34results about How to "Reduce search space" patented technology

The invention provides a power grid rapid scheduling command system and method based on artificial intelligence, belonging to the technical field of communications. The system comprises a processor, adisplay module, a voice acquisition module and a call control module, the voice acquisition module is used for acquiring real-time audio data of a dispatcher; the processor comprises a voice recognition module, a voice transliteration module, a keyword extraction module, a natural language processing module and a recording retrieval module. The call control module is used for initiating call control to the station according to the station recognition result and the confirmation information; the recording retrieval module is used for obtaining at least one retrieval keyword after processing bythe natural language processing module according to the received keyword instruction or voice instruction, and retrieving and extracting stored audio data and / or recording characters by utilizing theretrieval keyword. According to the invention, the functions of automatic recognition, voice transfer, intelligent retrieval and the like of the dispatcher dispatching voice instruction are realized,and the dispatcher dispatching and commanding work efficiency and the dispatcher dispatching and commanding accuracy are improved.

The invention discloses a deep learning-based user interest point recommendation method and system. The method comprises the steps of obtaining historical sign-in data of a user; training a deep learning model based on the historical sign-in data; inputting the latest sign-in data of the user into the trained deep learning model, and outputting predicted user interest points, wherein the deep learning model automatically extracts a POI category preference feature and a POI preference feature of a user, and the two features are expressed as two features Embedding; then, performing Euclidean distance calculation on the two kinds of characteristics Embedding and a candidate set POI Embedding, sorting through scores, and outputting POIs ranked as the top N, wherein N is a set value. Accordingto the invention, the situation that the recommendation accuracy can be influenced by the retrieval space with a huge POI corpus is considered, and the filtering module is linked after the preferenceencoder of the POI category, so that the POI retrieval space is reduced, the CatDM retrieval difficulty is reduced, and meanwhile, the recommendation accuracy is improved.

The invention discloses a wind tunnel test aircraft wing deformation full-field measurement method. The method comprises the following steps of making spot patterns on a surface of a wing; calibrating a stereoscopic vision measurement system to obtain an internal parameter Pa of the camera a, an internal parameter Pb of the camera b, a calibration parameter Ps of the stereoscopic vision measurement system and a transformation matrix R for converting a measurement coordinate system into an airplane model coordinate system; acquiring a test image sequence according to images shot by the camera a and the camera b before and during the blowing test; and calculating a wing deformation full-field measurement result based on the test image sequence. The method is used for solving problems of sparse measurement data, low spatial resolution and the like due to the fact that only discrete point measurement can be carried out in a wind tunnel test aircraft wing deformation measurement technology in the prior art, and dense measurement data with higher spatial resolution can be obtained; purposes of aircraft wing deformation full-field measurement and wing surface three-dimensional shape full-field measurement in the deformation state are achieved.

A pedestrian detection method based on weighted part model and selective search segmentation, which consists of a weighted part model training part and a pedestrian detection part. Model construction process: Based on the construction process of the deformable component model, different weights are set for different components according to the different occlusion conditions of pedestrians in crowded scenes, and the hidden support vector machine is used to train the weights. Pedestrian detection process: extract the HOG feature pyramid in the picture, and selectively search and segment the picture to achieve more objects with fewer effective windows. On this basis, cascade detection based on threshold clipping is used. Pedestrian detection. The method of the invention solves the problem of missed detection of pedestrians in a crowded scene from two angles, and has the advantages of reducing interference, improving the accuracy of pedestrian detection, and the like.

An automatic matching method for ICD surgery and operation codes based on deep learning, using a modular modeling method, each module only completes a relatively simple task, which greatly reduces the search space for model parameters and reduces the required amount of data. This method uses a bidirectional autoregressive language model to model the natural language sequence, uses each operation description and each ICD code to combine, calculates the semantic space weight between them, and uses the semantic space weight to reconstruct the operation description , and finally use the reconstructed surgery description to perform ICD code classification matching to solve the problem of concept splitting. In the calculation, the inherent hierarchical structure of ICD surgery and operation codes is also used for bidirectional autoregressive model modeling, incorporating business prior test knowledge. It solves the problems encountered in clinical practice, and can quickly and accurately perform ICD code matching.

