189 results about "Bayesian optimization" patented technology

The invention discloses an automated supervised learning method with multi-source data supported, which comprises steps of (1) data pre-processing, (2) feature engineering, (3) model and super parameter regulation and (4) Bayesian pipeline optimization. The traditional data analysis process is automated, the process of manually regulating a machine learning pipeline is fundamentally improved, under the high coupling degree of super parameter regulation and pipeline optimization, the extensibility of the system on a supervised learning algorithm is greatly improved, a genetic algorithm is creatively put forward for regulation on the super parameters for the machine learning pipeline, and the efficiency for automated parameter regulation is greatly improved. Besides, a Bayesian optimizer is adopted for optimizing a pipeline algorithm combination, the problem of combination space explosion can be greatly solved, and finally, the accuracy and the efficiency of the automated supervised learning method are improved in the result.

The invention discloses a LightGBM fault diagnosis method based on improved Bayesian optimization. The LightGBM fault diagnosis method comprises the following steps: 1) determining hyper-parameters needing to be optimized by a LightGBM model and a hyper-parameter value range; 2) improving the Bayesian optimization algorithm to obtain an improved Bayesian optimization algorithm GP-ProbHedge; 3) selecting an optimal hyper-parameter combination of the fault diagnosis model by using the method in the step 2) in combination with a five-fold cross validation mode; and 4) constructing an improved Bayesian optimization LightGBM fault diagnosis model, and giving a model iteration process and an optimization result. By adopting the technology, compared with the prior art, according to the invention,an improved Bayesian optimization algorithm is provided to carry out optimization selection on parameters of a fault model; by improving an acquisition function of a traditional Bayesian optimizationalgorithm and a covariance function of a Gaussian process of the traditional Bayesian optimization algorithm, an improved Bayesian optimization LightGBM fault diagnosis method is provided, and equipment faults are diagnosed and predicted.

The invention discloses a non-contact driver fatigue detection method based on a millimeter-wave radar. A millimeter-wave radar is used to collect a thoracic cavity vibration signal of a driver in a driving process; after the thoracic cavity vibration signal is preprocessed, the heart rate signals and respiratory signals of different frequency bands are separated; the heartbeat frequency and the respiratory frequency of the separated heart rate and respiratory signals are calculated by adopting a wavelet transform method, and seven derived physiological characteristics are calculated; by exploring the change rule of the physiological features along with time, the physiological features are found to show a good linear change trend along with accumulation of driving time; and the random forest algorithm based on Bayesian optimization effectively discriminates occurrence of fatigue time, and algorithm accuracy can be improved compared with algorithm accuracy of an original random forest model. According to the invention, the discomfort brought to a driver when the driver wears various devices to carry out physiological signal detection is solved, detection cost is reduced, the fatiguemoment of the driver can be accurately predicted, fatigue early warning is sent to the driver, and therefore, the traffic accident rate caused by fatigue is reduced.

The invention provides a data enhancement strategy generation method and device and a data enhancement method and device. The method comprises the following steps: obtaining training data and verification data; outputting a data enhancement strategy through a strategy generator, performing data enhancement processing on the training data through the data enhancement strategy, training a to-be-trained network by adopting the enhanced training data, verifying the trained network through verification data, and adjusting parameters in the strategy generator according to an obtained verification result; and circularly executing the steps until a preset condition is met, thereby obtaining a final data enhancement strategy. According to the method, a Bayesian optimization mode is introduced to search a data enhancement strategy; in the searching process, the strategy generator predicts the data enhancement strategy based on the historical verification result, then the parameters of the strategy generator are continuously adjusted in the subsequent verification process, and finally the optimal data enhancement strategy is obtained, so that the implementation difficulty of searching the optimal strategy is reduced, and the calculation cost is saved.

