30 results about "Unlabelled data" patented technology

The method is oriented to actual scenes of rolling bearings with scarce marks in cross-equipment and cross-working conditions, and solves the problems of instable precision and negative migration caused by domain migration under small sample and semi-supervised conditions. The invention provides a cross-domain transfer learning method of domain sensing data-meta initialization-teacher and student semi-supervision-curriculum type alignment-physical consistency-online updating. According to the method, order resampling, event anchoring slicing and robust scaling are matched with a self-supervision health index and order-preserving calibration to generate a soft / interval weak label; unlabeled is absorbed through consistency learning and uncertainty gating, and domain invariant representation is obtained according to low / middle / high level step-by-step alignment; negative migration is inhibited by combining monotone / integral consistency of the non-negative degradation rate with time deformation and spectrum keeping consistency, and rapid adaptation is realized by matching meta-learning and online small-step fine tuning; the resulting generic characterization and reusable initialization can be used for health grading, phase identification, and life-related estimation.

The invention discloses an image restoration method, system and device, a medium and a product, and belongs to the field of power grids, and the method comprises the steps: obtaining a to-be-restored image of a power scene under an extreme weather condition, inputting the to-be-restored image to an image restoration model, and obtaining a target image, the image restoration model is obtained by guiding an initial restoration model under a semi-supervised framework through an evaluation model to carry out iterative training, in each iteration, the label-free data is input into the first restoration model to obtain a restoration result, the restoration result is input into the evaluation model to carry out quality evaluation to obtain an evaluation signal, and the evaluation signal is sent to the semi-supervised framework; dynamically adjusting model parameters of a current second repair model based on the evaluation signal to guide the current second repair model to learn key features of the power equipment under extreme weather conditions, the evaluation model being obtained by performing all-parameter supervision fine tuning training on a multi-modal large language model based on power scene data; therefore, the generalization ability and the repairing effect of the repairing model can be improved by implementing the method and the device.

The application discloses a kind of semi-supervised identity recognition method and system based on millimeter wave radar, method includes: obtaining the point cloud data and range-velocity diagram data of millimeter wave radar;Point cloud data and range-velocity diagram data are preprocessed;The point cloud identity recognition model and range-velocity diagram identity recognition model designed are collaboratively trained using the point cloud data and range-velocity diagram data after preprocessing, the prediction probability of point cloud identity recognition model and the prediction probability of range-velocity diagram identity recognition model are obtained;The prediction probability of point cloud identity recognition model and the prediction probability of range-velocity diagram identity recognition model are fused.The application is trained by the method of semi-supervised collaboration to point cloud identity recognition model and range-velocity diagram identity recognition model, each other guide uses unlabelled data, improves model performance;Meanwhile, the prediction of two kinds of identity recognition models is fused, and higher identity recognition accuracy is realized.

In one embodiment, the present disclosure provides a computer-implemented method comprising: receiving a first labeled training dataset comprising a first plurality of images each associated with a set of labels; programmatically training a machine learning neural Teacher model on the first labeled training dataset; programmatically applying a machine learning model trained for NLP to an unlabeled dataset comprising a digital electronic representation of natural language text summaries of a second plurality of images, thereby generating a second labeled training dataset comprising the second plurality of images; programmatically generating soft pseudo-labels using the machine learning neural Teacher model; programmatically generating derived labels using the soft pseudo-labels; training one or more machine learning neural Student models using the derived labels; receiving a target image; applying an ensemble of the one or more Student models to output one or more classifications of the target image.

The application discloses a kind of training method, device, electronic equipment and storage medium of image segmentation model.The method comprises: obtaining multiple groups of training sample data, and training sample data includes sample image;In the iterative training process of image segmentation model, sample image is input to the encoder in image segmentation model, and the encoding feature image is obtained, the encoding feature image is input to the decoder in image segmentation model, and the first reconstruction image is obtained, the encoding feature image is input to the variational autoencoder in image segmentation model, and the second reconstruction image is obtained;Determine segmentation model loss based on the first reconstruction image, the second reconstruction image and sample image, adjust the parameter of current image segmentation model based on segmentation model loss, and obtain target image segmentation model.The above technical solution realizes self-supervised learning by increasing variational autoencoder, realizes unlabelled data training, reduces dependence on labeled data, and reduces data labeling cost.

