697 results about "Synthetic data" patented technology

The invention discloses a computer memory bank fault prediction method and system based on deep learning, and relates to the technical field of computer hardware fault diagnosis, and the system comprises a multi-source time sequence data collection module which is used for obtaining memory bank operation state data in real time; the dynamic feature enhancement module is based on a composite architecture of a generative adversarial network and transfer learning, comprises a fault mode generator, and generates synthetic data consistent with real fault distribution by using an LSTM network; aligning feature spaces of different hardware platforms through a maximum mean difference loss function; the multi-modal fusion deep learning model comprises a space-time convolutional network, a graph attention network and an adaptive weight adjustment mechanism; and the fault early warning analysis module is used for analyzing a fault probability predicted value, an interpretable thermodynamic diagram and a maintenance suggestion. According to the invention, passive maintenance is changed into active prevention and control, and preposition and precision of fault management are realized through dual mechanisms of long-term trend prediction and short-term risk early warning.

Techniques for determining a presence of an object, especially an object such as animal or debris, in a path of a vehicle, are discussed herein. For example, sensors of various modalities, which may include multispectral sensors, may capture data representing an environment the vehicle is traversing. In examples, one or more trained machine learned (ML) models, operating on a vehicle computing system, may detect and / or classify objects in the environment, based on input data of one or more modalities or spectral bands. The ML models may be pre-trained using training data including real sensor data, synthetic data, and / or augmented data, along with auto-generated annotations. In some examples, hyperspectral data may be used to identify materials associated with detected objects. A confidence score associated with the detection of the object may also be computed. The vehicle may be controlled based on detection of the object and its classification.

The invention relates to the technical field of wind field simulation, and provides a wind field numerical simulation method based on multiple meteorological data sources. The objective of the invention is to solve the problems of large simulation error, rough terrain boundary processing and poor turbulence model parameter adaptability caused by non-uniform coverage of a single data source, insufficient precision and unscientific multi-source fusion. The method is characterized by comprising the following steps: step 1, constructing a CFD three-dimensional grid based on a geometric model of a target area; 2, collecting original data of a ground station, satellite remote sensing, numerical forecasting and the like; 3, performing standardized preprocessing (abnormal value elimination, missing interpolation, radiation / geometric correction and resampling), determining a multi-source fusion weight by combining historical data analysis, and generating comprehensive meteorological data by adopting a weighted average method; and 4, inputting the comprehensive data into a CFD-RANS model, dynamically adjusting turbulence parameters, accurately setting terrain boundary conditions, and obtaining a wind field space-time distribution result through numerical solution. According to the invention, through combination of multi-source data fusion and CFD simulation, the wind field simulation precision and stability are improved.

A bridge component identification method based on unmanned aerial vehicle point cloud reconstruction and three-dimensional synthetic data is characterized in that a point cloud is generated through video acquisition of an unmanned aerial vehicle, accurate identification of bridge components is realized by combining with an improved PointNet + + model, and a technical process of data acquisition-point cloud generation-model training-component identification is constructed. The method comprises the following specific implementation steps: (1) acquiring a bridge member field video by using an unmanned aerial vehicle; (2) building a three-dimensional reconstruction framework to generate dense point cloud of bridge components; and (3) generating diversified bridge point cloud synthetic data in batches. (4) constructing a point cloud semantic segmentation data set with labels, and preprocessing and enhancing the point cloud semantic segmentation data set for model training; (5) realizing point cloud identification based on an improved PointNet + + model; and (6) post-processing an identification result, and extracting geometric parameters of the component for bridge structure state evaluation. According to the invention, efficient generation and high-precision identification of the point cloud of the bridge member can be realized, and reliable data support is provided for quality evaluation of the bridge structure.

The invention discloses a drainage pipeline multi-parameter sensing trenchless repair dynamic regulation and control method and system. Through a fusion technology path of multi-source comprehensive data acquisition, edge calculation real-time cross validation of abnormity, pipeline state simulation analysis of defect levels, intelligent algorithm dynamic adjustment of repair parameters and geographic information visualization platform full-process monitoring, closed-loop management from abnormity identification to repair regulation and control is realized. A comprehensive data set is formed through multi-source data acquisition, an edge computing technology is utilized to quickly identify abnormities, a simulation technology is combined to evaluate the severity of defects, then process parameters are optimized and repaired based on defect levels, and a regulation and control instruction is generated through linkage of a visual platform and a pump station dispatching system. And finally, a pipeline repair and operation regulation and control scheme is formed through integration, and the repair effect and the system stability are ensured. According to the method, the pipeline abnormity processing accuracy and the repairing efficiency are remarkably improved, and a technical guarantee is provided for safe and stable operation of an urban drainage system.