A system for enhanced power system model calibration is provided. The system is programmed to store a model of a device. The model includes a plurality of parameters. The system is also programmed to receive a plurality of events associated with the device, receive a first set of input calibration values for the plurality of parameters, sequentially analyze the plurality of events in a first sequence to determine a set of calibrated parameter values for the model, validate the set of calibrated parameter values for the model to determine fit, and perform Bayesian optimization on the determined fit, the set of calibrated parameter values for the model, and the plurality of events.

The invention provides an electromechanical cooperative design method of a skin antenna based on parallel Bayesian optimization in order to reduce calculation amount, improve design quality and optimize efficiency. Determine the design variables and initial design space, carry out initial sampling, extract all sample points in the sample point database and their corresponding response values, select Bayesian algorithm to construct the surrogate model of the objective function, use the pseudo-expectation enhancement criterion as the collection function to select multiple candidate solutions, and judge the degree of satisfaction of the constraint conditions. If the number of calculation times of the electromechanical coupling model reaches the set number or satisfies the convergence criterion, the iteration is stopped and the optimal solution is outputted. A large amount of optimization time is saved for complex problems, and only one sample point is randomly selected for initial sampling, so the optimization process is repeatable.

The embodiment of the invention discloses a short-term traffic flow prediction method and system based on Bayesian optimization. The method comprises the steps of: collecting original traffic flow data passing through a fixed road position in a fixed time interval, carrying out preprocessing on the original traffic flow data according to a seasonal model algorithm to generate time sequence trafficflow data; constructing a short-term traffic flow prediction model based on a support vector regression machine, and training the short-term traffic flow prediction model; calculating a mean absolutepercentage error of the short-term traffic flow prediction model, and acquiring the prediction precision of the short-term traffic flow prediction model according to the mean absolute percentage error; optimizing the model parameters corresponding to the prediction precision according to a Bayesian optimization algorithm until a target short-term traffic flow prediction model is generated; and predicting the short-term traffic flow according to the target short-time traffic flow prediction model. According to the embodiment of the invention, the generalization ability of the short-term traffic flow prediction model is improved, the prediction precision is improved, and convenience is provided for intelligent traffic.

The invention relates to a predictive operation and maintenance-oriented power battery health prediction method, and belongs to the technical field of battery management. According to the method, a usefulness evaluation system is used for evaluating health factors, a Bayesian optimization method is used for optimizing a threshold value, transfer learning is used for transmitting most similar attenuation battery model training information to a test single body, early attenuation data is used for fine adjustment of a model, and finally, the trained model is used for health prediction. The methodcomprises one-step attenuation prediction and extrapolation residual capacity prediction, and model self-correction is carried out by utilizing online extracted features. According to the method, theproblem that the on-line capacity is unmeasurable when the battery capacity is traditionally used for prediction is avoided, prediction of retirement points is provided for predictive operation and maintenance of the battery, self-correction can be carried out in real time in the using process, and the prediction precision is improved.

The invention relates to an intelligent automobile rapid test method based on Bayesian optimization. According to the method, representative test scenes are accurately selected, the test frequency isreduced, and the safety of a large number of samples is evaluated. The method comprises the following steps: firstly, obtaining driving scene key parameters of a vehicle on the basis of on-site traffic scenes, determining a value range and a sampling interval for the key parameters, and combining the key parameters to form a parameter space; then, based on the Bayesian optimization theory, selecting appropriate classifiers and acquisition functions according to different test purposes; and finally, initializing a classifier, calculating a numerical value of an acquisition function, and selecting a next intelligent automobile test scene which is more in line with requirements according to the numerical value of the acquisition function. Compared with the prior art, the method has the advantages of reducing the test times, improving the test efficiency, ensuring the test reliability and the like.

Systems and methods are provided for optimally allocating resources used to perform multiple tasks / jobs, e.g., machine learning training jobs. The possible resource configurations or candidates that can be used to perform such jobs are generated. A first batch of training jobs can be randomly selected and run using one of the possible resource configuration candidates. Subsequent batches of training jobs may be performed using other resource configuration candidates that have been selected using an optimization process, e.g., Bayesian optimization. Upon reaching a stopping criterion, the resource configuration resulting in a desired optimization metric, e.g., fastest job completion time can be selected and used to execute the remaining training jobs.

