955 results about "Iterative refinement" patented technology

Methods and apparatus are provided for iterative refinement of cohorts using visual exploration and data analytics. A cohort comprised of multiple data objects is defined by obtaining an initial cohort seeding; visualizing the initial cohort using a selected view to present a current cohort; reducing the current cohort using one or more visual filters; visualizing the current cohort using a selected view; expanding the current cohort using one or more selected analytics; and determining whether the current cohort should be further modified using one or more of additional reductions and additional expansions. Cohorts can be passed between views and analytics via drag-and-drop interactions as an analysis unfolds.

A method is presented for large media data base query and media entry identification based on multi-level similarity search and reference-query entry correlation. Media content fingerprinting detects unique features and generates discriminative descriptors and signatures used to form preliminary reference data base. The preliminary reference data base is processed and a subset-set of it is selected to form a final reference data base. To identify a media query a fast similarity search is performed first on the reference database resulting in a preliminary set of likely matching videos. For each preliminary likely matching video a further multi-level correlation is performed which includes iterative refinement, sub-sequence merging, and final result classification.

A method and computer program product are disclosed for determining an optimal classifier model for a pattern recognition system and updating the determined model to recognize new output classes. An initial plurality of classifier models are selected from a set of generated classifier models. An optimal representative classifier for each classifier model is selected according to an iterative optimization routine. The representative classifier having the highest associated value for a fitness function is accepted.

The invention relates to a three-dimensional wire frame structure method and system fusing a binocular camera and IMU positioning. On the basis of binocular vision, the invention initializes and fusesinertial measurement information by using a divide-and-conquer strategy, implements tracking, positioning and drawing, and can robustly run in indoor and outdoor environments and complex motion conditions. On the basis of accurate positioning, 3D wireframe reconstruction and iterative optimization are carried out based on the posture of the key frame. Linear segments are matched by local featuresand spatial geometric constraints and back-projected into three-dimensional space. Through the angle and distance constraints, the straight line segments are divided into different sets. Based on thegrouping results, the fitting region is determined and the straight line segments are merged. Finally, a 3-D wireframe structure is output. The invention fuses multi-source information to improve thestability and robustness of the system on the traditional vision-based positioning and mapping method. At the same time, line information is added to the key frame to sparsely express the structuralcharacteristics of the three-dimensional environment, which improves the computational efficiency.

The invention relates to a method which automatically classifies motion targets learning online, models an image sequence background and detects the motion targets, scene variation, coverage viewing angle and partitioning scene, extracts and clusters characteristic vectors, and marks region classes; the number of the motion targets in a sub-region and certain threshold value initialize Gaussian distribution and prior probability to accomplish initialization of a classifier in accordance with the characteristic vectors of all the motion target regions that pass through the sub-region; the motion targets in the sub-region are classified and parameters of the classifier are online iterated and optimized; classification results in the process of tracking the motion targets are synthesized to output the classification result of the motion result that learns online. The invention is used for detection of abnormalities in monitor scenes, establishing rules for various class targets, enhancing security of monitor system, identifying objects in the monitor scenes, lessening complexity of identification algorithm, improving rate of identification, and for semantized comprehension for the monitor scenes, identifying classes of the motion target and aiding to comprehension for behavior events occurring in the scenes.

A test coverage tool provides output that identifies differences between the actual coverage provided by a test suite run on a program under test and the coverage criteria (e.g., the coverage criteria required by the test/development team management). The output from the test coverage tool is generated in the same language that was used to write the coverage criteria that are input to an automated test generator to create the test cases which form the test suite. As a result, the output from the coverage tool can be input back into the automated test generator to cause the generator to revise the test cases to correct the inadequacies. This allows iterative refinement of the test suite automatically, enabling automated test generation to be more effectively and efficiently used with more complex software and more complex test generation inputs.

In preferred embodiments, test coverage analysis results of several different test suites, some manually generated and others automatically generated, are used to produce a streamlined automatically-generated test suite and/or to add missing elements to an automatically generated test-suite.

