14 results about "Labelling algorithm" patented technology

The invention provides an RDMA (Remote Direct Memory Access) out-of-order receiving and selective retransmission method, which comprises the following steps of: recording an arrival state of a data packet by adopting a Bitmap algorithm and / or a segmentation marking algorithm of a sliding window, dynamically generating ACK (Acknowledgement Character) or NAK (Negative Attached Keying) feedback information according to the algorithm, and only retransmitting a lost data packet according to the received NAK feedback information. The RDMA out-of-order receiving and selective retransmission method supports Bitmap statistics of a sliding window or a segment marking method of saving a memory, is compatible with an existing protocol and realizes an ACK mechanism of selective retransmission, and avoids the problem of redundant transmission in a traditional mechanism. The invention further discloses computer equipment, a data transmission system and a computer readable storage medium.

A computer-implemented method for a digital security system receives unlabeled event data associated with a computing environment, clusters via an unsupervised machine learning model the unlabeled event data into clusters of unlabeled event data where unlabeled event data in one cluster are more similar to each other than to unlabeled event data in other clusters, selects a respective subset of unlabeled event data for each cluster of unlabeled event data, translates via a large language model artificial neural network each unlabeled event datum in each respective subset of unlabeled event data into a description for the unlabeled event datum, and applies a label via a labeling algorithm to at least one unlabeled event datum in a respective cluster responsive to and representative of the respective description for the unlabeled event datum in the respective subset, thereby transforming the at least one unlabeled event datum to a labeled event datum.

The embodiment of the invention provides an industrial monitoring method and device, electronic equipment and a medium. According to the embodiment of the invention, the data model and the algorithm program of the industrial project are obtained, and the algorithm program is analyzed to generate the algorithm flow chart, so that the monitoring parameters in the data model and the program variables in the algorithm program are accurately bound based on the variable binding address. In the execution process of the industrial project, the real-time data of the monitoring parameters are dynamically mapped to the corresponding program variables in the algorithm flow chart, and the corresponding target flow chart nodes in the algorithm flow chart can be determined and marked according to the actual execution state of the industrial project, so that the operation and maintenance threshold is reduced, the automation level of industrial monitoring is improved, and the industrial monitoring efficiency is improved. Problems in the production process can be found and solved in time, and finally the stability and efficiency of industrial production are improved.

The application discloses a kind of based on single RGB image and realizes the fusion method of different human reconstruction model, method includes: first, the implicit expression network obtained by two different reconstruction methods is meshed by Marching Cubs algorithm, obtains the three-dimensional human reconstruction model of space and space resolution alignment, the pixel alignment depth map of each reconstruction model is obtained by rendering mode, and the thickness of space z is calculated, and the same area and different area of two models are obtained by thickness map comparison;Using the RGB image and depth image rendered by human model dataset, the depth map network of visible face and invisible face of person is trained, the predicted visible face and invisible face depth map are obtained by using RGB image input network, the z space position is determined and the z space thickness of different models is aligned using thickness scaling algorithm;Finally, the different areas of two models and the same area are interpolated and fused at boundary using edge distance marking algorithm.The application improves reconstruction accuracy.

The application discloses a large-scale multi-label text classification method based on label adaptive text representation. The application firstly explores label adaptive representation of the text to effectively process classification performance of head labels and tail labels under large-scale multi-label classification; a pre-trained language model is used to learn a representation pool for the text, so that different labels can focus on different representations to complete correlation discrimination. Considering the characteristics of deep models and long texts, text representation enhancement is proposed to ensure the difference and comprehensiveness of the representations in the pool. Therefore, the application can provide effective discriminative text features for large-scale labels to improve the prediction performance. Compared with current large-scale multi-label algorithms, the application can guarantee the overall classification performance of large-scale multi-labels on the one hand, and guarantee that tail labels can better focus on detailed text features on the other hand, and the performance is better than that of the current most advanced large-scale multi-label algorithm.