The invention discloses a transient stability evaluation method for Bayesian optimization LightGBM. The method comprises the following steps: obtaining a transient stability data set of a power system; training the Bayesian optimized LightGBM by using the data in the transient stability data set to obtain an optimal parameter of the LightGBM, and then obtaining the trained LightGBM; obtaining thedata used for evaluating the transient stability of the power system on line after the power system has a fault, preprocessing the obtained data, and inputting the preprocessed data into the trained LightGBM to obtain a power system transient stability evaluation result after the power system has the fault. According to the method, the transient stability state under multiple complex uncertain factor 'combinatorial number explosion' can be quickly and accurately evaluated, and the online evaluation of the transient stability of the power system is facilitated.

The invention discloses a diagnosis model for industrial equipment fault diagnosis, a construction method thereof and an application thereof. The construction method comprises the steps: employing a variational mode decomposition method to decompose and train each original vibration signal of industrial equipment, obtaining a plurality of sub-modal components, and selecting an optimal sub-modal component from the plurality of sub-modal components; adopting a Bayesian optimization one-dimensional fast non-local mean method to denoise all the optimal sub-modal components; and based on all the denoised optimal sub-modal components, adopting metric learning to change a sample distance metric function in the classifier and training to obtain a diagnosis model of industrial equipment fault diagnosis. According to the construction method, variational mode decomposition is adopted to separate fault features from original vibration signals; and a non-local mean denoising algorithm with high denoising performance is further introduced, and Bayesian optimization is carried out on the parameters, so that a good denoising effect on the vibration signal with the high signal-to-noise ratio is realized; and finally, metric learning is applied to training of the classifier. The construction method is high in diagnosis accuracy and wide in application range.

The invention belongs to the image recognition field and relates to a Bayesian optimization-based image table character segmentation method. The method includes the following steps that: cells in an image table to be recognized are detected; information content in each cell is segmented out wholly; spare segmentation points of characters are found for character information in each cell through using a projection method; a Bayesian classifier is utilized to judge the spare segmentation points, so that accurate segmentation points of the character information can be found out; and segmented character sub images are classified. With the Bayesian optimization-based image table character segmentation method adopted, favorable conditions are created for the accurate recognition of the character information; the completeness and accuracy of single segmented characters can be ensured; the problem of difficulty in image table segmentation in the image and character recognition field which has been difficult to be solved can be solved. The Bayesian optimization-based image table character segmentation method has very high accuracy in image table character segmentation and has a bright application prospect in the image and character recognition, information mining and information analysis field.

The invention discloses a water chilling unit fault diagnosis method and system based on Bayesian optimization LightGBM, and a medium. The method comprises the following steps: collecting and storing on-site historical data of a water chilling unit through a sensor; preprocessing the historical data; performing feature selection by using a two-step method combining an embedding method and a recursive feature elimination method; using the historical data subjected to data preprocessing and feature selection for training a LightGBM model, combining a Bayesian optimization algorithm with a ten-fold cross validation mode to determine an optimal hyper-parameter combination of the LightGBM model, and then obtaining a trained LightGBM diagnosis model; and preprocessing the real-time operation data and inputting the data into the diagnosis model to obtain a water chilling unit fault diagnosis result. Parameters of the diagnosis model can be rapidly determined, the operation state of the water chilling unit can be rapidly and accurately evaluated, key fault features can be extracted, and the method is used for guiding engineering practice and strengthening on-site operation maintenance of the water chilling unit.

The invention belongs to the field of automatic optimization design of analog circuit parameters in integrated circuit design, and particularly relates to a multi-target Bayesian optimization method based on a Gaussian process model. According to the method, in each iteration, a Gaussian process model is constructed for each performance index, a low-confidence interval function is further constructed, and a point for next circuit simulation is selected through multi-objective optimization of the low-confidence interval function. Compared with an international mainstream method at present, themethod provided by the invention can greatly reduce the number of times of circuit simulation and obtain a high-precision Pareto front edge.