An adaptive, interactive visual workspace for viewing groups of files based on their relationships. Relationships of files are visualized using iterative refinement of categories through a direct-manipulation graph-based layout. The visual workspace starts with a fully connected graph linking thumbnail images of related files that is then partitioned into neighborhoods in response to a user creating file stacks corresponding to different categories. Normalized spring lengths improve the overall quality of the layout. Different modes for membership in neighborhoods avoid confusing motion of files and help a user to manually organize the workspace. Additionally, retrieved files can be added without having to significantly move the previous files. Different visualization techniques indicate which files are related to each other. An elliptical-shaped window displaying rings of elliptic thumbnail images presents a group of files in an easily visible and space-efficient way. Different zoom rates are used for file location, and surrogate sizes allow users to increase the separation between files while still increasing the surrogate sizes.

The invention provides a deep transfer learning intelligent fault diagnosis method and device, a storage medium and equipment. The method comprises the steps of obtaining to-be-identified target domain data; inputting the target domain data into a fault diagnosis migration convolutional neural network model for fault judgment to obtain a health condition label; the training method of the fault diagnosis migration convolutional neural network model comprises the following steps: respectively inputting a source domain sample and a target domain sample into the fault diagnosis migration convolutional neural network model to extract features; minimizing a health state classification error on the source domain sample through a cross entropy loss function; calculating a covariance distance of the features between the source domain sample and the target domain sample through a domain adaptive module; and constraining the parameters by iteratively optimizing a loss function. According to the invention, the data distribution difference between the source domain and the target domain can be reduced, the purpose of predicting the target domain label is achieved, and the migration accuracy ofthe health condition label from the source domain to the target domain is effectively improved.

An “early exit” of an iterative refinement algorithm is implemented by effectively disabling read after write dependency stalls of newer instructions, as well as disabling the register write enable of these instructions, for the remainder of the algorithm, in addition to disabling the register write enable of these instructions. By doing so, the latency of the algorithm is reduced and the performance is increased without the complexity and potential poor performance of compare and branch instructions that might otherwise be required.

The invention relates to a safety beam forming method and device based on an IRS. The method comprises the steps of establishing an SWIPT Internet of Things system model based on IRS assistance; jointly optimizing a base station transmitting beam forming matrix, a jammer covariance matrix and an IRS phase shift, and constructing an optimization problem of maximized energy collection; and using a relaxation variable, a semi-definite relaxation method, an auxiliary variable and a sequence parameter convex approximation method for converting a non-convex quadratic problem into an equivalent convex problem, and providing an alternate iterative optimization algorithm for obtaining a feasible solution of an original problem. According to the invention, the IRS and a jammer are combined to establish a secret communication link of the SWIPT Internet of Things system based on IRS assistance, and the base station transmitting beam forming matrix, the jammer covariance matrix and the IRS phase shift are optimized in a combined manner, so that the communication security is improved, and the energy collection capability of the system is enhanced at the same time.

The present invention includes a system and a method for automatically authoring video compositions from longer units of digitized video (or a source video) by using short segments of video (or video “cliplets”). The video composition authoring process provides an aesthetically-pleasing layout of data elements to create a video composition. The data elements include multimedia elements, parameter information and description information. Any data elements that are missing but required are automatically selected by the system. The user may then review preliminary video composition results and refine the results if desired. The video composition authoring system includes an element selection and layout module for selecting the data elements in the video composition, and an iterative refinement module that allows the user to change and refine the preliminary results.

The invention relates to a 6D pose estimation method based on an instance segmentation network and iterative optimization. The 6D pose estimation method comprises the following steps: 1) converting adepth image into an HHA feature map and a scene point cloud; 2) inputting the HHA feature map and the color image into an instance segmentation network added with a spatial transformation network, andperforming instance segmentation of any pose object to obtain an object type identification result and a mask segmentation result; 3) segmenting a target point cloud on the scenic spot cloud according to an instance segmentation result; And 4) according to an improved 4PCS algorithm and an ICP algorithm, performing matching and pose finishing on the segmented target point cloud and the model point cloud of the target CAD model, thereby obtaining an accurate pose estimation result. Compared with the prior art, the method has the advantages of accurate recognition, multi-type object recognition, high detection performance, high pose matching precision and the like.

