129 results about "Semantic gap" patented technology

In the realm of managing relational databases, a system that uses both the data in a relational database and domain knowledge in ontologies to return semantically relevant results to a user's query. Broadly contemplated herein, in essence, is a system that bridges a semantic gap between queries users want to express and queries that can be answered by the database using domain knowledge contained in ontologies. In accordance with a preferred embodiment of the present invention, such a system extends relational databases with the ability to answer semantic queries that are represented in SPARQL, an emerging Semantic Web query language. Particularly, users may express their queries in SPARQL, based on a semantic model of the data, and they get back semantically relevant results. Also broadly contemplated herein is the definition of different categories of results that are semantically relevant to a user's query and an effective retrieval of such results.

The embodiments of the present invention provide a computer-implemented method and system for translating user keywords into semantic queries based on domain vocabulary. The system receives the user keywords and search for the concepts. The concepts are transformed into a connected graph. The user keywords are translated into precise access paths based on the information relationship described in conceptual entity relationship models and then converts these paths into logic based queries. It bridges the semantic gap between user keywords and logic based structured queries. It enables users to interact with the semantic system by articulating the information in a structured query language. It improves the relevance of search results by incorporating semantic technology to drive the mechanics of the search solution.

A system and method for efficient security protocols in a virtualized datacenter environment are contemplated. In one embodiment, a system is provided comprising a hypervisor coupled to one or more protected virtual machines (VMs) and a security VM. Within a private communication channel, a split kernel loader provides an end-to-end communication between a paravirtualized security device driver, or symbiont, and the security VM. The symbiont monitors kernel-level activities of a corresponding guest OS, and conveys kernel-level metadata to the security VM via the private communication channel. Therefore, the well-known semantic gap problem is solved. The security VM is able to read all of the memory of a protected VM, detect locations of memory compromised by a malicious rootkit, and remediate any detected problems.

Disclosed are various embodiments that facilitate automatically bridging the semantic gap in machine introspection. It may be determined that a program executed by a first virtual machine is requested to introspect a second virtual machine. A system call execution context of the program may be determined in response to determining that the program is requested to introspect the second virtual machine. Redirectable data in a memory of the second virtual machine may be identified based at least in part on the system call execution context of the program. The program may be configured to access the redirectable data. In various embodiments, the program may be able to modify the redirectable data, thereby facilitating configuration, reconfiguration, and recovery operations to be performed on the second virtual machine from within the first virtual machine.

The invention discloses an automatic question-answering method based on deep learning, and aims to provide an algorithm-based fully-automatic question-answering scheme for a user. According to the method, question and answer pairs crawled from websites are used as data sources, and questions with more complicated forms can be answered. According to the method, on the basis of traditional similar-question retrieval, a BOW model, a TFIDF model and a Word2Vec model are utilized to represent text contents of questions as vectors, similar questions are resorted and screened out through calculatingsimilarity between vectors, semantic knowledge can be introduced, the semantic gulf problem in traditional question retrieval processes can be solved, and validity of candidate answers can be improved. In addition, based on deep learning, the method utilizes a neural network model, which is obtained by training, for matching scoring on a question and the candidate answers, and can automatically extract high-layer matching features between the question and the answers, automatically give an answer of the question, improve accuracy of an automatic question-answering system, reduce manual intervention at the same time, and reduce system development costs.

The invention discloses an automatic image annotation method based on deep learning and canonical correlation analysis. The method includes: using a depth Boltzmann machine to extract the high-level feature vectors of images and annotation words, selecting multiple Bernoulli distribution to fit annotation word samples, and selecting Gaussian distribution to fit image features; performing canonical correlation analysis on the high-level features of the images and the annotation words; calculating the Mahalanobis distance between to-be-annotated images and training set images in canonical variable space, and performing weighted calculation according to the distance to obtain high-level annotation word features; generating image annotation words through mean field estimation. The depth Boltzmann machine comprises I-DBM and T-DBM which are respectively used for extracting the high-level feature vectors of the images and the annotation words. Each of the I-DBM and the T-DBM sequentially comprises a visible layer, a first hidden unit layer and a second hidden unit layer from bottom to top. By the method, the problem of 'semantic gap' during image semantic annotation can be solved effectively, and annotation accuracy is increased.