The present application relates to a skeleton line intersection region redundancy repairing method based on distance constraint and center replacement, and belongs to the field of image processing.The repairing method comprises: for each foreground pixel point in the skeleton image, extracting its 3*3 neighborhood and temporarily setting the center pixel as a background value, and detecting the intersection point through an 8-neighborhood connected marking algorithm; based on the spatial proximity relationship of the intersection point, realizing the intersection point clustering through the Euclidean distance calculation and the breadth-first search algorithm; performing group screening, ignoring the group containing only a single intersection point, and performing topological reconstruction for the group containing multiple intersection points.The present application accurately identifies the redundant pixel cluster, improves the skeleton line quality, accurately identifies and removes the redundant pixels in the diagonal region through the template matching algorithm, ensures that the skeleton line strictly meets the single-pixel width requirement, and avoids the false end point and the line length measurement error.

The application relates to a knowledge representation learning method based on group detection, wherein the method comprises the following steps: determining a target knowledge node and associated knowledge nodes, constructing a knowledge graph based on the target knowledge node and the associated knowledge nodes, processing the knowledge graph according to a preset rule to obtain group detection information corresponding to the target knowledge node, adding different negative sample labels to the associated knowledge nodes based on the group detection information, and applying different penalty weight coefficients to the associated knowledge nodes according to the negative sample labels in the process of knowledge representation learning of an algorithm model, and generating a knowledge representation vector. Through the application, the problem that the distinguishing degree of different entity vectors of the same type is low in knowledge representation learning is solved, and the distinguishing degree of the knowledge representation vector is improved.

The invention discloses a multi-target detection method based on a millimeter-wave radar, and the method comprises the steps: collecting the echo data of multiple targets in a complex scene through the millimeter-wave radar, and carrying out the preprocessing of the echo data, and obtaining the distance-Doppler two-dimensional data to be processed; based on the obtained two-dimensional data to be processed, adopting a CFAR detection algorithm to obtain a coarse estimation CFAR mark detection result; the maximum module value square of each column of pulses on the two-dimensional data to be processed is stored, and a false target interference mark is obtained through dynamic adaptive threshold estimation; and based on the obtained CFAR mark and the false target interference mark, accurate detection of the target is realized through a comprehensive mark. According to the method, a CFAR-based comprehensive marking algorithm is adopted, target detection is realized by dynamically estimating background noise and a self-adaptive threshold value, multi-dimensional information such as distance, speed and time-frequency characteristics is fused, a real target and clutter are effectively distinguished, and the target can be accurately identified and related information of the target can be extracted in a complex scene.

The invention discloses a shell type transformer fault diagnosis method under weak turn-to-turn short circuit and a related device, and relates to the technical field of transformer fault diagnosis, and the method comprises the steps: collecting operation parameters of a shell type transformer, and carrying out the preprocessing of the operation parameters; performing multi-scale decomposition on the preprocessed operation parameters by adopting wavelet packet decomposition to generate a two-dimensional gray feature map and generate a sample set; carrying out model training and adaptive optimization on the improved connected domain labeling algorithm to obtain an optimized improved connected domain labeling algorithm; identifying the real-time two-dimensional gray feature map to obtain an abnormal connected domain in the two-dimensional gray feature map; and extracting characteristic parameters of the abnormal connected domain, matching the characteristic parameters with a preset fault sample database to determine a fault type and a fault level, and carrying out spatial positioning. According to the invention, the problem that the fault of the shell type transformer under weak turn-to-turn short circuit cannot be accurately diagnosed due to inaccurate fault feature extraction in a complex scene in the prior art is solved.