The invention discloses a processing equipment technological parameter optimization method and system. The method comprises the following steps: collecting production line data in real time and obtaining a target value of a quality index; setting an initial value of a process parameter and a preset deviation value of the quality index; utilizing a section of historical production line data which comprises a current moment and the process parameters to respectively predict and obtain a predicted value of the quality index and an uncertainty range of the quality index; determining an optimization target according to the preset deviation value and the prediction value of the quality index and the uncertainty range of the quality index, and optimizing and updating the process parameters and the preset deviation value; performing an iterative optimization process by utilizing the updated process parameters and the preset deviation value until a termination condition is met; and after iterative optimization is completed, outputting the value of the current process parameter as the optimal process parameter under the current incoming material information and the environmental parameter. According to the invention, the accuracy and stability of the process parameter optimization result can be improved.

The invention discloses a digital twinning intelligent monitoring system for an unmanned fully mechanized excavation face of a mine. The system is composed of a physical entity, a digital twinning body, an intrinsic safety type intelligent gateway, a cloud server, a cloud database, an intrinsic safety type 5G base station, a machine vision device, a communication interface, a network switch and avideo display terminal. According to the method, a digital twinning model is constructed, and perception analysis, analogue simulation, iterative optimization and decision control are carried out by utilizing a deep learning algorithm based on machine vision and edge calculation; based on data twinning and data driving, real-time monitoring, intelligent sensing, accurate positioning and health prediction of the remote physical space mine unmanned fully mechanized excavation face are achieved through the virtual space digital twinning unmanned fully mechanized excavation face. According to theinvention, the problems of non-uniform communication interfaces, weak equipment integration capability, poor video monitoring real-time performance, fuzzy video images and the like of the existing monitoring system are solved, and the real-time performance, accuracy and robustness of the intelligent monitoring system for the unmanned fully mechanized excavation face of the mine are improved.

The invention relates to a multi-process planning comprehensive evaluation system and method based on digital twinning and deep learning, and belongs to the field of process planning virtual simulation. The system comprises a physical equipment layer, a data sensing layer, an information processing layer, a virtual space layer and a digital twinning layer, the data sensing layer collects related real-time data and historical data of the physical equipment layer in the machining process and sends the data to the information processing layer for data fusion analysis and processing; meanwhile, the acquired data is transmitted to a virtual space layer, and a dynamic virtual model corresponding to the physical entity is constructed under the guidance of requirements of related models in a digital twinning layer; the digital twinning layer is used for leading the other layers together, carrying out virtual-real interaction feedback, fusion analysis and iterative optimization, and carrying out comprehensive evaluation analysis on different process schemes of the to-be-processed part, so that process parameter optimization and process route improvement are realized. According to the method, the flexibility and dynamic adaptability of process planning are improved, the resource utilization rate can be improved, the processing period is shortened, and the production cost is reduced.

A prediction processing system, method, and method for doing business is disclosed. The prediction processing system can collect, process and publish event-outcome information. The prediction processing system can dynamically filter participants into groupings and iteratively optimize odds calculations over time. A proposal framework consisting of a various abstract proposal types can model and settle propositions. Game propositions can be automatically generated based on event categorization relationships. The prediction processing system can provide game players or others with access to collective intelligence, including prediction information and information derived from prediction information. The prediction processing system can create groupings of better-performing predictors and provide them with additional information not generally available. Better-performing predictors can be provided with additional stakes to increase the weight of their predictions in odds calculations. The prediction processing system can provide access to event-outcome information on a for-fee subscription basis.

