17 results about "Risk Estimate" patented technology

The invention relates to the technical field of machine learning training fault tolerance, and discloses a LoRA training fault tolerance method based on state awareness. The method comprises the steps that training indexes are collected through a sensor network, a time sequence knowledge base is constructed, and sampling frequency is adjusted in a self-adaptive mode; historical data is utilized to generate a state change trend baseline, and abnormity is preliminarily identified through deviation degree comparison. And for the abnormal behavior, performing verification and risk assessment by fusing the gradient frequency domain characteristics and the loss curve form of the abnormal behavior. Encoding the anomalies into a high-dimensional point set, analyzing and identifying a continuous homologous structure of the high-dimensional point set by adopting topological data, and judging a systematic gradient anomaly mode according to topological characteristics; and deducing a compensation coefficient according to geometric attributes of the topological characteristics, calibrating a risk estimation value, and dynamically reconstructing a monitoring strategy. According to the method, systematic training faults can be accurately identified from a data structure level, adaptive intelligent fault-tolerant control is realized, and the reliability and efficiency of a training process are improved.

The invention discloses a dynamic risk prediction method, system and device based on a causal enhanced space-time Transform (CAST), and a medium, and the method comprises the steps: carrying out the fusion of multi-modal time series data through a causal enhanced space-time Transform (CAST) model, and obtaining a causal enhanced fusion feature representation; estimating posterior distribution parameters of generalized Pareto distribution in an extreme value theory based on the feature representation; calculating an expected tail risk estimation value as a dynamic risk prediction value; and performing early warning or dynamic pricing decision according to the risk value. According to the method, a causal inference mechanism is introduced, so that pseudo-correlation interference in multi-modal data is effectively inhibited, and high adaptability to a dynamic environment is realized; meanwhile, historical prior knowledge is combined with real-time data, so that the accuracy and stability of tail risk parameter estimation are improved; in addition, a differential privacy mechanism is introduced in a risk value output link, and it is ensured that individual privacy is not leaked. The method has wide application prospects in the fields of intelligent transportation, financial science and technology, industrial Internet of Things and the like.

The invention belongs to the technical field of energy system digitization and risk assessment, and particularly relates to a virtual power plant operation risk assessment method and equipment based on adaptive important sampling. According to the scheme, through introduction of the relaxation probability distribution sequence and parameter optimization of the adaptive important sampling density function considering the risk level, the improved important sampling method can more accurately and rapidly approach optimal sampling distribution of the virtual power plant operation risk, the evaluation efficiency is improved, and through construction of the multi-stage fusion risk estimation function, the risk estimation efficiency is improved. All sample information generated in a pre-simulation iteration process and a main simulation stage is comprehensively utilized, so that the accuracy and the stability of evaluation indexes are remarkably improved; therefore, the problem of'computational disaster 'faced by traditional Monte Carlo simulation is effectively solved, the defects that a traditional important sampling method is low in sample utilization efficiency and does not distinguish risk levels are overcome, and key risk scanning data support can be provided for real-time market decision and internal resource scheduling of the virtual power plant.

To identify a risk from a target object for supporting risk assessment.SOLUTION: A target object information acquisition unit 111 acquires a target object which is a character string or an image of a target object for which a risk is identified. An information processing unit 112 generates a prompt to identify a risk estimated from the target object acquired by the target object information acquisition unit 111, and supplies the generated prompt to a generative AI to generate identification result information indicating a result of identifying a risk.SELECTED DRAWING: Figure 2

Embodiments present techniques for determining a list of recommended items in response to a user query. An embodiment can determine a first ordered list of items including a plurality of items stored by a content platform. Based on a reward discount parameter, a first total discounted future reward for the first ordered list of items can be determined. Based on a risk discount parameter, a first risk estimate for the first ordered list of items can be determined. Similarly, a second ordered list of items can have a second total discounted future reward and a second risk estimate. The second ordered list of items can be the list of recommended items when the second total discounted future reward is larger than or equal to the first total discounted future reward, and the second risk estimate is less than or equal to the first risk estimate.

Embodiments of the present specification provide a risk detection method and device, wherein the risk detection method comprises: obtaining feature data generated by a target software product in a plurality of target feature dimensions, the plurality of target feature dimensions being obtained by analyzing historical risk data of the target software product; for any target feature dimension, using a risk analysis strategy corresponding to the target feature dimension to analyze the feature data generated in the target feature dimension, and obtaining a risk estimate of the target feature dimension; and determining a risk detection result of the target software product according to the risk estimates of the plurality of target feature dimensions. The above method can improve the accuracy of risk detection by obtaining target feature dimensions with a relatively high correlation with the risk of the target software through analysis of the historical risk data of the target software product, and determining the risk detection result of the target software product through the feature data generated in the plurality of target feature dimensions.

