34 results about "Bayesian Prediction" patented technology

The statistical analysis described and claimed is a predictive statistical tree model that overcomes several problems observed in prior statistical models and regression analyses, while ensuring greater accuracy and predictive capabilities. Although the claimed use of the predictive statistical tree model described herein is directed to the prediction of a disease in individuals, the claimed model can be used for a variety of applications including the prediction of disease states, susceptibility of disease states or any other biological state of interest, as well as other applicable non-biological states of interest. This model first screens genes to reduce noise, applies k-means correlation-based clustering targeting a large number of clusters, and then uses singular value decompositions (SVD) to extract the single dominant factor (principal component) from each cluster. This generates a statistically significant number of cluster-derived singular factors, that we refer to as metagenes, that characterize multiple patterns of expression of the genes across samples. The strategy aims to extract multiple such patterns while reducing dimension and smoothing out gene-specific noise through the aggregation within clusters. Formal predictive analysis then uses these metagenes in a Bayesian classification tree analysis. This generates multiple recursive partitions of the sample into subgroups (the “leaves” of the classification tree), and associates Bayesian predictive probabilities of outcomes with each subgroup. Overall predictions for an individual sample are then generated by averaging predictions, with appropriate weights, across many such tree models. The model includes the use of iterative out-of-sample, cross-validation predictions leaving each sample out of the data set one at a time, refitting the model from the remaining samples and using it to predict the hold-out case. This rigorously tests the predictive value of a model and mirrors the real-world prognostic context where prediction of new cases as they arise is the major goal.

Distributed computing methods and systems are disclosed, wherein intensive fatigue-risk calculations are partitioned according to available computing resources, parameters of the fatigue-risk calculation, time-sensitive user demands, and the like. Methods are disclosed wherein execution-cost functions are used to allocate accessible computing resources. Additional methods include partitioning calculation tasks by user-prioritized needs and by general mathematical features of the calculations themselves. Included herein are methods to calculate only prediction-maximum likelihoods instead of full probability distributions, to calculate prediction likelihoods using Bayesian prediction techniques (instead of full re-tabulation of all data), to collate interim results of fatigue-risk calculations where serial results can be appropriately collated (e.g., serial time-slice independence of the cumulative task involved), to use simplified (e.g., linear, first-order) approximations of richer models of fatigue prediction, to assign user-identified priorities to each computational task within a plurality of such requests, and the like.

The present invention relates to a method of implementing, using and also testing a machine learning system. Preferably the system employs the Naïve Bayesian prediction algorithm in conjunction with a feature data structure to provide probability distributions for an input record belonging to one or more categories. Elements of the feature data structure may be prioritised and sorted with a view to selecting relevant elements only for use in the calculation of a probability indication or distribution. A method of testing is also described which allows the influence of one input learning data record to be removed from the system with the same record being used to subsequently test the accuracy of the system.

The present invention relates to a method of implementing, using and also testing a machine learning system. Preferably the system employs the Naïve Bayesian prediction algorithm in conjunction with a feature data structure to provide probability distributions for an input record belonging to one or more categories. Elements of the feature data structure may be prioritized and sorted with a view to selecting relevant elements only for use in the calculation of a probability indication or distribution. A method of testing is also described which allows the influence of one input learning data record to be removed from the system with the same record being used to subsequently test the accuracy of the system.

A system and method for making program recommendations to users of a network-based video recording system utilizes expressed preferences as inputs to collaborative filtering and Bayesian predictive algorithms to rate television programs using a graphical rating system. The predictive algorithms are adaptive, improving in accuracy as more programs are rated.