The application discloses a confidence-driven pseudo-label generation method for noise labels, and belongs to the fields of deep learning and image classification; specifically, first, original labels are collected as training data, a part of which is labeled to obtain labels q containing noise labels; another part of training samples x is input into a feature extractor for feature extraction, and then is respectively input into a classifier and a linear mapping module of a deep neural network model, the classifier outputs a distribution p, and the linear mapping module outputs a confidence value conf between 0 and 1; then, pseudo labels are constructed and loss training is performed, and a loss L CDPL is obtained, gradients are returned to the feature extractor and the classifier; finally, the confidence conf is trained as a pseudo label allocation ratio, and gradients are returned to a mapping function h conf of the linear mapping module, so that a more reasonable and balanced pseudo label information allocation ratio can be constructed at each stage. The application solves the problem of unbalanced information allocation in the existing pseudo label algorithm, and greatly reduces the influence of noise.

The invention discloses an automatic data classification and grading method and system based on a label algorithm, and relates to the technical field of data processing, and the method comprises the steps: obtaining original data, carrying out the preprocessing, and forming a standard data format; performing feature extraction on the standard data, synchronously generating a metadata tag, obtaining a service context tag and a descriptive tag, and forming a multi-dimensional attribute tag set through tag fusion; constructing a classification model and a grading model based on the standard data and the corresponding label set; correspondingly training a classification model and a grading model based on a pre-collected labeled label set and historical standard data corresponding to the category result and the grade result; after obtaining new data and obtaining a corresponding label set, calling the trained classification model and grading model, and outputting an automatic classification result and a grading result; and storing data generated in the process. According to the invention, automatic classification and grading of data can be realized.

The present application belongs to the technical field of anti-seepage structure evaluation, and particularly relates to an anti-seepage structure integrity evaluation method and system based on resistivity imaging, which comprises the following steps: firstly, collecting complex admittance data of monitoring points and extracting phase loss angle features to strip the interference of underground environment humidity fluctuation through phase information; then, constructing virtual part admittance gradient features in combination with structure anisotropy correction factors to enhance the recognition degree of small damage edges; calculating seepage probability weight by using reference gradient deviation, and identifying seepage clusters based on adaptive seed point screening and eight-neighbor connected domain marking algorithm to obtain continuous seepage area; finally, calculating integrity score through a nonlinear model by introducing risk sensitivity index and area penalty weight to realize accurate early warning and positioning. The present application effectively solves the problem of poor seepage recognition accuracy under complex background interference, and significantly improves the robustness and scientificity of the evaluation result.

The application discloses a lotus phenotype identification method and device based on a pseudo-label algorithm and a MobileNetV2 network, and the method comprises the following steps: step one, a grid model is constructed, the model selects the MobileNetV2 network as a feature extraction network for lotus identification, applies an SE attention mechanism to a feature processing unit of the MobileNetV2, and simultaneously uses a pseudo-label algorithm to perform pseudo-labeling on unlabeled lotus data; step two, a model is trained, a training set in a lotus data set is used to pre-train a model to initialize the MobileNetV2 feature extraction network, then the model obtained through pre-training is used to predict the unlabeled data, the minimum entropy, i.e. the highest confidence, is selected to perform pseudo-labeling on the lotus data, and finally all the labeled data is retrained to obtain an optimal model; and step three, the model after training is used to identify lotus phenotypes. The application can improve the expression and generalization capabilities of the model, reduce the labeling amount of a large data set, and be more suitable for various unbalanced and complex data distributions.

The invention provides an intent-driven marketing activity automatic generation method and system, and the method comprises the steps: obtaining intra-bank business data and a tag library, carrying out the processing of customer data through a big data portrait and a tag algorithm model, and generating and outputting enhanced customer group data and a customer 360 portrait; based on the generated customer group data and customer 360 portraits, classifying and layering customers, establishing and managing a high-quality marketing material library through a management background, tracking and analyzing use data of materials, and outputting user demands and behavior analysis results; issuing an online marketing task to the mobile terminal of the employee, automatically tracking and collecting task completion data, and generating a precision marketing strategy for different clients; meanwhile, using data of the materials are tracked and analyzed, user requirements and behavior analysis results are output, marketing strategies are changed according to behaviors of clients, market changes can be quickly responded, and marketing efficiency is improved.