The invention provides a traffic risk estimation method and system based on splicing frames and causal attention, and relates to the technical field of traffic scene risk assessment. The method comprises the following steps: acquiring RGB space video data of a traffic scene, obtaining an original space diagram through frame-level segmentation, preprocessing the original space diagram to obtain a gray scale weight map and a single-channel depth map, and splicing based on a frame splicing and supplementing strategy to obtain five-dimensional input data; inputting the five-dimensional input data into the improved ResNet-50 network for risk estimation; according to the improved ResNet-50, an SCA module is added behind each residual error layer, and dimension remodeling is carried out before a feature map is input into the SCA module and when the feature map is transmitted between the residual error layers so as to ensure that frame sequence information is correctly processed; and finally obtaining a risk classification result through a global average pooling layer and a full connection layer. Through combining a frame splicing and complementing strategy of strong spatial information and sparse causal attention, the space-time correlation consistency capability of the model is enhanced, and the accuracy and robustness of risk estimation in a complex traffic scene are improved.

The present invention provides a program for estimating a risk that may occur in the future to a subject, on the basis of time-series information relating to the subject.　A risk estimation assistance program according to the present disclosure includes an information acquisition procedure, a risk calculation procedure, and a risk output procedure, and is for causing a computer to execute the procedures. The information acquisition procedure acquires time-series information indicating a feature of a subject, the risk calculation procedure calculates a risk that may occur in the future to the subject on the basis of the time-series information, and the risk output procedure outputs the risk.

Embodiments present techniques for determining a list of recommended items in response to a user query. An embodiment can determine a first ordered list of items including a plurality of items stored by a content platform. Based on a reward discount parameter, a first total discounted future reward for the first ordered list of items can be determined. Based on a risk discount parameter, a first risk estimate for the first ordered list of items can be determined. Similarly, a second ordered list of items can have a second total discounted future reward and a second risk estimate. The second ordered list of items can be the list of recommended items when the second total discounted future reward is larger than or equal to the first total discounted future reward, and the second risk estimate is less than or equal to the first risk estimate.

The invention discloses an asset loss rate estimation method, and relates to the technical field of financial risk management, and the method comprises the following steps: S1, data acquisition: obtaining internal data and external market data of a target customer, and obtaining a guarantee mode, pledge details and present value data of a target debt item; s2, determining default probability and subject rating: based on the internal data and the external market data of the target customer in the step S1, calculating future one-year default probability (PD) of the target customer by using a Merton model; mapping the default probability to a standard Puer subject rating, and outputting an internal subject rating according to an internal and external rating mapping rule; according to the technical scheme of the invention, the default probability prediction based on the Merton model and the default loss rate calculation based on multi-model fusion are organically combined through the integrated automatic evaluation process, and the full-link automatic measurement from data input to risk output is realized by means of a standardized rating mapping mechanism, so that the accuracy and efficiency of risk estimation are remarkably improved.

The invention discloses a training sample data set construction method with privacy protection and a multi-classifier, sensitive information in training data is protected by constructing a multi-hidden-label mechanism, firstly, a label space division module is utilized to divide a label set into sensitive labels and non-sensitive labels, and the sensitive labels and the non-sensitive labels are classified according to a multi-hidden-label strategy; and dividing the labels into disjoint hidden label groups to form hidden label input. And then, an unbiased risk estimation module is adopted to calculate model loss, and effective training of a multi-class classifier is realized on the premise that real label explicit exposure is not needed. Furthermore, a learning strategy based on two-step optimization is provided, the first step is to optimize the positive risk item, and the second step is to optimize the negative risk item, so that the model recognition capability is improved. According to the method, a multi-hidden label mechanism is introduced, while a plurality of sensitive labels are effectively hidden, clear supervision information is still provided for part of non-sensitive labels, a plurality of sensitive and non-sensitive class labels can be effectively recognized, and privacy protection and model learnability are both considered.