The invention relates to a pathological section intelligent auxiliary differential diagnosis system based on multi-modal fusion, in particular to the field of clinical pathology, semantic unification of multi-modal data is achieved through meta-task construction and a cross-modal alignment technology, and transferable diagnostic knowledge is extracted by utilizing a meta-learning framework; the method combines a generative model and knowledge constraints to generate high-quality synthetic data, and finally fuses real and synthetic samples through a self-adaptive diagnosis mechanism, thereby remarkably improving the differential diagnosis capability of rare lesions, effectively solving the problem of model generalization in a training data scarcity scene, and improving the accuracy of model identification. And efficient and reliable intelligent auxiliary decision support is provided for clinical pathological diagnosis.

The invention relates to the technical field of attribute graph collection and release, in particular to an attribute graph collection and release method, system and device based on local differential privacy and a medium. Receiving disturbed data from each client; based on all disturbed data, estimating community division of the original graph structure and triangular counting of each node; obtaining a target graph structure based on community division and triangular counting; and on the basis of the target graph structure and the disturbed data, node attributes are optimized, so that node attribute distribution and community division are kept consistent, and a final publishable target attribute graph is obtained. The method has the advantages that strict requirements of local differential privacy are met on the privacy protection level, and multi-dimensional features such as a community structure, triangular counting and node attribute homogeneity are reserved in the published attribute graph at the same time; and the practical value of the synthetic data in applications such as downstream graph analysis, community discovery and personalized service is greatly improved.

Aspects of the subject disclosure may include, for example, a system including: one or more probes in a communication network; a processing system including a processor; and a memory that stores executable instructions that, when executed by the processing system, facilitate performance of operations of: collecting data from the one or more probes and application program interfaces of network elements in the communication network; converting the data into semantic vectors using an embedding model; storing the semantic vectors in a vector database; using foundation models to generate outputs based on the semantic vectors; generating synthetic data using a generative adversarial network, wherein the synthetic data is used test the foundation models; using federated reinforcement to incorporate human feedback into the semantic vectors; and managing the communication network based on the outputs. Other embodiments are disclosed.

A system and method for facilitating training of large language model based recommender systems are provided. The system may utilize one or more LLMs to create probability distributions for binary classification tasks associated with specific user-item pairs. The probabilities may be utilized to rank one or more tasks directly. The training of the one or more LLMs may involve the use of Knowledge Distillation methods and may be based on incorporating a dual-label system such as, for example, hard labels and soft labels. The one or more LLMs training data may consist of user-item pairs and their corresponding features. The labels used in the training process may include binary classification labels and their respective probabilities. The system may further implement the trained one or more LLMs to determine rankings or recommendations associated with user engagement of one or more content items.

The invention discloses an LIBS rare earth multi-element quantitative detection method based on a deep learning model, and the method comprises the steps: constructing a multi-scale feature fusion deep learning architecture MSFF-Net, carrying out the weighted fusion of local features extracted by a 1D-CNN and long-range dependence features of BiLSTM modeling through a channel attention mechanism, and enhancing weak signal features in combination with a first-order derivative spectrum. Meanwhile, a synthetic data enhancement strategy based on a physical model is adopted to simulate a matrix effect and a spectral line overlapping scene, and spectral feature migration under different matrixes is achieved in cooperation with a matrix self-adaptive migration learning strategy. And after data acquisition and preprocessing, a trace element concentration prediction value is synchronously output through a multi-task learning framework. According to the method, the detection precision and the anti-interference capability of the LIBS technology on trace elements are effectively improved, and the method has remarkable application value in the industrial fields of mineral resource exploration, strategic metal recovery and the like.

Systems and techniques are described herein for adapting a pretrained machine learning model. For instance, a process can include encoding a training image into a first feature vector, the training image including a first object located at a first location; generating a second feature vector based on a set of sinusoidal functions using a set of weights; combining the first feature vector with a second feature vector to generate a combined feature vector; processing the combined feature vector using a visual language model to obtain a second location for the first object; and adjusting the set of weights based on a comparison between the first location and the second location.

Embodiments here in disclose a method and system for network configuration generation using a network digital twin (NDT) comprising a spatio-temporal graph neural network (STGNN) for performance-based simulation in a communication network system is disclosed. A simulation request is received from a management service (MnS) consumer, including network parameters related to live network nodes. Performance data is retrieved from a management server, and performance is assessed against a performance threshold. Multi-domain telecom data is processed using a spatio-temporal graph neural network (STGNN) by a feature fusion attention mechanism, trained via a hybrid feedback loop with real and synthetic data. Upon meeting performance thresholds, a configuration recommendation is generated by a configuration recommendation engine, including an explanation and confidence value. The configuration recommendation is transmitted to the MnS consumer, enabling network automation without altering the live network directly.