The invention discloses a point cloud registration method based on Bayesian optimization, a system and a readable storage medium. The method comprises the following steps: obtaining a contour point cloud of a target object and a template contour point cloud; Build the initial data set; According to the Bayesian optimization algorithm, the initial angle is selected. According to the initial angle,the contour point cloud is rotated and translated to obtain a new contour point cloud. The new contour point cloud and the template contour point cloud are registered by ICP, and the fitness is obtained. And extracting a rotation parameter and a translation parameter of the target object based on the first fitness to estimate a pose of the target object. The invention utilizes the contour point cloud of the target object to carry out registration and optimization based on the Bayesian optimization algorithm, and finds out the relatively optimized initial position and posture of the target object, so that the robot adjusts the picking action according to the initial position and posture, and quickly completes the picking of the target object.

The invention relates to an image data enhancement strategy selection method and a face recognition image data enhancement method. The image data enhancement strategy selection method comprises the steps of obtaining a data enhancement strategy search space and a target object image set; randomly selecting a plurality of data enhancement strategies in the search space; respectively calculating image classification errors corresponding to a plurality of randomly selected data enhancement strategies to obtain an initial data enhancement strategy error pair set; and searching and selecting an optimal data enhancement strategy from the search space by adopting a Bayesian optimization method. The face recognition image data enhancement method comprises the step of applying the selected optimaldata enhancement strategy to a to-be-enhanced face recognition image. According to the data enhancement strategy selection method, the optimal data enhancement strategy only needs to be selected onceaccording to the existing target object image set, the optimal data enhancement strategy can be applied to the same kind of target object image data for image data enhancement, and the use efficiencyof the image data enhancement method is improved.

The invention belongs to the technical field of bearing fault diagnosis, and discloses a bearing fault diagnosis method, system, device and terminal, and the method comprises the steps: extracting the time-frequency features of an original vibration signal of a bearing through continuous wavelet transform, and converting the time-frequency features into a 32 * 32 pixel two-dimensional image; using an improved AlexNet model to carry out fault feature extraction on the time-frequency spectrogram; for fault diagnosis classification, selecting optimal model parameters through an LGBM classification algorithm and Bayesian optimization. The bearing fault diagnosis method provided by the invention has optimal fault diagnosis accuracy. Through experimental comparison, compared with other seven methods, the method provided by the invention has the highest accuracy rate of 99.712%, the prediction time consumption of 1.47 seconds for 1800 samples is also at the same order of magnitude as the time consumption of other models, the variance of the accuracy rates of five times of prediction is only 0.063, and the method provided by the invention is more stable than other six methods and has optimal comprehensive performance.

The invention discloses a transient stability evaluation method and system based on automatic and interpretable machine learning, and the method comprises the steps: firstly obtaining monitoring data,extracting key feature data, and inputting the key feature data into an automatic and interpretable machine learning model; evaluating the transient stability condition of the power system by the transient stability evaluation sub-model, and calculating the contribution degree of each piece of key feature data to an evaluation result by the interpretable sub-model; when the transient stability evaluation sub-model is trained, screening original data, extracting low-dimensional key characteristic data, using a Bayesian optimization model to automatically adjust and optimize hyper-parameters ofthe transient stability evaluation sub-model, and evaluating evaluation performance of the transient stability evaluation sub-model according to accuracy, recall rate and area below an ROC curve. Thekey feature data can be automatically extracted, the hyper-parameters of the evaluation model can be automatically adjusted, the transient stability evaluation of the power system is realized, and the transient stability evaluation result is explained.

