing is facilitated.

The invention discloses a screen drawing software. In the present invention, a computer is caused to function as a wizard controller (111a) and a highlight movement controller (111b). The wizard controller (111a) stores function-dependency-relation data (122) representing a sequence for setting up a plurality of sub functions included in the function of a programmable display (200) into a storage (12); stores setting-item dependency-relation data (123) representing a sequence for setting up each setting item on a sub-function setting screen into the storage (12); displays a list of a plurality of sub functions on a display (14) each sub function bearing a sequential number based on the function-dependency-relation data (122); and displays the setting screen for a sub function selected from the displayed list of sub functions on the display (14). The highlight movement controller (111b) highlights each setting item on the sub function setting screen sequentially on the basis of the setting-item dependency-relation data (123).

The invention discloses a defect prediction method for a numerical control system software module, which comprises the following steps of: 1) preprocessing a training data set, and determining parameters in a classifier; 2) performing data processing on the training data set, and dividing test data and training data; 3) establishing an improved naive Bayesian model based on a feature dependency relationship, processing features of the training data set in a pairwise manner, and performing model learning; 4) preprocessing the to-be-tested data by using the parameters in the classifier determined in the step 1); and 5) performing defect prediction on to-be-tested data by using the trained improved naive Bayesian model based on the feature dependency relationship. The dependency relationship is quantized and then added into the naive Bayesian classifier, so that the improved naive Bayesian classifier can be effectively applied to the numerical control system, and the purpose of accurately and efficiently detecting whether defects exist in the software module of the numerical control system can be achieved.

A method for the computerized design of real-world object uses knowledge of already existing designs and their physical performance numbers. During the design process, a plurality of design variables are adapted, and the result of the design has improved physical performance numbers when converted into the real-world Object. The design process is decomposed in parallel streams of optimization modules, with each optimization module individually optimizing a group of interrelated design variables. The functional interrelation between design variables is evaluated using an interaction information representing the functional dependency between the design variables and the physical performance numbers.

The invention provides a pedestrian re-identification method integrated with position awareness attention, which comprises the following steps: introducing a position awareness attention module into a ResNet50 network, the module is an effective improvement of a non-local attention module, and embedding position information into the non-local attention module which captures a long-range feature dependency relationship, so as to identify pedestrian re-identification. And the expression ability of the extracted features is effectively improved. The position awareness attention module provided by the invention belongs to a lightweight structure, and the module is integrated into the ResNet50 network, so that distinguishable features of pedestrians can be effectively extracted, and features with small association degree with a pedestrian recognition task are inhibited at the same time; compared with a traditional network model and other related methods, a better recognition effect is achieved on a plurality of popular pedestrian re-recognition standard data sets.

The present invention relates to a method and apparatus for accomplishing a requested operation selectively based upon features of other equipment between equippments connected through a digital interface such as the IEEE 1394. The feature-dependent operating apparatus consists of a connector sending and receiving data through the digital interface; a memory storing feature information of a connected equipment received through the connector; a pickup detecting video and / or audio data from a data storage medium; a controller controlling the pickup to detect data from the data storage medium based on the stored feature information, converting the format of the detected data into a format specified in the stored feature information, and generating an advisory message informing a mismatch between the format of data to offer and the format specified in the feature information; and a data sender transmitting data detected by the pickup or the generated message to the connected equipment through the digital interface. This feature-dependent operating apparatus is able to adjust functions and processing capacity of a equipment to the maximal or universal capacity of the connected equipment automatically, or to inform a viewer of the cause of possible malfunction if the features of the connected equipment are not same, when data by a requested operation are to be transmitted to the connected equipment, thereby improving the convenience of a viewer.

The invention provides a risk event prediction method, a risk event prediction system, a risk event prediction system generation method, an electronic device, a computer readable storage medium and acomputer program product. The risk event prediction system generation method comprises: a data preprocessing step of receiving multiple event data associated with a risk event in a predetermined timeperiod, and performing preprocessing on the multiple event data to generate to-be-processed data; a feature data selection step of selecting data, the correlation between which and the risk event satisfies a predetermined condition, from the to-be-processed data as feature data; and a prediction system training step of training and obtaining a prediction system by using the feature data as training data. The generated prediction system has the advantages of high accuracy, strong robustness and low dependence on characteristics, so that the prediction of the first occurrence of the risk event can be realized under the condition of lack of enough past risk event history.

The invention belongs to the field of computer vision, and in particular relates to a method for detecting that a monitoring video of a road area is blocked by leaves. Firstly, a deep learning platform based on Mask R-CNN is established to train the network model of road area and leaf target, and then the detected road area results are optimized to meet the recognition expectations in urban video surveillance scenes; The display and understanding of the positional relationship of the three-dimensional space on the two-dimensional plane in monitoring, the defined occlusion rules, and finally the judgment result of whether there is a road area blocked by leaves. The present invention can replace the traditional manual method of browsing a large number of videos to check and confirm the occlusion of leaves one by one, get rid of the characteristic dependence based on subjective experience disputes such as the size of the leaf area, distribution area, distance, etc., and realize the three-dimensional space by simulating in two-dimensional space Occlusion-aware applications.

The embodiment of the invention discloses a business object labeling method, device, server and storage medium. The method includes: extracting text information from the business object, and writing it into a text collection; extracting candidate phrases from the text information, and candidate words in the candidate phrases have feature dependencies; within the scope of the text collection , respectively calculating the correlation between the candidate words for the candidate phrases; selecting the target candidate phrases from the candidate phrases based on the correlation; the candidate words in the target candidate phrases are marked as the Label information of the business object. There is no need for the user to manually select a suitable word as the label of the business object, which reduces the workload and improves the efficiency.

The invention discloses a failure recovery method based on precursor dependence to mainly solve the problems that in an existing failure recovery technology, a failure point cannot be accurately positioned, and multi-system repeated backup atom functions are arranged with the dependence of a single detection point. According to the technical scheme, a functional dependence tree is structured for an application case needing failure data recovery; secondly, backup of the atom function of the functional dependence tree is built; thirdly, recovery processing is carried out when a root node of the functional dependence tree breaks down, an external storage device is searched for the latest full backup, the failure atom function is recovered according to the full backup, all increment backup after the latest full backup are searched for, increment recovery is carried out on the failure atom function one by one according to the sequence, and finally the root node is recovered. By means of the failure recovery method, function detailing can be carried out on an application, and targeted recovery can be carried out when a failure occurs; furthermore, a double-layer backup mode is used, the backup efficiency is improved, and the method can be used for failure data recovery of the application case.