The invention discloses a retrieval system based on a multi-lesion region characteristic and oriented to a medical image database. In the invention, the image is omni-directionally described by advanced image description symbols, including image region contents, pathological representation and anatomical position information, thus realizing the quick positioning and matching method on a plurality of lesion regions; by adopting the method of combining the semantic navigation and high-dimension data index, the quick retrieval of large-scale image characteristic value can be realized, the retrieval efficiency is improved, and the 'semantic gap' phenomenon existing in the traditional image retrieval technology is relieved to certain degree. The retrieval system sufficiently utilizes the historical image and diagnosis data in a PACS (picture archiving and communication system) database, can be used as an effective measure for computer aided diagnosis, and can be widely applied to the fields such as medical clinics, researches and teaching.

The invention discloses a problem similarity calculation method based on a plurality of features, includes steps: For the input new question sentence, Compared with the stored historical questions andcorresponding answers, the similarities between the new questions and the historical questions are calculated based on character features, semantic features of words, semantic features of sentences,implied topic features of sentences and semantic features of answers. The final similarity is the product of the above five similarities and their corresponding weights, which are trained by linear regression method. The invention adopts a plurality of features to increase the diversity of sample attributes, and improves the generalization ability of the model. At that same time, the soft cosine distance is utilized to convert the TF-IDF is fused with editing distance, word semantics and other information, which overcomes the semantic gap between words and improves the accuracy of similarity calculation.

The invention discloses a transfer learning-based multi-view commodity image retrieval and identification method. The method comprises the steps of 1, establishing a multi-view image basic library according to a commodity list, performing fine adjustment on a pre-trained deep residual error network by using a small amount of commodity images through a transfer learning technology, extracting features of the image basic library by using the network, performing dimension reduction on the features, constructing a feature library, and finally according to corresponding relationships among the feature library, the image basic library and commodity types, establishing a mapping table; 2, after to-be-identified commodity images are obtained, extracting features of the images by using the networkand performing dimension reduction; and 3, performing distance measurement on the features of the to-be-identified commodity images and the features of the images in the basic library, taking the mostsimilar image with the shortest distance as a matching result, and through the mapping table, obtaining commodity type names of the to-be-identified commodity images. The features with strong representation capabilities can be automatically extracted; a semantic gap is further broken through; and the retrieval efficiency and the identification precision are improved by only utilizing a small amount of image basic libraries and low-dimensional features.

The invention discloses a cross-modal retrieval method based on multilayer semantic alignment, which comprises the following steps of: acquiring a remarkable fine-grained region by utilizing a self-attention mechanism, promoting the alignment of entities and relationships among modal data, providing an image text matching strategy based on semantic consistency, extracting semantic tags from a given text data set, and performing global semantic constraint through multi-label prediction, so that more accurate cross-modal association is obtained. Therefore, the problem of cross-modal data semantic gap is solved.

The invention discloses a dialogue short text clustering method based on form and semantic similarity. The form similarity adopts character string editing distance similarity, and the semantic similarity is based on HowNet and WordNet knowledge bases; weight values of the short text and words are introduced during the calculation of the short text similarity. The dialogue short text clustering method based on the form and semantic similarity solves the problems of certain irregular and input wrong noise information, synonyms and semantic gaps included in the dialogue short text to a certain extent, and consequently, relatively great improvement is realized in comparison with a word bag vector based clustering method.