The statistical analysis described and claimed is a predictive statistical tree model that overcomes several problems observed in prior statistical models and regression analyses, while ensuring greater accuracy and predictive capabilities. Although the claimed use of the predictive statistical tree model described herein is directed to the prediction of a disease in individuals, the claimed model can be used for a variety of applications including the prediction of disease states, susceptibility of disease states or any other biological state of interest, as well as other applicable non-biological states of interest. This model first screens genes to reduce noise, applies k-means correlation-based clustering targeting a large number of clusters, and then uses singular value decompositions (SVD) to extract the single dominant factor (principal component) from each cluster. This generates a statistically significant number of cluster-derived singular factors, that we refer to as metagenes, that characterize multiple patterns of expression of the genes across samples. The strategy aims to extract multiple such patterns while reducing dimension and smoothing out gene-specific noise through the aggregation within clusters. Formal predictive analysis then uses these metagenes in a Bayesian classification tree analysis. This generates multiple recursive partitions of the sample into subgroups (the “leaves” of the classification tree), and associates Bayesian predictive probabilities of outcomes with each subgroup. Overall predictions for an individual sample are then generated by averaging predictions, with appropriate weights, across many such tree models. The model includes the use of iterative out-of-sample, cross-validation predictions leaving each sample out of the data set one at a time, refitting the model from the remaining samples and using it to predict the hold-out case. This rigorously tests the predictive value of a model and mirrors the real-world prognostic context where prediction of new cases as they arise is the major goal.

A system and method for predicting, proposing and / or evaluating suitable medication dosing regimens for a specific individual as a function of individual-specific characteristics and observed responses of the specific individual. Mathematical models of observed patient responses are used in determining an initial dose. The system and method use the patient's observed response to the initial dose to refine the model for use to forecast expected responses to proposed dosing regimens more accurately for a specific patient. More specifically, the system and method uses Bayesian averaging, Bayesian updating and Bayesian forecasting techniques to develop patient-specific dosing regimens as a function of not only generic mathematical models and patient-specific characteristics accounted for in the models as covariate patient factors, but also observed patient-specific responses that are not accounted for within the models themselves, and that reflect variability that distinguishes the specific patient from the typical patient reflected by the model.

The invention discloses a visible light dynamic positioning method based on optical flow method detection and Bayesian forecasting. According to the dynamic positioning method, an optical flow methodand a Bayesian forecasting method are utilized to detect a target in each frame of image of a moving object and calculate the distance so as to carry out dynamic positioning, wherein the optical flowmethod detection process comprises the steps of image acquisition, preprocessing and edge detection; the traveling speed displacement data of a target are acquired by utilizing an LED lamp and a high-speed camera arranged in front of and behind the moving object so as to obtain dynamic positioning information; and a search center of a next frame is set through Bayesian forecasting, the Bayesian forecasting process comprises the steps of setting the search center, forecasting the mass center position and finding out a target output mass center, the dynamic positioning method is simple and feasible, the hardware cost is low, the positioning precision is high, a system is stable, and the dynamic positioning method is not susceptible to interference and influence and has a wide application prospect.

A system and method for predicting, proposing and / or evaluating suitable medication dosing regimens for a specific individual as a function of individual-specific characteristics and observed responses of the specific individual. Mathematical models of observed patient responses are used in determining an initial dose. The system and method use the patient's observed response to the initial dose to refine the model for use to forecast expected responses to proposed dosing regimens more accurately for a specific patient. More specifically, the system and method uses Bayesian averaging, Bayesian updating and Bayesian forecasting techniques to develop patient-specific dosing regimens as a function of not only generic mathematical models and patient-specific characteristics accounted for in the models as covariate patient factors, but also observed patient-specific responses that are not accounted for within the models themselves, and that reflect variability that distinguishes the specific patient from the typical patient reflected by the model.

A system, method, and computer program product provides a useful measure of the accuracy of a Naïve Bayes predictive model and reduced computational expense relative to conventional techniques. A method for measuring accuracy of a Naïve Bayes predictive model comprises the steps of receiving a training dataset comprising a plurality of rows of data, building a Naïve Bayes predictive model using the training dataset, for each of at least a portion of the plurality of rows of data in the training dataset incrementally untraining the Naïve Bayes predictive model using the row of data and determining an accuracy of the incrementally untrained Naïve Bayes predictive model, and determining an aggregate accuracy of the Naïve Bayes predictive model.