The invention relates to the field of data management, in particular to an enterprise risk early warning method and system based on multi-source data fusion. Comprising the following steps: receiving an internal document and a knowledge base, extracting an enterprise data set, and constructing a dynamic knowledge graph; in the influence propagation model based on the time-space diagram neural network, executing multi-layer iterative message passing based on a double-layer attention mechanism, and generating a health state vector; inputting the health state vector into a multi-task decoding framework, and purifying the features of each risk task through an advisory depolarization network to generate a risk specific vector; enabling the risk specificity vector to pass through a plurality of task solution terminals to obtain a risk estimation value; and the influence propagation model is updated according to the difference value between the risk estimation value and the real loss through a conflict resolution optimizer based on gradient projection. According to the method, systematic risk opening quantification is carried out through a double-layer attention mechanism, and pseudo-correlation interference in multi-source data is reduced through an adversarial depolarization network.

The invention discloses a stomach cancer postoperative prognosis evaluation system based on nonlinear Cox, and relates to the technical field of medical information, and the system comprises a data acquisition module which is used for obtaining stomach cancer postoperative pathological characteristic data of a target object; the nonlinear survival risk modeling module is used for processing the pathological feature data on the basis of a nonlinear Cox proportional risk model fused with Gaussian process prior and generating scores representing individualized continuous survival risks; the risk curved surface construction and boundary extraction module is used for constructing the continuous survival risk scores into a three-dimensional risk curved surface or a contour map and extracting an equal-risk contour boundary with clinical stable boundary significance; and the prognosis result output module is used for mapping the continuous survival risk score into risk layering information and outputting a prognosis evaluation result. According to the scheme, the complex nonlinear coupling effect between the lymph node load and the tumor infiltration depth is finely depicted, and the accuracy of risk estimation is improved.

This invention discloses a postoperative prognostic assessment system for gastric cancer based on a nonlinear Cox proportional hazards model, belonging to the field of medical information technology. The system includes: a data acquisition module for acquiring postoperative pathological feature data of the target subject; a nonlinear survival risk modeling module for processing the pathological feature data based on a nonlinear Cox proportional hazards model incorporating Gaussian process priors, and generating a score characterizing individualized continuous survival risk; a risk surface construction and boundary extraction module for constructing the continuous survival risk score into a three-dimensional risk surface or contour map, and extracting isorisk contour boundaries with clinically stable demarcation significance; and a prognostic result output module for mapping the continuous survival risk score to risk stratification information and outputting the prognostic assessment result. This application's solution achieves a fine characterization of the complex nonlinear coupling effect between lymph node burden and tumor invasion depth, improving the accuracy of risk estimation.

A computer-implemented method, an apparatus, and a computer program product for estimating an aortic rupture risk are disclosed. The method comprises: obtaining imaging data comprising three-dimensional imaging data representing an aorta of a subject; obtaining additional data related to the subject, wherein the additional data comprises multi- dimensional stress data of the aorta of the subject; fitting an aortic structural model to the imaging data to determine a geometry of the subject's aorta, and processing the determined geometry and the multi-dimensional stress data by a first estimator configured to determine a first aortic rupture risk estimate; processing the imaging data by a second estimator configured to determine a second aortic rupture risk estimate, wherein the second estimator has been trained using training data comprising three-dimensional imaging data of a plurality of aortas; determining a combined aortic rupture risk estimate based on the first aortic rupture risk estimate and the second aortic rupture risk estimate; and outputting the combined aortic rupture risk estimate and including the indicative rupture direction of the aorta.

Disclosed techniques include receiving a user model as a computational graph and a list of risk metrics indicating types of risk factors to assess, and selecting at least one composable wrapper for wrapping the user model based on a risk metric(s) in the list. The techniques further include determining a set of graph-level transformations to apply to the computational graph representation that implement a risk estimate for the user model, applying the set of graph-level transformations to modify the operations performed by the user model, and generating, based on the transformed computational graph, a modified executable variant of the user model that, when executed within a machine learning framework, produces: outputs that preserve the structure and interpretation of the user model outputs; and corresponding uncertainty estimates. Still further, the techniques include executing the modified executable variant risk-aware variant of the user model to produce prediction output(s) and / or corresponding uncertainty estimate(s).

The invention relates to the technical field of medical data processing, and discloses a diabetic nephropathy risk assessment method and model. The method includes collecting time sequence medical data of a diabetic patient, including blood detection values, anthropometric data, and diagnosis and treatment records. For each point in time, feature elements are extracted and a temporary risk estimate is calculated. And connecting a historical medical database, matching similar patient categories, and correcting the temporary estimation value by using the risk history to obtain a unified risk estimation value. And calculating a difference value between the temporary estimation value and the unified estimation value as an accuracy index. And predicting the future nephropathy risk grade of the patient in combination with the accuracy index and the characteristic time sequence change. According to the method, individual assessment is calibrated through group historical data, the built-in accuracy index is generated, and the accuracy of risk assessment and the reliability of clinical decision making are improved.