ion discloses a cross-modal retrieval method based on a sketch retrieval three-dimensional model, comprising the following steps: 1) selecting a data set; 2) rendering the three-dimensional model datain the dataset, and obtaining a plurality of two-dimensional views from one three-dimensional model, wherein the two-dimensional views are used for representing the three-dimensional model, and the sketch data is uniformly sized into 256 plus 256; 3) training a sketch classifier and a view classify; 4) constructing the depth metric learning space: taking the sketch and view of the original dataset as the input of the network, taking the parameters of the sketch and view classifier as the parameters of the network, and achieving the retrieval goal through training. The invention is mediated byviews, minimizing the semantic gap between the 3D model and the sketch. A novel network structure of sketch retrieval 3D model is constructed, which achieves very good classification accuracy on SHREC2013 and SHREC2014. The experimental results fully show that our network framework can retrieve the corresponding 3D model from the sketch.

The invention relates to a method for optimizing a deep convolutional neural network for image classification, which comprises the steps of first, building an image classification convolutional neural network; second, training the image classification convolutional neural network; third, testing the image classification convolutional neural network. The process of testing the image classification convolutional neural network comprises the steps that a preprocessed test data set is fed into a trained network model, and an Accuracy layer of the network outputs an accuracy value according to a probability value outputted by a Softmax layer and a label value of an input layer, wherein the accuracy value is the probability that a test image is correctly classified. According to operations performed in the steps, optimization for the deep convolutional neural network for image classification can be realized. The method provided by the invention for optimizing the deep convolutional neural network for image classification can effectively reduce the semantic gap and is high in classification accuracy.

The invention discloses a method for automatically annotating remote sensing images on the basis of deep learning. The method for automatically annotating the remote sensing images includes extracting visual feature vectors of the to-be-annotated remote sensing images; inputting the visual feature vectors into a DBM (deep Boltzmann machine) model to automatically annotate the to-be-annotated remote sensing images. The DBM model implemented in the method sequentially comprises a visible layer, a first hidden layer, a second hidden layer and a tag layer from bottom to top, and is acquired by means of training. The method for automatically annotating the remote sensing images has the advantages that the deep Boltzmann machine model implemented in the method comprises the two hidden layers (namely, the first hidden layer and the second hidden layer respectively), accordingly, the problem of 'semantic gaps' in image semantic annotation procedures can be effectively solved by the two hidden layers, and the integral annotation accuracy can be improved.

The invention belongs to the technical field of image processing and particularly provides a semantic propagation and mixed multi-instance learning-based Web image retrieval method. Web image retrieval is performed by combining visual characteristics of images with text information. The method comprises the steps of representing the images as BoW models first, then clustering the images according to visual similarity and text similarity, and propagating semantic characteristics of the images into visual eigenvectors of the images through universal visual vocabularies in a text class; and in a related feedback stage, introducing a mixed multi-instance learning algorithm, thereby solving the small sample problem in an actual retrieval process. Compared with a conventional CBIR (Content Based Image Retrieval) frame, the retrieval method has the advantages that the semantic characteristics of the images are propagated to the visual characteristics by utilizing the text information of the internet images in a cross-modal mode, and semi-supervised learning is introduced in related feedback based on multi-instance learning to cope with the small sample problem, so that a semantic gap can be effectively reduced and the Web image retrieval performance can be improved.

The invention provides a semantic-based three-dimensional model retrieval system and method. The system comprises a semantic retrieval module and a semantic database. The semantic retrieval module is used to segment and filter semanteme of a retrieved sentence so as to obtain a semantic concept. The method comprises the steps of inquiring the synonymous semantic concept in the WordNet according to the semantic concept, calculating the similarity between the semantic concept and the inquired semantic concept and screening the semantic concept according to the similarity to obtain other semantic concepts associated with the semantic concept; and combining the semantic concept with other associated semantic concepts to retrieve corresponding three-dimensional information in the semantic database to obtain the retrieval result. According to the three-dimensional model retrieval system and method provided by the invention, the semantic gap of high-level semantic information and low-level semantic information is complemented, the retrieval range of the three-dimensional model is expanded, and the retrieval precision is improved.