The invention discloses a method and a device for inquiring a path between two figure nodes in a family tree graph. The method comprises the steps of acquiring a family tree graph corresponding to a target family tree and two nodes of a to-be-queried path,wherein the family tree graph at least comprises character nodes and attribute information, and the two nodes comprise an initial node and an end node; searching a first precursor node set corresponding to the initial node and a second precursor node set corresponding to the end node, each precursor node set comprising: at least one precursornode for indicating preset attribute information, the precursor node being used for indicating a travelable next node in the path; when the same precursor nodes exist in the first precursor node setand the second precursor node set, sequentially reading the precursor nodes in the first precursor node set and the second precursor node set to obtain a path node set; and determining the shortest path between the two nodes according to the path node set.

The invention discloses a Web service discovery method based on a probabilistic latent semantic analysis model, wherein the probabilistic latent semantic analysis model is utilized to perform modeling analysis on analyzed Web service descriptive documents, semantic concepts hidden behind service description are dug out for semantic clustering, demand service and service-centralized service are subjected to similarity matching at the relatively advanced concept hierarchy, and in combination with spectral clustering on semantic hierarchy, a service data set is subjected to irrelevant data filtration by a spectral clustering-based algorithm prior to the semantic clustering, thereby compressing the computation complexity. As proved by tests, the method is quite superior in both precision ratio and recall ratio of service discovery.

The invention discloses a method and a system for searching a diamond ring based on a sketch and a distance field. The method comprises the following steps: performing three-perspective projection on a metal ring in the diamond ring, acquiring a metal ring three-perspective projection drawing and extracting a projection image outline, thereby acquiring a three-perspective outline drawing of the metal ring; calculating a distance value at each point on the three-perspective outline drawing of the metal ring, converting the distance value into a corresponding gray value, collecting the gray value at each point of the three-perspective outline drawing of the metal ring and forming a three-perspective distance field grey-scale map of the metal ring and then establishing a three-perspective distance field grey-scale map database of the metal ring; acquiring a three-perspective outline sketch of the metal ring inputted by the user, searching for the three-perspective distance field grey-scale map with the highest similarity with the three-perspective outline sketch of the metal ring inputted by the user from the three-perspective distance field grey-scale map database of the metal ring and then outputting as a target metal ring image; and performing vertical view projection on diamond in the diamond ring and searching for a target diamond image according to a metal ring method and then outputting.

The invention provides a picture retrieval method, device and electronic equipment, which relate to the technical field of retrieval. If a picture to be retrieved is received, the characteristic information of the picture to be retrieved is extracted; according to the characteristic information of the picture to be retrieved, a Determine the picture information set matching the feature information in the information base; according to the feature information of the picture to be retrieved, from the determined picture information set, filter the similarity with the picture to be retrieved to meet the preset second similarity requirement The image ID of the picture; the retrieval result of the image to be retrieved is obtained through the screened out image ID. The embodiment of the present invention divides the picture identifiers of similar pictures into the same picture information set by pre-constructing the picture information database. The retrieval space is reduced, thereby shortening the retrieval time and improving the quality of retrieval services.

A stamping part instance retrieval method, in order to make full use of the successful experience of the stamping process, from the perspective of process similarity, the nearest neighbor method is used to retrieve the stamping part instance. Specifically, first extract features and constraints between features from the 3D model of the stamping part, and obtain the material thickness, strength and hardness of the material according to the stamping part material information table; then calculate the similarity between the old and new stamping parts from the perspective of process similarity ; The calculation of the similarity of the old and new stamping instances is carried out in two stages: the first stage calculates the similarity of the topological relationship of the stampings based on the plane, filters out dissimilar old instances, and obtains similar old instances to narrow down The retrieval space of the subsequent stage; the second stage obtains a most similar old instance from a batch of old instances obtained in the first stage by comparing the comprehensive similarity of features, constraints and materials between the old and new instances. This method has the advantages of easy implementation, less resource consumption, high computational efficiency, and more accurate and reasonable.

The invention provides an implementation method of Web service discovery based on neural network clustering optimization, and the method provided by the invention is used for rapidly and accurately discovering Web service satisfying user requirements under an open network environment. In the method provided by the invention, a service request subsystem, a service providing subsystem, a service registration center and a service discovery intermediary subsystem are utilized, wherein the service discovery intermediary subsystem is the core component of the system and comprises a service showing module, a service classifying module and a service matching module. According to the invention, the semanteme Web service discovery primary system of clustering optimization is realized; the evaluating result on the true Web service data set shows that the implementation method ensures relatively higher service discovery efficiency and precision, and the method is easy to implement; the system is flexible, and the cost is low; and the method is easy to popularize.