The invention discloses a gait identity recognition method based on a generative self-supervised visual pre-training model BEiT and DAS technology. The method comprises the following steps: firstly, carrying out scene construction and data preparation, collecting label-free data and label data, extracting data signals for signal preprocessing, and constructing a time frequency characteristic data set; constructing a pre-training image reconstruction network of a self-supervised visual pre-training model BEiT, performing feature extraction on an input image, constructing a deep learning network based on the self-supervised visual pre-training model BEiT as a feature extractor, and training the deep learning network by using unlabeled spatio-temporal data in an experimental scene to obtain weight parameters; and constructing a downstream classification task network, performing classification learning of downstream tasks by using a small amount of data with labels to obtain a new weight, and completing identification and classification of non-label data by using the obtained new weight. According to the method, the demand quantity of the tagged data is reduced, and the accuracy is ensured.

The application discloses a wall heat flow prediction method considering material ablation response under a semi-supervised learning framework, and relates to the field of aerospace. The method comprises the following steps: constructing a data set; building a neural network model and introducing an ablation physical information loss term into a loss function of the neural network model; pre-training the neural network model; predicting unlabelled data through the pre-trained neural network model to generate pseudo labels; mixing labelled data and unlabelled data to semi-supervise and train the pre-trained neural network model to obtain a final neural network model; and predicting wall heat flow of material ablation response through the final neural network model. The method reduces the demand of the model on the labelled data set, enhances the model generalization ability under the condition of small sample training, increases the physical interpretability of the network model, and improves the wall heat flow prediction precision.

The invention relates to a computational lithography modeling and hot spot detection method and device based on self-supervised learning, and the method comprises the steps: constructing a neural network model which comprises a layout feature encoder, a geometric reconstruction branch and a lithography space image prediction branch, and enabling a differentiable optical simulation layer to be built in the lithography space image prediction branch so as to integrate physical prior; after an input layout is masked based on label-free mask plate graph data, the model is driven to execute pre-training of geometric reconstruction and photoetching space image prediction at the same time, an original unmasked layout is processed through the differentiable optical simulation layer to generate a pseudo-true value, and a supervision signal is provided for photoetching space image prediction; and on the basis of the pre-training model, performing fine tuning by using a small amount of labeled data to realize hotspot detection. The differentiable optical simulation layer is integrated into the self-supervised learning framework, so that the model can learn the optical law from massive label-free data, the dependence on labeled data is reduced, and the hotspot detection capability and accuracy of unknown graphs are improved.

The invention discloses a rotating machine fault classification method based on semi-supervised transfer learning, and relates to the field of fault diagnosis, and the method comprises the following steps: S1, collecting a vibration signal containing transient noise and a small number of fault labels; s2, positioning and removing transient noise by using a first-order Markov model, and dividing a 80% training set and a 20% test set; s3, building a semi-supervised transfer learning model; s4, training a semi-supervised transfer learning model; s5, the model is tested, and it needs to be ensured that the accuracy rate is larger than or equal to 95% under the label rate of 5 And S6, performing online application. According to the method, semi-supervised and transfer learning are fused, the non-label data value is effectively mined, transient noise is suppressed, the blank of low-label and high-interference working condition diagnosis is filled, a data driving method is promoted to move from a laboratory to an industrial complex scene, and key technical support is provided for field improvement of fault detection accuracy and reliability and predictive maintenance upgrading.

One aspect of the invention provides a computer-based method for providing corrective feedback about exercise form, the method comprising; recording a user performing a specific exercise: evaluating, by the computer, with machine learning, computer vision, or deep learning models that have been previously trained in order to evaluate the form of a user by training on labelled and or unlabeled datasets that consist of: both correct and incorrect exercise form for the different types of exercises being evaluated; identifying the user throughout the video, the exercise type, each repetition of the exercise, the user's errors in form; and then generating, by the computer, corrective feedback for the user on how to improve exercise form for subsequent repetitions; and communicating, via an output device, the corrective feedback to the user.

A computer-implemented method providing a neural network for identifying radio technologies employed in an environment. The neural network includes an autoencoder having an encoder, and a classifier. The method has the steps of sensing a radio spectrum of the environment thereby obtaining a set of data samples, labelling a subset of the data samples by a respective radio technology thereby obtaining labelled data samples, training the autoencoder in an unsupervised way by unlabelled data samples, training the classifier in a supervised way by the labelled data samples, and providing the neural network by coupling the output of an encoder network of the autoencoder to an input of the classifier.

