62 results about "Graphical model" patented technology

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.

Embodiments of the following disclosure provide a feature importance system and method to identify features relevant to determining whether a target variable will achieve a particular value. One example method includes generating a probabilistic graphical model to represent the performance of one or more entities in a supply chain and selecting or more target variables. The method further includes collating a list of features pertaining to the one or more selected target variables, pruning at least one of the one or more features from the list and generating one or more bins in which to distribute the one or more features in the list. The method further includes modeling a network graph incorporating the one or more features in the list and bins and determining one or more inferences pertaining to the one or more supply chain entity target variables.

A machine learning system is provided to enhance various aspects of machine learning models. In some aspects, a substantially photorealistic three-dimensional (3D) graphical model of an object is accessed and a set of training images of the 3D graphical mode are generated, the set of training images generated to add imperfections and degrade photorealistic quality of the training images. The set of training images are provided as training data to train an artificial neural network.

A building management system including one or more storage devices storing instructions thereon that, when executed by one or more processors, cause the one or more processors to ingest asset information; cause a graphical model of the building to include a fault indicator based on the asset information, the fault indicator corresponding to a fault occurring on a first physical asset corresponding to a first virtual asset; cause a display device of a user device to display the graphical model within a user interface; receive a selection of the fault indicator from a user via the user interface; and in response to receiving the selection, cause the user interface to navigate to a fault-driven view of the graphical model depicting the first virtual asset and one or more second virtual assets corresponding to one or more second physical assets affected by the fault occurring on the first physical asset.

This invention discloses a low-light image quality enhancement system and method based on enhanced night scene modeling. The system includes: a sample data construction module, which generates enhanced low-light scene images based on an initial image dataset using a dual-channel hybrid generative network architecture and constructs an optimized training sample library; and a hybrid enhancement model, constructed based on an attention module and a probabilistic graphical model module. During the training phase, the hybrid enhancement model uses the optimized training sample library as input and employs a region-adaptive weighted loss function for parameter optimization. During the application phase, the optimized model receives the low-light image to be processed, estimates the illumination and noise distribution of the low-light image, and generates the enhanced image. This invention significantly improves the coverage of night scene training data and achieves adaptive enhancement of image features under low-light conditions, significantly improving the noise suppression effect in dark areas and the ability to restore image details.

The invention discloses a multi-source data fusion method and system for mobile Beidou high-precision positioning, and relates to the technical field of high-precision navigation positioning, the problem of satellite observation value quality degradation in a complex environment is inhibited through tight coupling layer explicit modeling and online estimation of multipath errors, and in the tight coupling depth fusion and calculation process, the real-time performance of the satellite observation value is improved. The constructed first factor graph model takes the multi-path error as a state parameter to be estimated for explicit modeling, the multi-path error is incorporated into a Beidou original observation factor, and in the joint optimization solution process, the first factor graph model takes the multi-path error as an independent degree of freedom for online real-time estimation and updating, so that the multi-path error is estimated in real time; according to the method, the part, attributable to multi-path interference, in the observation residual error is actively distributed and absorbed into the state parameters, so that a tight coupling resolving module can still output a more reliable tight coupling optimal pose sequence with higher anti-interference capability and provide a stable absolute position even if the satellite observation condition is seriously polluted by multiple paths.

The invention discloses a data processing method and device for conversion from a Revit model to a YJK model and a medium, and relates to the technical field of electric digital data processing, and the method comprises the steps: building a three-dimensional diagram model based on building structure three-dimensional model data in a Revit model; carrying out collinear analysis on all plane edges in the three-dimensional graph model, extracting a plurality of plane edges which are located on the same straight line and are adjacent end to end as collinear edge groups, and marking grouping identifiers; tracking continuous frames at the same plane position in a vertical space based on a collinear edge group, and calculating projection weighted center coordinates for each continuous frame based on a cross section projection overlapping region of upper and lower vertical members and a cross section projection overlapping region of a horizontal beam member; rounding the weighted center coordinate according to a preset modulus to obtain an adjusted node coordinate; and updating the original coordinates of the corresponding nodes by using the adjusted node coordinates, and recording the deviation between the original coordinates and the adjusted node coordinates as an eccentric distance to generate a YJK structure calculation model.