The invention discloses a dangerous road section traffic accident early-warning method for a vehicle-mounted short-distance communication network. According to the invention, a roadside node is arranged beside a dangerous road section. Meanwhile, the information containing the speed and the type of a vehicle and periodically sent by the vehicle is collected through a communication channel of the vehicle-mounted short-distance communication network. By utilizing a bayesian network having the historical data learning function, the occurrence probability of traffic accidents of the vehicle is predicted. The implementation method comprises five parts. During the first part, a factor for determining accident prediction and a value range of the factor are determined. During the second part, a bayesian prediction model is determined according to the prediction factor and the value range of the factor. During the third part, the learning of the bayesian network is carried out and the probability of each condition is determined based on the historical data of traffic accidents and expert experiences. During the fourth part, the reasoning method of the bayesian network is determined. Duringthe fifth part, the method for collecting the vehicle information and the method for predicting the vehicle traffic accident are realized through the roadside node.

The invention discloses an image blind denoising system, and belongs to the technical field of image denoising. The method aims to solve the problem that an existing image denoising method depends onnoise and clear image pairs for model training, and feasibility is poor. A self-supervised learning module comprises a blind spot network based on hole convolution and a noise level estimation networkrelated to an image, and is used for carrying out self-supervised loss optimization on a noise sample image in a noise sample image set and obtaining a preliminary blind denoised image based on Bayesian prediction to obtain a first data set; an image-related noise level estimation network further processes the clear sample images in the clear sample image set to generate corresponding noise images, and a second data set composed of the clear sample images and the corresponding noise image pairs is obtained; and a knowledge distillation module is used for training a convolutional neural denoising network based on multistage wavelets by adopting the first data set and the second data set in a fully supervised mode to obtain a denoising model. According to the system, blind denoising is realized based on non-paired images.

The invention discloses a method for estimating the scale of a host infected by a malicious code based on DNS cache detection. The method comprises the steps that DNS parsers within the range of a specific region are detected, cache information of malicious domain names in the DNS parsers is collected, a Bayesian forecasting filter model is constructed based on the information, and the scale of the host infected by the malicious code in a corresponding network domain is estimated. According to a system, the problems existing in a traditional monitoring method such as privacy protection and network authorization are effectively solved.

The invention provides a code completion method. The method comprises the steps of obtaining a plurality of training source codes; generating abstract syntax trees of a plurality of training source codes, and generating parent abstract syntax trees and child abstract syntax trees of the plurality of training source codes according to the abstract syntax trees; generating parent control flow graphsof the plurality of training source codes according to the parent abstract syntax trees of the plurality of training source codes, and generating child control flow graphs of the plurality of training source codes according to the child abstract syntax trees of the plurality of training source codes; constructing a Bayesian prediction model according to the parent control flow graphs and the child control flow graphs of the plurality of training source codes; obtaining a source code to be completed; and inputting the to-be-completed source code into the Bayesian prediction model for code completion to obtain a subsequent code of the to-be-completed source code. The invention further provides a code completion device, a computer device and a computer readable storage medium. According to the method, complex codes can be complemented, and the software code writing efficiency is improved.

The invention provides a fault diagnosis method and system based on a case library. The method comprises the steps: acquiring a fault case sample library; determining a correlation degree between eachrelated factor in the fault case sample library and each equipment fault type by utilizing a machine learning method to obtain a target correlation degree directed graph; based on the target association degree directed graph, establishing a Bayesian prediction model of the equipment fault; and performing fault diagnosis operation on the to-be-diagnosed equipment by utilizing the Bayesian prediction model to obtain a fault diagnosis result, the fault diagnosis result comprising any one of the following items: the to-be-diagnosed equipment has a fault and the to-be-diagnosed equipment does nothave a fault. The technical problems that in the prior art, the equipment overhauling process is tedious, and the operation efficiency is not high are solved.