The invention relates to the technical field of infrared spectroscopy wavelength, and more specifically, relates to a novel infrared spectroscopy wavelength selection method based on sparse Bayesian study. The sparse optimized wavelength selection method utilizes the spectral structure priori knowledge and the related priori knowledge of co-linearity between spectrums. The provided method has the advantages of little calculation quantity, few adjustable parameters, and strong robustness. For the first time, the spectral structure priori knowledge and the related priori knowledge of co-linearity between spectrums are utilized, and the spectral partitioned sparse structure can be determined in a self-adapting mode. Then sparse Bayesian study algorithm is adopted to calculate out the optimal solution of the sparse optimization problem so as to screen out the best wavelength point combination. The provided method can be widely applied to the field of solid-phase / liquid-phase / gas-phase infrared spectrum wavelength selection.

The embodiment of the invention provides a searching method and device for distributed model parameters and an electronic device, and the method and device specifically comprise the steps: carrying out the verification calculation of each gridding space through employing a pre-obtained verification sample set, and obtaining a cross verification mean value of an evaluation index corresponding to each gridding space; selecting a current optimal cross validation mean value from the obtained cross validation mean values, and selecting a most possible local search space according to the current optimal cross validation mean value; and performing Bayesian optimization calculation in the local search space according to a pre-selected optimal sampling point to obtain an optimal parameter combination in the local search space. According to the scheme, the search space is reduced on the basis of the corresponding rule, and then the parameter search is carried out in the reduced search space, sothat the parameter optimization efficiency is improved when the parameter space dimension is relatively high.

One embodiment of a method includes determining, by a Bayesian optimizer, a first sparsity ratio associated with a limit on an accuracy loss caused by compressing the machine learning model. The method further includes selecting, by the Bayesian optimizer, a second sparsity ratio that optimizes a predefined objective function for the machine learning model within a search space bounded by the first sparsity ratio. The method further includes generating a compressed version of the machine learning model having the second sparsity ratio.

The invention discloses a machine learning hyper-parameter adjusting method and system, relating to the technical field of machine learning. According to the invention, a Bayesian optimization framework is adopted, a prior probability model is generated by a Gaussian process through an initial observation set, a next evaluation point is searched by using a chaotic particle swarm method, a new evaluation point is evaluated to obtain an observation value corresponding to the new evaluation point, a Gaussian process probability agent model is updated by updating the observation set, and an optimal hyper-parameter combination is obtained through multiple iterative search updates. According to the method, the evaluation frequency of hyper-parameter tuning of a machine learning model can be effectively reduced, the defect that traditional Bayesian optimization falls into the local optimal point is overcome, and hyper-parameter tuning can be more accurately and efficiently carried out.

The invention provides a preprandial insulin dosage individualized decision-making system based on a Gaussian process. The system utilizes an artificial intelligence method to mine information contained in patient blood glucose monitoring and insulin infusion data at the same time, establishes a postprandial blood glucose prediction model conforming to a human body metabolism rule, utilizes a risksensitive optimization control thought to determine the risk sensitivity coefficient in an individualized mode, and uses a Bayesian optimization method to solve the optimization problem, so the safeand effective preprandial insulin dosage can still be determined even under the condition that deviation exists in model prediction, and postprandial blood glucose management is improved. Therefore, in order to improve postprandial blood glucose management, the preprandial insulin dosage individualized decision-making system based on Gaussian process learning risk sensitive control is designed byutilizing historical data of a patient, the system fully excavates information of historical data of blood glucose metabolism of the patient, and the postprandial blood glucose prediction model is established, so subsequent postprandial blood sugar control or postprandial high and low blood sugar early warning can be conveniently implemented.

The invention provides a method for automatic optimal model selection based on big data. The method comprises: step S1, classifying mining targets; step S2, using information gain to perform rapid feature selection on a whole data set; step S3, establishing a training set and a verification set; step S4, selecting an effective data mining algorithm and a parameter combination thereof; step S5, using a Bayes optimization method to select effective parameter combinations of each algorithm; step S6, selecting an optimal data mining algorithm K; step S7, using cross validation, selecting and determining a parameter value combination of the data mining algorithm K, to obtain a final model; step S8, if a result obtained by the model is relatively poor, repeating steps S2-S7, selecting the optimal model again, until a model result is satisfied; if the result is relatively satisfied, outputting the model. The method can save time consumed by automatic modeling, and modeling efficiency is improved, and the optimal algorithm can be searched rapidly from large quantity of algorithms, and parameter combinations in the optimal algorithm are selected by cross validation.