This invention relates to the field of financial risk early warning technology, specifically a method for dynamic prediction of repayment ability and overdue early warning based on multi-source time-series data. The method includes: acquiring multi-source time-series data of a target object over a continuous time period; generating a clean time-series data stream through time axis normalization and data quality restoration; constructing a feature set of income stability, expenditure volatility, and debt pressure based on this data stream, forming a multi-dimensional risk feature space; inputting this data into a probabilistic graphical model based on state transitions, outputting a probability distribution of repayment ability states; and calling an inversion inference engine to trace the key characteristic variables and change trajectories of abnormal states through state inversion technology, comparing them with preset risk thresholds to determine the risk critical point and level. This method achieves dynamic prediction of repayment ability and overdue early warning, improving the accuracy and targeting of risk identification.

The invention discloses a design processing method and device of a state machine, a storage medium and electronic equipment, relates to the technical field of nuclear power instrument control systems, and mainly aims to solve the problems of high cost and low reuse rate when the state machine in a nuclear power DCS (Distributed Control System) is designed by adopting existing graphical model design software. Performing modularization customization processing on each state machine element to obtain each corresponding component module; the state machine elements comprise a state machine, a state, migration and bifurcation; tool options corresponding to the state machine, the state, the migration and the bifurcation are set on a toolbar of a software page, and an incidence relation is established between the corresponding tool options and the component module; and when the tool option is selected, calling the corresponding component module based on the incidence relation, and designing a state machine structure on a software page according to the sequence of the engineering structure.

Disclosed herein are system, method, and computer program product embodiments for determining a probability of a device at risk. The device may be associated with a plurality of security parameters. For a security parameter, the device can be in multiple states. A probability value corresponding to a security parameter can indicate the security parameter being in a state among the multiple states. A probabilistic graphical model may be used to represent dependences of the plurality of security parameters. A device security risk prediction module may determine a probability of the device at risk based on the probabilistic graphical model and the probability assignments to the plurality of nodes of the probabilistic graphical model, and further determine a user action instruction to be provided to a user of the device based on the probability of the device at risk.

This invention relates to a method for simulating and analyzing rainfall-induced landslide geological disaster scenarios, belonging to the field of image synthesis, and more specifically to the field of 3D geographic modeling for computer mapping. The method includes: collecting customized model data of a given mountain at the current moment based on a 3D graphical model of the mountain; and using a landslide intelligent prediction model to intelligently predict whether a landslide geological disaster will occur on each side of the given mountain within the next time interval, starting from the current moment, based on the customized model data. This invention also relates to a rainfall-induced landslide geological disaster scenario simulation and analysis system. Through this invention, the technical problem of difficulty in predicting future landslide data for landslide-prone physical entities such as mountains is addressed by intelligently predicting whether a landslide will occur on each side of the given mountain within a future time interval based on a 3D graphical model of the mountain at the current moment, thereby solving the aforementioned technical problem.

A machine learning system is provided to enhance various aspects of machine learning models. In some aspects, a substantially photorealistic three-dimensional (3D) graphical model of an object is accessed and a set of training images of the 3D graphical mode are generated, the set of training images generated to add imperfections and degrade photorealistic quality of the training images. The set of training images are provided as training data to train an artificial neural network.

This invention discloses a method for constructing a digital twin for a water electrolysis hydrogen production system, relating to the field of digital twin technology. The method includes: collecting multi-source operational data from the water electrolysis hydrogen production system, extracting all sampling points, constructing proximity relationships between all sampling points to form a proximity connection graph; traversing the proximity connection graph, organizing the shortest path distances between all pairs of sampling points to form a high-dimensional spatial distance matrix, using multi-dimensional scaling transformation to map the sampling points to a low-dimensional space, establishing a correspondence between the low-dimensional coordinate vectors and the high-dimensional original data, forming a digital twin state representation model; based on the state representation model, constructing a probabilistic graphical model containing three types of nodes: components, sensors, and state variables; setting initial node confidence and information transmission rules, iterating until the confidence stabilizes; performing corrections based on node state deviations; and iterating in real-time according to the correction cycle to achieve real-time synchronization between the digital twin and the water electrolysis hydrogen production system.