The invention discloses a method for estimating the scale of a host infected by a malicious code based on DNS cache detection. The method comprises the steps that DNS parsers within the range of a specific region are detected, cache information of malicious domain names in the DNS parsers is collected, a Bayesian forecasting filter model is constructed based on the information, and the scale of the host infected by the malicious code in a corresponding network domain is estimated. According to a system, the problems existing in a traditional monitoring method such as privacy protection and network authorization are effectively solved.

By combining the digital phase-locked loop technique with the particle filtering technique, the invention provides a particle filtering ultra-wideband pulse timing tracker with feedback control. Timing deviation is modeled as a random walk by combing with a digital phase-locked loop, the parameters of the random walk are determined by using the Bayesian forecasting technique, and thus, a state space model of a digital filter is built. The period of a digital clock is finally controlled by using an output signal of the digital filter, so that the timing deviation capture range is enlarged, timing precision and timing stability are increased, and loss of lock does not occur at low signal to noise ratio. In addition, because the particle filtering algorithm is adopted, the tracker can work in the colored noise environment, and the scope of application is widened.

The invention provides an intelligent unmanned chariot position loss finding method based on fog calculation. A main control unit globally controls the intelligent unmanned chariot according to the information and control algorithms of all intelligent unmanned chariots; meanwhile, each intelligent unmanned chariot can form a formation with the adjacent intelligent unmanned chariot; the position ofthe intelligent unmanned chariot is acquired by means of sensor information of the intelligent unmanned chariot. The intelligent unmanned chariot cannot obtain the position information due to weatheror geographical reasons; the main control unit adopts particle filtering and Bayesian prediction based on fog calculation; the intelligent unmanned chariot group is communicated with other intelligent unmanned chariot around the intelligent unmanned chariot at the lost position to obtain the position information of the intelligent unmanned chariot and the associated information of the intelligentunmanned chariot at the lost position, so that the position information of the intelligent unmanned chariot at the lost position is calculated, the intelligent unmanned chariot group is notified, andcoordinated fighting of the intelligent unmanned chariot group is guaranteed. Fog calculation is used for storing, analyzing, processing and mining data; data forwarding and processing are carried out; the characteristics of nonlinearity and non-Gaussian are mostly presented under the interference of various noises; particle filtering and Bayesian prediction are adopted, motion estimation and tracking are conducted through a particle filtering method, position information of the intelligent unmanned chariot group and intelligent unmanned chariot associated information of the lost position areobtained, and therefore the position information of the intelligent unmanned chariot at the lost position is calculated.

The invention provides a two-stage discrimination defect report severity prediction method based on a Spacy word vector, and the method comprises the following steps of: firstly, searching a historical defect report from a defect tracking system where a project is located, extracting corresponding contents of a description information summery attribute and a severity degree severity attribute to obtain a defect report training data set, and then performing preprocessing and generating a corresponding vector; and finally, executing a two-stage discrimination process to construct a severity prediction model. The method has the advantages that: the Naive Bayes algorithm is adopted in the two-stage discrimination process, the algorithm is easy to implement and good in effect, and the accuracyof the prediction model can be guaranteed; according to the severity prediction model, the same data is applied twice, so that on the one hand, secondary utilization of the same batch of data is realized, and the performance of the model is improved; and on the other hand, two-stage discrimination is realized, so that the pressure of the multi-classification Naive Bayes prediction model can be reduced under the condition of correct large-class classification, and the accuracy of the prediction model is further improved.