Exemplary embodiments for data security include a data access proxy coupled with a database, further coupled with a server configured to operate the data access proxy to: identify a user and request to access a data item; validate the user and request, including inspecting the user's identity, evaluating the user's history, and evaluating permissions and restrictions associated with the user and the data item; access the database to retrieve the data item; inspect security attributes related to the data item; and transform the data item based on one or more privacy rules, including redacting the at least one data item, deleting information from the at least one data item, substituting information from the at least one private data item with other information, adding information to the at least one data item, providing synthetic data as a private data item, or providing proxy data for the data item.

The invention relates to the technical field of computers, and discloses a subject classification model construction method and system, electronic equipment and a storage medium, and the method comprises the steps: obtaining academic paper bibliography data based on an academic database, and constructing an initial training data set; identifying minority category subjects of which the sample quantity is lower than a preset threshold value, generating synthetic data containing chapters and keywords, and labeling corresponding subject categories; mixing the synthetic data with real data in the initial training data set, and constructing a balanced mixed training data set; splicing a text based on the chapter and the keyword of each sample in the mixed training data set, and generating a multi-level fusion feature; and taking the multi-level fusion features as input, accessing a full-connection classification layer to construct a model, carrying out end-to-end training based on a mixed training data set, and adjusting and optimizing model hyper-parameters to obtain a final subject classification model. According to the method, data imbalance can be effectively relieved, existing labeling resources are fully utilized, and the model generalization ability is improved.

The invention relates to a virtual simulation emergency drilling method and system under a mine fire dynamic catastrophe condition based on deep reinforcement learning, and belongs to the field of mine safety management. According to the method, a mine comprehensive database is constructed, environment, equipment, fire source and ventilation system data of a mine are collected, a mine structure model is generated by using a three-dimensional modeling technology, and a mine fire spreading process is simulated in real time through fire dynamics simulation, deep reinforcement learning optimization and a virtual reality technology. The drill personnel enter the fire disaster simulation environment through the virtual reality equipment to carry out emergency operation, and the system carries out real-time feedback according to the operation of the drill personnel and optimizes the fire disaster spreading simulation model. Parameters such as wind speed and wind quantity of the ventilation system and fire spreading are subjected to coupling calculation, so that the accuracy and dynamic response of fire simulation are ensured. According to the system, through cooperative work of a plurality of modules, intelligence, individuation and high efficiency of mine fire emergency drilling are realized, and the emergency response capability of drilling personnel is effectively improved.

The invention discloses a wharf risk assessment and management method based on artificial intelligence, particularly relates to the field of wharf risk assessment and management, and is used for solving the problems that an existing wharf is difficult to quantify and delay influence on a supply chain network level, and prospective intervention cannot be performed on supply chain risks. According to the method, a comprehensive data platform of wharf operation and supply chain cargo linkage is constructed, and the bottleneck time point of a future delivery window is analyzed and predicted based on the ship arrival density, the operation bearing capacity and the stockpiling structure; searching a historical supply chain interruption event and extracting corresponding cargo and node information; tracking a conduction path of an interruption event in the supply chain network, quantifying the influence of a bottleneck on a downstream node, and establishing a supply chain risk influence model; generating a supply chain risk influence map in combination with future bottleneck prediction; and finally, a composite risk portrait is constructed, and targeted risk intervention is carried out on the wharf and the supply chain node. And intelligent prediction and management of wharf and supply chain full-chain risks are realized.

The invention relates to the technical field of industrial detection, and discloses a lithium niobate metasurface defect intelligent identification method based on reactive ion beam etching, which systematically solves the problem of scarcity of experimental data through physical driving simulation and a physics-based rendering algorithm, and provides large-scale field specific training data for a deep learning model. The bidirectional coupling simulation model generates defect morphology prediction data conforming to an etching dynamics law based on real process parameters and material parameters, the prediction data is converted into a synthetic image with real microscopic image statistical characteristics by a physical rendering algorithm, and the quality and consistency of the synthetic data are ensured by an automatic labeling and statistical verification process. According to the method, the number of samples of the ternary association database is multiplied, and extreme process conditions and rare defect modes which are difficult to obtain through experimental collection are covered.