The invention relates to the technical field of data processing, and discloses a building remote sensing interpretation method and device based on streaming processing and electronic equipment, and the method comprises the steps: obtaining multi-source heterogeneous geographic data in a target region, dividing the multi-source heterogeneous geographic data into a plurality of target sub-rectangular regions based on a preset reference coordinate system, generating a multi-band sample of each target sub-rectangle in parallel; performing graph AI model reasoning and vector conversion operation on each target sub-rectangle multi-band sample in sequence to obtain a geometrically complete building contour vector; and performing attribute AI model reasoning based on a preset target attribute, and performing multi-stage streaming attribute fusion based on the geometrically complete building contour vector to generate a vector interpretation result with attributes in the target area. According to the method, the problems of two engineering bottlenecks of low reasoning speed and high storage occupation when massive high-resolution remote sensing data is processed in the prior art are effectively solved.

The present disclosure relates to methods and systems that provide querying and analysis of clinical trials using probabilistic graphical models. The methods and systems train a probabilistic graphical model using clinical trial data and use the probabilistic graphical model to perform inferences in response to queries for clinical trials. The methods and systems use the probabilistic graphical model to handle multimodal datatypes of the clinical trial data and predict multiple attributes of the clinical trial for an input query.

The invention relates to a rainfall-induced landslide geological disaster scene simulation analysis method, belongs to image synthesis, and more particularly relates to the field of 3D geographic model modeling for computer mapping, and the method comprises the following steps: collecting various customized model data of a set mountain at the current moment based on a 3D graphic model of the set mountain at the current moment; and adopting an intelligent landslide prediction model to intelligently predict whether landslide geological disasters occur on each mountain side surface of the set mountain in the next time interval with the current moment as the starting moment according to the customized model data. The invention also relates to a rainfall-induced landslide geological disaster scene simulation analysis system. According to the invention, for the technical problem that future landslide data of a landslide-prone physical entity such as a mountain is difficult to predict in the prior art, intelligent prediction of whether landslide occurs on each side surface of the mountain in the future time interval is completed according to the 3D graphic model of the set mountain at the current moment, so that the technical problem is solved.

Provided are systems and methods for generating refined agent trajectories, leveraging a combination of neural-based trajectory prediction systems and probabilistic graphical models (PGMs). In particular, example implementations of the present disclosure utilize a probabilistic graphical model to refine agent trajectories initially predicted by a neural-based system, enhancing their adherence to fundamental movement constraints such as smooth trajectory continuity and realistic acceleration patterns. This refinement process ensures that the trajectories are not only more accurate but also comply with certain physical and practical constraints.

A computing system generates transformation error probabilities by analyzing a training data set containing training strings, each transformation error probability indicating a probability that a per-character transformation applied to a character of a training string results in a typographical error in a resulting transformation string, wherein the training data set includes strings from a historical dataset of strings including typographical errors. The computing system populates a probabilistic graphical model with the transformation error probabilities corresponding to each resulting transformation string and predicts a likelihood that an input string contains a typographical error based on the probabilistic graphical model.

The invention relates to the technical field of software development, and discloses a bidirectional mapping method, device and equipment for a structure tree and a graphical model.The method comprises the steps that a mapping script is defined, and the mapping script is used for establishing a mapping relation for elements in the structure tree and the graphical model; receiving a first operation instruction for a first area, wherein the first area is a structure tree display area or a graphical modeling area; querying related target element information from the mapping script according to the first operation instruction; generating a second operation instruction according to the first operation instruction and the target element information; the second operation instruction is applied to a second area, the second area is a structure tree display area or a graphical modeling area, and the second area is different from the first area. According to the method, the synchronization effect of the structure tree and the graphical model is realized during hierarchy change.

The invention discloses an AI-based three-dimensional model simplification method and system, a medium and equipment, and belongs to the technical field of lightweight processing of three-dimensional graphic models.The method comprises the steps that an original three-dimensional model is obtained, and texture information (including texture numbers and images) and primitive data (including primitive numbers, texture numbers and vertex data) of the original three-dimensional model are extracted; the vertex data comprises a position, a UV coordinate and a normal; inputting the texture images into a recognition model, screening pure color textures and outputting representative color values; and deleting UV coordinates and normal lines of the primitives associated with the pure-color textures, adding vertex color attributes and assigning values as representative colors to obtain first primitive data, carrying out surface reduction based on an edge collapse geometric surface reduction algorithm, and carrying out rendering in combination with texture information: if the vertexes contain the color attributes, directly rendering, otherwise, using texture images, and finally outputting a simplified model. By implementing the method and the device, the technical problem of low efficiency in the surface reduction process of the existing three-dimensional model in the prior art can be solved.