The invention discloses an image search method based on soft constraint non-supervision cross-modality hash. The method sequentially comprises the following steps of building a picture and a dataset of text data corresponding to the picture to obtain an image feature data matrix and a text feature data matrix; configuring an image similarity matrix and a text similarity matrix, distributing hash codes according to the image similarity matrix and the text similarity matrix, and thus obtaining guiding hash codes; adopting the guiding hash codes to optimize final hash codes and a corresponding projection matrix; calculating a Hamming distance between the hash codes and hash codes of a sample in a search base, and then descendingly outputting search results according to the size order of the Hamming distance. When the image search method is utilized, quantification loss of the hash codes can be lowered, semantic gaps can be shortened, discrete solutions can be obtained, and then the accuracy and efficiency of crossed searching of pictures and texts can be improved.

The invention discloses a content-based retrieval method and a content-based retrieval system facing radiographs in a PACS database. The content-based retrieval to the radiographs is realized by associating a radiodiagnosis report in a radio information system with the radiographs in the PACS database, establishing the reverse index of the radiodiagnosis report and a feature index library of the radiographs, and then using a two-phase search algorithm. The retrieval method, by effectively integrating text retrieval technology and content-based image retrieval technology, overcomes the defect that the prior content-based image retrieval technology is not suitable for large-scale image databases, alleviates the semantic gap in the prior content-based image retrieval technology to a certain extent, and has high application value.

The invention discloses a salient object-based image retrieval method and system. The method comprises the steps of performing saliency detection on a query image containing a salient object to determine a region where the salient object of the query image is located; determining visual features of the region where the salient object of the query image is located; determining a semantic type of the salient object of the query image; performing similarity measurement on the visual features of the salient object of the query image and visual features of salient objects of images with the same semantic type in an image library, and determining the images, meeting a condition that the similarity between the images and the query image is greater than a similarity threshold, in the image library. According to the method and the system, the image retrieval is carried out through the visual features of the region where the salient object of the image is located, so that the background interference is avoided; by determining the semantic type of the salient object of the query image, the images with different semantic types in the image library are filtered, so that the semantic gap of the image retrieval is reduced, the image retrieval complexity is lowered, and the image retrieval accuracy is further improved.

The invention discloses a cross-modal retrieval method based on a graph convolutional neural network. The cross-modal retrieval method comprises four processes of network construction, data set preprocessing, network training and retrieval and precision testing. Semantic representations in an image mode and a text mode are respectively learned by using a graph convolutional neural network; the cross-modal retrieval method can help to process the potential relationship among modal features, introduces the associated data of the third modal into the cross-modal retrieval method to reduce the semantic gap among the modals, and can significantly improve the accuracy and stability of cross-modal retrieval, thereby realizing accurate cross-modal retrieval.

The invention belongs to the sub-field of the learning theory and application in the technical field of computer application, and relates to a CCA and 2PKNN based automatic image annotation method, in order to solve problems of a semantic gap, a weak annotation, and category imbalance that exist in an automatic image annotation task. The method comprises: firstly, for a semantic gap problem, mapping two features to a CCA sub-space, and solving a distance between the two features in the sub-space; for a weak annotation problem, establishing a semantic space for each annotation; for a category imbalance problem, by combining a KNN algorithm, finding out k nearest neighbors of a test image in the semantic space of each annotation, constituting the k nearest neighbors to an image sub-set, and by using a visual distance between the sub-set and the test image, and by combining a Bayesian formula, assigning a few annotations with the highest score to the test image; and finally, optimizing an image annotation result by using correlation between annotations. The method disclosed by the present invention has a greater degree of improvement for image annotation performance.