The invention relates to the technical field of traffic tool fault prediction, and discloses a shared traffic tool fault prediction method and system. The method comprises the steps of collecting vehicle condition fault data which are uploaded by operation and maintenance personnel and users and are reserved in a local server log file, reading the log file data through Apache Flume, uploading the log file data to an HDFS file system in Hadoop, then cleaning data stored in an HDFS, collecting the cleaned data in a MongoDB, screening out a new feature index set from the initial feature index set by using Spark and adopting a random forest model for the data collected in the MongoDB, establishing a naive Bayesian prediction model of vehicle faults on the basis of the new feature index set, and writing the prediction result back to the data document of each vehicle in the MongoDB to realize the prediction function of the vehicle fault, and feeding back information to the user and the maintenance department subsequently, so that the next user using the vehicle is warned, the working efficiency of the maintenance department is improved, and an improvement direction is provided for the technical department.

The invention discloses an alarm handling and receiving information scoring method based on Bayes prediction. Firstly, personal information is matched through a regular expression for word segmentation; then, a text type is predicted according to a word segmentation result, and the category which a text belongs to is judged through a naive Bayes algorithm according to the word segmentation resultand internal classification feature samples to obtain the probability that one text belongs to a case category; finally, data collision is performed through a data collision model graph to complete warning condition correlation. The category probability of one warning condition text classification is obtained by adopting a Bayes classifier according to the warning condition word segmentation result and a warning condition keyword library, then the category probability and warning condition weight are accumulated through a decision-making tree to obtain a scoring result, and collision completedbased on importance distinguishing of warning conditions after analysis of all feature information of the warning conditions can be achieved in the mode that relevant texts are associated by extracting special identifiers of warning condition texts, for example, identity numbers.

Digitized computer democracy systems allow for agents to make decisions on the behalf of clients and or beneficiaries by virtually polling their beliefs as conditions change over time. In the proposed innovation artificial intelligence, machine learning, and data analytics are used to dynamically predict the most important states to survey in a process. Beneficiaries are then surveyed about their preferences over various options in these surveyed states. When one of the states that is surveyed occurs, then voting logic and economic theory is used to make a decision on the appropriate course of action. When a state that has not been surveyed occurs, Bayesian methods are used to dynamically predict what a user's ballot would have looked like if that state had been surveyed. This Bayesian prediction is then tested, and if the result is found to be robust then a decision is reached using voting logic and economic theory. If the results are not robust, then the process is rerun to determine the most important states to survey. Using these analytic tools and techniques, the proposed invention provides a framework for practical and robust digital democracy applications for client and beneficiary management.

The application discloses a method and a device for determining geographic information points. A specific implementation of the method includes: acquiring user location information, wherein the location information includes user location coordinates; using the user location coordinates as an input value of a pre-trained Bayesian prediction model, and according to the The Bayesian prediction model obtains the probability value that the user is at each geographic information point in at least one geographic information point, wherein the Bayesian prediction model uses the basic information of the geographic information point as sample data training, wherein the The basic information includes location coordinates of geographic information points and historical location information of historical visiting users; the geographic information point corresponding to the maximum probability value is determined as the geographic information point where the user is located. This embodiment realizes determining the geographical information point where the user is located.

An image blind denoising system belongs to the technical field of image denoising. The invention aims at the problem that the existing image denoising method relies on the pair of noise and clear image to train the model, and the feasibility is poor. Its self-supervised learning module includes: a blind spot network based on hole convolution and an image-related noise level estimation network, which are used to optimize the noise sample images in the noise sample image set through self-supervised loss, and obtain preliminary blindness removal based on Bayesian prediction The noise image is obtained to obtain the first data set; the image-related noise level estimation network also processes the clear sample image in the clear sample image set to generate a corresponding noise image, and obtains a second data set composed of the clear sample image and the corresponding noise image pair; The knowledge distillation module is used to train a multi-level wavelet-based convolutional neural denoising network using the first data set and the second data set in a fully supervised mode to obtain a denoising model. The present invention realizes blind denoising based on unpaired images.