The application relates to the technical field of probabilistic graphical models, in particular to a power consumption portrait construction method and system based on a dynamic Bayesian network. Multiple hidden variables are introduced to describe the preference degree of a user for various power consumption attributes, and a power consumption portrait based on a DBN is constructed to intuitively and reasonably explain the influencing factors of power consumption rules; a VQVAE model and a GPN model are introduced to reduce the power consumption preference hidden variables, the reduction does not change the original power consumption rule distribution, the power consumption portrait accurately describes the dynamic relationship between the variables, the construction cost of the power consumption portrait is reduced, the power consumption portrait template is constructed to uniformly learn the relatively fixed relationship within and between time slices, and the related local structure in the template is updated in combination with the power consumption data of adjacent time slices, so that the dynamic relationship existing within and between time slices is captured, and efficient construction of the power consumption portrait is realized. The application aims to solve the problem of how to optimize the construction of the power consumption portrait.

Certain aspects of the disclosure provide apparatuses and methods for graphical geological process models that use FM to generate maps for offshore cable routes. A method includes receiving, by a geological process model (GPM), a data input; selecting, an FM algorithm of a plurality of FM algorithms, wherein the FM algorithm is based on a type of the data input; generating, a data output using the FM algorithm, wherein the data output comprises a prediction result of a sedimentation factor; and generating, by the GPM, a graphical layer of an interactive graphical model based on the data output, wherein the graphical layer comprises a graphic associated with the prediction result.

This invention discloses a three-dimensional backfilling system and method for pipe gallery trenches based on earthworm-inspired peristaltic shear-axial compaction coupling and artificial intelligence closed-loop control. The system includes an earthworm-inspired peristaltic shear vibration compaction unit, an axial compaction pulse unit, boundary fitting and guiding constraint components, a multimodal sensing unit, and a digital twin and artificial intelligence control unit. The peristaltic unit adopts a segmented sleeve structure, generating a directional horizontal shear wave field for initial compaction within the narrow trench through a peristaltic sequence of "anchoring phase-shear phase-propulsion phase"; the axial unit applies adjustable compaction pulses vertically for intensified compaction. The multimodal sensing unit collects soil state data in real time, and the AI ​​control unit constructs a digital twin model, employing a reinforcement learning and graphical model collaborative control strategy to optimize compaction parameters online. This invention achieves directional energy transfer and adaptive pore pressure control within a narrow space, significantly improving backfill density and uniformity, and is suitable for high groundwater conditions.

This application relates to a method for detecting genuine documents using intelligent seals. It involves acquiring color images of the stamped area and distance time series data from multiple photoelectric ranging sensors. Visual features are extracted and compared with a pre-stored template to obtain visual Mahalanobis distance. Ranging statistical features are extracted and compared with a dynamic benchmark model to obtain ranging Mahalanobis distance. Spatial consistency analysis is performed on each sensor using the ranging statistical features and ranging Mahalanobis distance to obtain ranging anomaly scores. The visual anomaly scores are weighted and fused with the ranging anomaly scores of the corresponding sensors to obtain a local fusion score. Spatial consistency propagation processing is applied to the local fusion score using a graphical model to obtain a posterior anomaly probability. The authenticity of the document is determined based on the posterior anomaly probability, visual Mahalanobis distance, and ranging Mahalanobis distance. Through multimodal feature comparison and dual verification of spatial distribution contradictions, visual forgery and physical evasion behaviors cannot simultaneously deceive both modalities, thus improving the reliability of detecting forged documents.

The present disclosure relates to a computer-implemented method and system for generating anatomical labels of an anatomical structure, the method comprising receiving an anatomical structure with extracted centerlines, or a medical image containing the anatomical structure with extracted centerlines; and predicting, by at least one processor, anatomical labels of the anatomical structure based on the centerlines of the anatomical structure using a trained deep learning network, the deep learning network being composed of a branching network, a graph neural network, a recurrent neural network and a probabilistic graphical model connected in sequence, wherein the branching network comprises at least two branch networks arranged in parallel. The computer-implemented method in the present disclosure can utilize the trained deep learning network to automatically generate anatomical labels of the whole anatomical structure in the medical image in an end-to-end manner, and the method still has high prediction accuracy and reliability in the case of large individual variation of the anatomical structure, which is conducive to helping doctors accurately and reliably diagnose.