The invention is applicable to the technical field of image processing and provides a semantic mapping method of the local invariant feature of an image. The semantic mapping method comprises the following steps of step A: extracting and describing the local invariant feature of the colorful image; step B: after extracting the local invariant feature, generating a visual dictionary for the local invariant feature extracted from the colorful image on the basis of an algorithm for supervising fuzzy spectral clustering, wherein the visual dictionary comprises the attached relation of visual features and visual words; step C: carrying out semantic mapping and image description on the attached image with the local invariant feature extracted in the step A according to the visual dictionary generated in the step B. The semantic mapping method provided by the invention has the advantages that the problem of semantic gaps can be eliminated, the accuracy of image classification, image search and target recognition is improved and the development of the theory and the method of machine vision can be promoted.

The invention discloses a method for marking picture semantics based on a Gauss mixture model, which belongs to the technical field of image retrieval and automatic image marking. The method comprises the following steps: S1, obtaining a relationship between a low-level visual feature of the image and a semantics concept through monitoring Bayesian learning, and obtaining an image feature set; S2, establishing two Gauss mixture models for each semantics concept by means of an expectation-maximization algorithm, and adding a step of eliminating a noise area; and S3. according to the image feature set, calculating the color posterior probability of the pattern posterior probability of an area layer, arranging the calculated posterior probabilities which belong to all concepts of the image according to a descending order, and obtaining the color ordering value of each concept; similarly, arranging the pattern posterior probabilities and obtaining the pattern ordering value of each concept; and selecting a concept class marking image with a least summation of front R ordering values. According to the method of the invention, the difference between the low-level visual feature of the image and the high-level semantics concept expression is remarkably reduced, thereby effectively settling a semantic gap problem.

The invention provides a video monitoring system image acquisition method. The method comprises the following steps of acquiring one frame in a monitoring video, which means that an image expressed by the frame is acquired; according to a preset noise removing rule, removing noise data in the image; according to a preset target identification rule, identifying a target object in the image where the noise data is removed; adding a label to the frame, wherein the label can express a preset characteristic of the target object based on semantics; storing the label corresponding to the frame. Through mapping an extracted visual bottom layer characteristic to high-layer semantic information which is convenient for intuitive understanding of human according to a preset algorithm, based on that, classification and marking of video monitoring image data are realized and a meaning of the video monitoring image data is well expressed. ''A semantic gap'' between an image bottom layer characteristic and rich semantic contents of human is reduced and even eliminated and rapid and high-efficient acquisition of a video monitoring image is realized.

The invention discloses a CAD semantic model search method based on a design intent. The method comprises the steps of A, establishing a three-dimensional CAD model database and carrying out three-dimensional annotation of the design intent through utilization of a PMI module of UG according to modeling, analyzing and manufacturing features of each model; B, carrying out classification on the annotation information of three-dimensional models according to modeling information, analyzing information and manufacturing information, and establishing a design intent semantic tree of each model; C, establishing a field-based body semantic model tree according to a three-dimensional semantic tree database; D, establishing a search index according to the body semantic tree; E, comparing similarity of a target search word set and semantic tree nodes and returning the same or similar nodes and sub-nodes thereof; and F, calculating corresponding model semantic similarity according to the returned nodes, returning the three-dimensional models with high semantic similarity and feeding back the three-dimensional models to a user. According to the method, the semantic gap problem of the content-based search method is solved, the similarity calculation is carried out by matching target search words and semantic annotation words, and the recall ratio of search is improved.

The invention discloses a linear generalization regression model based cross-media retrieval method which includes: extracting semantic features of different model objects, establishing regression relations among models by the aid of the linear generalization regression model to realize mutual conversions of the features of models; estimating posterior probability distribution of the model objects by multi-class Logistic regression algorithm after features are converted; and measuring distance between a calculation test sample and a database sample by distance measurement, and outputting index to obtain the most similar samples of prior N databases. The linear generalization regression model based cross-media retrieval method can cross semantic gaps of different models, protect different models from revealing effective information during converting process to the greatest extent, guarantee effectiveness of information transmission of different models accordingly, improve robustness and accuracy for cross-media retrieval further, and has good application prospect and considerable marketing value.