The invention provides a cigarette ventilation rate prediction method based on a gradient boosting regression tree, and the method comprises the steps: carrying out the data preprocessing to form an original data set Dataset, wherein the data comprises the characteristic data: cigarette paper air permeability, tipping paper air permeability, filter stick suction resistance, cigarette length, cigarette circumference, cigarette hardness, cigarette quality and cigarette suction resistance; dividing the original data set Dataset into a training set Transfer and a test set Test set; performing feature selection by adopting the maximum information coefficient; performing parameter optimization on the cigarette ventilation rate prediction model based on the gradient boosting regression tree by adopting a Bayesian optimization method; according to a parameter optimization result, performing model verification by utilizing data in the test set Test set, and realizing cigarette ventilation rateprediction by utilizing the verified model. The model established by the invention has the advantage of high precision, and can accurately predict the ventilation rate of the cigarette.

The invention relates to a grape wine classification method based on Bayesian optimization and an electronic nose, and the method comprises the following steps: S1, employing a LightGBM algorithm, employing a Leaf-wise tree building method, finding a leaf with the maximum splitting gain from all current leaves each time during tree building, then splitting, and repeating the above steps; the LightGBM uses the maximum tree depth to prune the tree, and excessive fitting is avoided; S2, building a Bayesian optimization algorithm; S3, building a BO-LightGBM, and performing self-optimization adjustment on hyper-parameters of the LightGBM by using a Bayesian hyper-parameter optimization algorithm; enabling bayesian optimization to use a probability model to replace a complex optimization function, introducing the prior of a to-be-optimized target into the probability model, thus the model can effectively reduce unnecessary sampling. The Bayesian optimization method has the advantages that the Bayesian optimization method determines the optimization method of the next evaluation point by constructing the probability model of the function to be optimized and utilizing the probability model, the most advanced result is achieved on some global optimization problems, and the Bayesian optimization method is a better solution for hyper-parameter optimization.

The invention discloses an air conditioner fault diagnosis method based on a Bayesian optimization PCA-limit random tree. The air conditioner fault diagnosis method comprises the following steps: 1) acquiring operation data of an air conditioner under normal and different faults and normalizing the operation data; 2) carrying out dimensionality reduction on the normalized data through a PCA algorithm, and taking the normalized data as the input of an ExtraTree model; 3) establishing a limit random tree classification model, training and testing a classifier, and obtaining a PCA-limit random tree fault diagnosis model for an air conditioner; 4) utilizing a Bayesian optimization algorithm to optimize the feature number and the CART decision tree number of a PCA-extreme random tree fault diagnosis model after the PCA dimension reduction to obtain the optimal feature number and the optimal CART decision tree number after the dimension reduction; and 5) then, taking the calculated optimal PCA dimension-reduced feature quantity value and CART decision tree quantity value as parameters of a PCA-limit random tree model, training a sample to obtain a PCA-limit random tree fault diagnosis model, and then using the diagnosis model to diagnose real-time data.

The invention provides a pedestrian re-identification method and device based on a Bayesian optimization technology and a computer system, and the method comprises the steps: carrying out the detection of a pedestrian image data set in a video through employing a depth target detection algorithm, and obtaining all pedestrian image data of a to-be-extracted feature; performing depth feature extraction on all detected pedestrian image data to obtain all feature value data of pedestrians in the video, thereby establishing a retrieval feature database; and inputting a to-be-queried picture, carrying out depth feature extraction on the picture to obtain a to-be-retrieved image feature value, carrying out Euclidean distance calculation and sorting on the to-be-retrieved image feature value and all feature values in the retrieval feature database, and outputting a sorting result. According to the scheme, the precision of an existing model is improved by improving the prior art, so that the reasoning speed is guaranteed, and the retrieval precision is improved.