Embodiments of the following disclosure provide a feature importance system and method to identify features relevant to determining whether a target variable will achieve a particular value. One example method includes generating a probabilistic graphical model to represent the performance of one or more entities in a supply chain and selecting or more target variables. The method further includes collating a list of features pertaining to the one or more selected target variables, pruning at least one of the one or more features from the list and generating one or more bins in which to distribute the one or more features in the list. The method further includes modeling a network graph incorporating the one or more features in the list and bins and determining one or more inferences pertaining to the one or more supply chain entity target variables.

A graphical modeling computer system configured to: (i) generate a graphical model including each payment node of a plurality of payment nodes, each funding node of a plurality of funding nodes, and a transaction edge between the payment nodes and funding nodes representing transactions; (ii) retrieve personally identifiable information (PII) for each transaction; (iii) train the graphical model to include PII embeddings; (iv) convert the graphical model into natural language sentences describing each transaction and the associated PII embeddings; (v) generate, using the GAI tools, n-dimensional vector embeddings, including the associated PII, for the transactional edge; (vi) build and train a final graphical model including approved transactions and n-dimensional vector embeddings to apply the final graphical model to a subsequent transaction between a payment node and a funding node to determine if the subsequent transaction is a fraudulent transaction.

The invention discloses an environmental hidden danger checking method and system based on AI and VR deep fusion, and the method comprises the steps: S1, constructing a high-fidelity virtual environment and a structured hidden danger library, S2, obtaining immersive hidden danger checking interaction and multi-modal behaviors, S3, carrying out the evaluation of hidden danger recognition and behaviors based on a parallel deep learning model, and S4, carrying out the recognition of hidden dangers. The method comprises the steps of S1, establishing a probability time sequence interval, S4, triggering hidden dangers and generating a dynamic comprehensive risk field based on the probability time sequence interval, S5, performing optimal evacuation path planning based on the dynamic comprehensive risk field, S6, generating an emergency strategy, and S7, pushing multi-modal information in real time. According to the invention, virtual reality hidden danger investigation training and artificial intelligence emergency decision guidance are seamlessly connected, so that real-time behavior data and risk perception information generated in the training process are directly converted into key input of emergency decision, seamless conversion and complete inspection from prevention ability cultivation to emergency decision execution are realized, and the training efficiency is improved. Meanwhile, dynamic risk simulation and personalized intelligent guidance based on a probabilistic model are introduced; the optimal path based on the environment is provided, ability evaluation results of individual users can be deeply fused, customized emergency strategies and guiding strength are generated, technologies such as virtual reality immersion rendering, artificial intelligence vision and behavior analysis, probabilistic graph models, dynamic path planning and case reasoning are deeply integrated into a unified framework, and the method has the advantages of being high in practicability and easy to popularize. Parameter circulation among the modules is natural, and synergistic interaction is achieved.

This invention belongs to the technical field of defect detection, specifically relating to a method for detecting defects in railway contact wires based on eddy currents. It addresses the technical problem that existing methods struggle to adapt to the dynamic changes in the geometric characteristics of the contact wire in different sections, leading to unstable signal decomposition results. The detection method includes: S1, generating a composite excitation signal; S2, mapping the response impedance signal and background magnetic field distribution data to an instantaneous surface coordinate system to obtain multimodal sensing data with a unified spatiotemporal reference; S3, simultaneously extracting the extreme values ​​of the local surface curvature as a third type of defect feature; and S4, constructing a probabilistic graphical model with the first, second, and third types of defect features as inputs and the defect type and severity level as outputs. This invention not only achieves the identification of defect types and levels but also quantifies the uncertainty of diagnostic results, improving the accuracy, robustness, and reliability of the detection.

The invention provides a high-order interactive prediction method and device with mixed graph deep learning, and the method comprises the steps: firstly constructing a drug molecule graph, a microorganism weighted graph, a disease weighted graph and a hypergraph connecting the three graphs based on multi-source heterogeneous data, such as a drug molecule structure, microorganism classification information and a disease semantic network, and forming a mixed graph structure; and then, through a mixed graph deep learning module fusing a graph convolutional network and a hypergraph neural network, nonlinear structure features and high-order interaction features of each entity are extracted, and adaptive fusion of the features is realized by using an attention mechanism. Next, the fused deep features are mapped into priori expectation of a potential factor matrix in a Bayesian logic tensor decomposition model, a probability graph model is constructed, and joint adaptive inference is performed on model parameters, latent variables and deep learning mapping through a variational expectation maximization algorithm, so that the probability graph model is obtained under the condition that negative sampling is not needed; and high-order association probability prediction of the full tensor space is realized.