The invention discloses a visible light dynamic positioning method based on optical flow method detection and Bayesian prediction. The dynamic positioning method utilizes the optical flow method and Bayesian prediction method to detect objects in each frame of images for moving objects, calculate The distance is used for dynamic positioning. Among them, the optical flow method detection process includes image acquisition, preprocessing, and edge detection; use LED lights and high-speed cameras mounted on the front and back of the moving object to collect the speed and displacement data of the target to obtain dynamic positioning information; through Bayesian prediction sets the search center of the next frame. Bayesian prediction includes setting the search center, predicting the position of the centroid, and finding the target output centroid. This method is simple, low in hardware cost, and high in positioning accuracy. The system It is stable, not easily affected by interference, and has broad application prospects.

The invention discloses a dangerous road section traffic accident early-warning method for a vehicle-mounted short-distance communication network. According to the invention, a roadside node is arranged beside a dangerous road section. Meanwhile, the information containing the speed and the type of a vehicle and periodically sent by the vehicle is collected through a communication channel of the vehicle-mounted short-distance communication network. By utilizing a bayesian network having the historical data learning function, the occurrence probability of traffic accidents of the vehicle is predicted. The implementation method comprises five parts. During the first part, a factor for determining accident prediction and a value range of the factor are determined. During the second part, a bayesian prediction model is determined according to the prediction factor and the value range of the factor. During the third part, the learning of the bayesian network is carried out and the probability of each condition is determined based on the historical data of traffic accidents and expert experiences. During the fourth part, the reasoning method of the bayesian network is determined. Duringthe fifth part, the method for collecting the vehicle information and the method for predicting the vehicle traffic accident are realized through the roadside node.

The present invention provides a two-stage discriminant defect report severity prediction method based on Spacy word vectors. First, collect historical defect reports from the defect tracking system where the project is located, and extract description information summary attribute and severity attribute corresponding content to obtain defect report training The data set is then preprocessed and corresponding vectors are generated, and finally a two-stage discriminant process is performed to construct a severity prediction model. The beneficial effect of the present invention is: the present invention adopts naive Bayesian algorithm in the two-stage discrimination process, and this algorithm is easy to realize, and effect is good, can guarantee the accuracy rate of prediction model; On the one hand, it realizes the secondary utilization of the same batch of data, which is conducive to improving the performance of the model; on the other hand, the realization of two-stage discrimination can reduce the pressure of multi-classification Naive Bayesian prediction model when the classification of large categories is correct. Further improve the accuracy of the prediction model.

The present invention proposes an intelligent labeling device and system for switchgear based on the Internet of Things, specifically by constructing an SML model to detect and predict faults on the collected equipment operating parameters, and then provides a data collection module, data processing The information interaction technology platform integrating the central module and the marking device overcomes the defects of the existing signs such as easy displacement or untimely information update or information confusion, and effectively improves the efficiency of on-site operators' maintenance and troubleshooting and fault detection. Judgment ability, and then use the neural network algorithm to optimize the SML model, improve the accuracy and authenticity of the SML model, and further build a Bayesian prediction model for fault risk prediction, effectively contain the fault before it occurs, and realize scientific intelligence The management system ensures the safe and stable operation of the power system.

The invention discloses a storage hard disk detection and early warning method and system, which comprises the following steps: collecting physical hard disk state information; obtaining the detection type of the hard disk state information; obtaining the detection result of the hard disk state information according to whether the hard disk damage information is wrong; Whether the write rate information is higher than the preset read / write rate threshold obtains the detection result of the hard disk status information; obtains the predicted life of the hard disk based on the Bayesian prediction algorithm; obtains the detection result of the hard disk status information according to whether the predicted life of the hard disk is higher than the preset life threshold. The present invention obtains the hard disk state information detection result according to the hard disk state information detection type after obtaining the hard disk state information detection type, and monitors the running state of the physical hard disk on the server through multiple types of hard disk state information detection results, which can be very good To ensure the stable and normal operation of the distributed storage system, and reduce the impact on the user's business system due to physical hard disk problems.