The invention discloses a thoracic surgery complication risk intelligent pre-judgment and prevention and control system, and relates to the technical field of medical surgery, the system comprises a data acquisition module, an edge fusion module, a baseline storage module, a two-dimensional evaluation module and a prevention and control execution module; according to the invention, intraoperative multi-source dynamic data are comprehensively collected through the data acquisition module, after feature quantification is carried out on image data through the edge fusion module and normalization processing is carried out on other data, intraoperative real-time comprehensive data with unified dimensions are generated through fusion operation, and meanwhile, preoperative baseline data are called by relying on the baseline storage module; the two-dimensional evaluation module is used for correspondingly matching intraoperative comprehensive data with preoperative baseline data, and a pre-judgment result of a complication occurrence condition is formed through multi-step quantitative calculation, so that full-process collaboration from data acquisition, fusion processing to evaluation and pre-judgment is realized, the pre-judgment process depends on integration and multi-dimensional analysis of multi-source data, and the pre-judgment efficiency is improved. And a complete logic closed loop is formed.

An example operation may include at least one of producing, by a class-conditioned sample generator executing on at least one processor communicatively coupled to a memory on a host platform, a synthetic feature set based on a label sequence and class information derived from received data, transmitting, by the host platform, a finalized synthetic sample to a computing device when the synthetic feature set satisfies a fidelity threshold, generating, by the computing device, a fidelity score based on a comparison of the finalized synthetic sample to the label sequence and the class information, retraining, by the computing device, the class-conditioned sample generator based on the fidelity score, and validating, by the computing device, the class-conditioned sample generator by transmitting a test prompt to the host platform, receiving a synthetic response generated by the class-conditioned sample generator, and comparing the synthetic response to previously stored synthetic data to validate the class-conditioned sample generator.

The invention provides a fault detection semi-supervised learning method for a three-dimensional seismic data sparse labeling scene. The fault detection semi-supervised learning method is used for realizing stable and continuous three-dimensional fault identification under the condition that only a small number of two-dimensional slices are labeled and a large number of voxels are not labeled. According to the method, firstly, a supervision mask is generated in a three-dimensional space based on two-dimensional marking, and supervision learning is executed only in an effective area to adapt to marking scarcity characteristics; and then, constructing a double-student model, keeping prediction coordination in an unlabeled area through mutual generation of pseudo labels, probability consistency and confidence constraint, and applying consistency constraint to overlapped areas of adjacent three-dimensional sub-blocks in combination with a teacher model so as to maintain spatial continuity of a fault structure. Through joint optimization of synthetic data pre-training and field data migration training, the model can still obtain reliable fault structure expression under the field conditions of strong noise, weak reflection and obvious cross-work-area difference.

A system, method, and computer-readable medium for generating factual and / or counterfactual data are described. This may have the effect of improving the complexity of data available for training machine learning models. The models may include, but not limited to, a probabilistic graphical model (PGM) and / or an agent-based model (ABM). Further aspects may provide for scrubbing actual data to create a data model that does not reveal the content of the underlying source data. Yet further aspects may provide for validating a data model.

Systems, apparatuses, methods, and computer program products are disclosed for generating representative training data. An example method includes comparing, by data analysis circuitry, a labeled dataset to a target dataset. The example method also includes generating, by training data circuitry, a training dataset based on the comparison, wherein the training dataset comprises at least a portion of the labeled dataset supplemented by a synthetic dataset. The example method also includes training, by modeling circuitry, a model using the training dataset.

The invention relates to the technical field of query characterization, and discloses a database query feature vector generation method, which comprises the following steps of: obtaining a connection bitmap and a database mode graph of a database queried by a target query statement; determining a database mode sub-graph related to the target query statement according to the database mode graph; encoding the database mode sub-graph through a preset gating graph neural network to generate a mode item vector; and determining a query feature vector corresponding to the target query statement according to the connection bitmap and the mode item vector. According to the method, by obtaining the database mode graph and the connection bitmap, the structure association information between the table and the column in the database and the predicate condition features of the query can be synthesized, the semantic association represented by the features is enhanced based on the coding capability of the graph neural network, and the expression capability of the query feature vector for query intention and data association is effectively improved.

The invention relates to an urban planning management method, system and device based on big data analysis and a medium, and the method comprises the steps: obtaining original urban planning data of multiple departments, and generating an encrypted data set with spatial attributes through homomorphic encryption and spatial grid coding; injecting antagonism abnormal data in the ciphertext space to generate a spatial correlation synthetic data set; normal and abnormal event prediction gradients are fused based on double-branch federal training, and a global prediction model is constructed; the privacy risk and the model precision are dynamically balanced through reinforcement learning, and the fusion weight and the synthesis proportion parameter are optimized; and solving the optimal solution of the infrastructure layout in combination with spatial constraints, and generating an encryption planning scheme. The method breaks through the limitation that privacy protection and data utility are difficult to consider in the traditional technology, solves the problems of abnormal simulation failure caused by encrypted data distortion, precision-safety contradiction caused by static parameter stiffness and lack of multi-target collaborative optimization of a planning scheme, and improves the scientificity and adaptability of urban planning decision.