A twin neural network training method for face verification comprises: a training sample set is prepared; the images in the training sample set are normalized in size and then input to the artificialneural network for training. Artificial neural network consists of two identical sub-neural networks. The training samples are divided into datasets data_p and data_p which are equal in number. The datasets data_p and data_p are sent to two sub-neural networks to extract the eigenvectors of the samples. By comparing the loss function to realize the iterative optimization of the neural network, until the iterative number reaches the set value, then jump out of the iteration, at this time the trained artificial neural network is the twin neural network for face verification; The contrast loss function represents similarity between two sets of eigenvectors.

The invention belongs to the technical field of autonomous environment sensing of robots and relates to an automatic calibration method between three-dimensional laser and monocular vision. The method is characterized in that the automatic combined calibration between the three-dimensional laser and the monocular vision can be completed through a designed calibration device. The process of extracting laser data feature corners can be divided into two stages of detection and correction. The mapping association of original three-dimensional laser data with a distance measurement data matrix after binarization and the statistical analysis are carried out during the detection stage; and the matching through a standard template constituted by black and white corners in a calibration plate is carried out during the correction stage, thereby improving the corner extraction precision and further adopting the iterative optimization method for solution. The automatic calibration method has the beneficial effects of effectively reducing the noise and the affects of an incident angle on laser distance measurement data, and the calibration process realizes the automation and the integration of the extraction and the matching of laser-vision data feature corners, thereby laying a foundation for multi-sensor information fusion.

The invention provides a network intrusion detection method based on an improved convolutional neural network. The method comprises the steps of acquiring a data set, performing numeralization, performing normalization processing, improving convolutional neural network model training, performing iterative optimization of model parameters, acquiring five classifiers, outputting a five-dimensional confusion matrix to serve as a classification result, and evaluating the classification result; the network intrusion detection method based on the improved convolutional neural network provided by theinvention is based on an improved convolutional neural network model, model training is performed by using a preprocessed original sample data set in combination with a cross-layer design manner, themodel achieves a good convergence effect through continuous feature extraction and iterative optimization, and then a classification test is performed by using the trained classifier, so that the feasibility and effectiveness of an intrusion detection effect can be improved by the method.

The invention relates to the technical field of energy internet planning, in particular to a game theory-based regional energy internet planning mathematical model and planning method, which comprisesthe following steps of: establishing a model of a distributed power supply, an energy storage system and a combined cooling heating and power system which are suitable for game theory planning and analysis; using A public energy operator, a service provider of distributed energy and a user as game participants, planning construction capacity, using pricing strategies and energy using behaviors asrespective decision variables, using the full life cycle income of equipment as earnings, and establishing different game models for regional energy internet planning; Establishing a specific game model under multiple modes according to the decision strategy of each participant; And carrying out solving through an iterative optimization algorithm, comparing and analyzing Nash equilibrium solutions under various game modes, and finally selecting the Nash equilibrium solution under the benefit maximization of all parties as a planning scheme of the regional energy internet.

The invention provides an active learning based network data anomaly detection method, relating to an SVM (support vector machine) based detection method. The method solves the problems of more training samples, high complexity and poor classification feature of SVM in the traditional network data anomaly detection method. The method comprises the following steps: 1, choosing C points from unmarked samples as candidate cluster centers, carrying out iteration optimized clustering operation on candidate unmarked sample set A, and choosing representative samples from iteration clustering results to construct a training sample set B; 2, training an SVM on the training sample set B to obtain a training hyperplane; 3, choosing a sample which can best improve classification feature from the candidate unmarked sample set A according to sample choosing standards, making the sample with classification and then adding to the training sample set B; 4, retraining the SVM on the updated training sample set B; and 5, ending if detection precision reaches a preset value; if not, returning the step 3. The algorithm can effectively lower complexity if applied to anomaly detection research.

Aspects of the present invention include systems and methods for generating a novel view interpolation. In embodiments, feature correspondences and geometrical contexts are used to find additional correspondences based on the assumption of the local linear transformation. The accuracy and the number of correspondence matches may be improved by iterative refinement. Having obtained a set of correspondences, a novel view image can be generated.