Techniques and systems for generating synthetic data 200 are described. The described techniques and systems provide realistic synthetic data by preserving observable and missing data distributions. A method of generating synthetic data includes receiving data having missing elements having missing elements; evaluating the data for patterns with respect to the missing elements; labeling the data according to the patterns; generating labeled synthetic data using the labeled data; and inserting blanks into the labeled synthetic data according to associated labels of the labeled data to generate synthetic data with corresponding missing elements.

Certain embodiments of the disclosure provide techniques for knowledge refinement for language model fine-tuning. A method generally includes obtaining a raw information item; partitioning the raw information item into a plurality of first contextual units, wherein each first contextual unit comprises a first portion of the raw information item; generating, via a first language model, first synthetic data based on: the plurality of first contextual units; the raw information item; and at least one of fine-grained synthesis, interleaved generation, or assembly augmentation; and fine-tuning a second language model based on the first synthetic data.

The invention discloses a content security reasoning auditing method based on knowledge enhancement. The method comprises the following steps: establishing an attribute system based on illegal content in a pre-established standardized knowledge rule base, and structuring the attribute system into an attribute set; a prompt text is generated based on the attribute set and used for calling a teacher model to generate candidate answers through reasoning, and a synthetic data set is established through the candidate answers and the standardized knowledge rule base; using the synthetic data set to guide the student model to supervise and fine-tune to obtain a content category judgment and explanation text, and then optimizing the student model through reinforcement learning training; reasoning explanations are generated through the student model after training optimization and used for manual rechecking and responsibility tracing, and finally violation category labels are output. On the premise of ensuring detection precision and robustness, judgment reasons and bases are explicitly output, transparency and consistency of auditing results are enhanced, reasoning and deployment cost is remarkably reduced, and comprehensive requirements of an internet platform in the aspects of real-time performance, compliance and interpretability are met.

The invention provides a multi-parameter water quality data fusion analysis method and system. The method comprises the steps that water quality parameters are collected to form a three-dimensional data cube; constructing a space-time tensor model by using Tucker decomposition and a graph convolution network, and generating a core tensor matrix; constructing a dynamic constraint library and embedding the generative adversarial network; training a generative adversarial network by using Transform and physical constraint loss, generating synthetic data and verifying the synthetic data; performing space-time fusion by using meta learning weight distribution and Bayesian deep learning to generate a weight matrix and a confidence interval; missing data are restored through physical constraint interpolation and Gaussian process regression, and SHAP and LIME interpretation and path diagrams are generated; and performing real-time analysis by using an edge-cloud collaborative architecture to generate an intelligent report. Through physical constraint modeling, dynamic weight distribution and edge-cloud collaborative architecture, the problems that synthetic data violates physical laws, weight staticization, response lag and insufficient interpretability are solved.

Provided in the present application are an image enhancement method and apparatus, an electronic device, and a storage medium, relating to the technical field of image processing. The method comprises: acquiring a medical data set, and collecting speckle images by means of an optical fiber endoscope; generating a synthetic data set on the basis of the speckle images and medical images in the medical data set; and performing adversarial training on a deep learning model on the basis of the synthetic data set and the medical data set to generate an image reconstruction model, so as to perform image reconstruction on endoscopic images on the basis of the image reconstruction model to obtain image reconstruction results. The present application solves the problem of poor image enhancement effect of image enhancement in the related art.

The invention belongs to the technical field of computer vision and ecological monitoring, and particularly relates to a large-scale bird flock counting system and method based on pure synthetic data training. In a model training stage, a large-scale bird flock composite image and a pixel-level label thereof are used as a unique training set, and end-to-end training is performed on a point prediction deep learning counting network using VGG19-BN as a backbone network to obtain a trained bird flock counting model; in the model training stage, real world bird flock images to be counted are input into the trained bird flock counting model, the model directly outputs predicted bird individual position points, a final counting result of the bird flock is obtained by counting the number of prediction points with high confidence, and a bird flock density distribution diagram can be generated for visualization. The invention provides a method and a system for realizing real scene high-precision counting only through synthetic data training based on a point prediction network aiming at the technical problems of strong dependence of real labeled data and insufficient cross-domain generalization ability of a model in a bird flock counting task.