A computer-implemented method of neural architecture search. The method includes: training a supermodel by sampling an architecture and training the supermodel with the sampled architecture on labeled training data and updating weights of the supermodel by gradients with respect to the sampled models; determining Pareto-optimal submodels of the supermodel based on at least two performance metrics by iteratively carrying out the following steps: computing outputs of a reference model for the unlabeled data, wherein the reference model a largest submodel of the supermodel; sampling a plurality of submodels from the supermodel; computing by the submodels their outputs of the unlabeled data; computing a difference between the outputs of the reference model and the submodels; employing an optimization algorithm to iteratively sample and evaluate submodels based on a plurality of objectives.

The invention belongs to the technical field of hydrogen safety, and discloses a hydrogen leakage positioning method, device, equipment and medium, the method comprises the following steps: carrying out Fourier transform on unlabeled multi-source hydrogen concentration sensor sequence data to obtain frequency domain characteristics, and respectively enhancing time domain data and frequency domain data; time-frequency domain characterization is extracted through a Transform double encoder, and comparison loss is constructed; mapping to a joint space through a projection network, and obtaining a final time-frequency representation through triple comparison loss optimization; and inputting BP neural network training to realize positioning. According to the method, high generalization features are extracted from label-free data, dependence on labeled data is reduced, the weak signal detection capability is improved by fusing time-frequency domain features, and a leakage source can be quickly positioned.

The application discloses an active anomaly detection method of data independent subset driving multi-expert hybrid training. Through active learning feedback in each round, samples are divided into two parts according to the setting of a threshold value according to the ranking of an anomaly score, and multiple sample subsets are divided based on the similarity and diversity of features. A plurality of weak expert groups are independently trained on each subset in the form of an expert hybrid network, reasoning is performed by means of integration, and the most abnormal samples are cross-selected on each expert network, so that the interference caused by the single-class setting of unlabelled data to the model is reduced. The application specifically applies a multi-task expert hybrid network model to single-task anomaly detection, intelligently learns the weight of integration and combination, strengthens the connection between data and expert networks, improves the detection performance, and provides a new idea for semi-supervised anomaly detection.

A radar target detection method and device based on cross-view cross-modal contrast learning and a medium are provided.The method constructs a self-supervised learning framework for radar target detection based on a distance-azimuth-Doppler (RAD) tensor, captures the relationship between different views of the RAD cube through cross-view contrast learning, enables the model to obtain spatial distribution knowledge from multiple views, and captures the relationship between the RAD and a visual modality through cross-modal contrast learning, enabling the model to obtain semantic knowledge from the visual modality.The present application combines cross-view and cross-modal contrast learning, enhances the feature representation capability of the radar, allows the RAD model to learn transferable feature representations from unlabelled data, and significantly improves the accuracy of radar target detection downstream tasks in the case of limited labelled data.

Broadly speaking, the present techniques provide an automatic way of classifying data items within an environment (e.g. a business, workplace, organisation, etc.), and applying data management policies based on the classifications. This is advantageous over existing techniques which require manual classification of data items, which is time consuming in environments where hundreds of new data items may be generated in a day or week. The present techniques use an embedding machine learning, ML, model to automatically determine the relevant classification label(s) for an unlabelled data item, which is then used to select and apply the relevant data management policy(ies).

The application discloses a kind of data and physical alternation driven semi-supervised seismic wave impedance inversion method, belong to petroleum geophysical exploration field.The method includes: obtaining the poststack seismic record of research area and well impedance curve, constructs label data set and unlabelled data set;Establish inversion and forward network and carry out preheating training;Freeze network parameter and predict initial model, combined with physical forward operator and physical constraint term iterative optimization generates physical pseudo label;Unfreeze network parameter, introduce physical pseudo label into semi-supervised joint update training of inversion and forward network;Cyclically alternately execute physical pseudo label generation and semi-supervised joint update training until meeting stopping criterion, obtain trained inversion network;Finally, the poststack seismic record of research area is input into the network, and the wave impedance inversion result is output.The application overcomes the problem of insufficient physical constraint strength in semi-supervised inversion, and can still achieve high-precision wave impedance inversion under the condition of limited well labels.

The present application relates to an unsupervised heterophilic edge graph analysis model using an edge discriminator and a multi-channel encoder, and a learning method using the same, and generates a feature information-based representation, a connection information-based representation and a weighted graph-based representation. Then, these representations are combined to generate a final node representation vector, and the consistency of the representation is enhanced through contrastive learning. Compared with the conventional single channel model, the node representation power is increased, which leads to excellent node classification accuracy. In addition, the model of the present application may be applied to unlabeled datasets through unsupervised learning, and particularly exhibits excellent performance on heterophilic edge graph. The present application may be applied to data analysis with complex relationships in the fields such as social network analysis, recommendation systems, and bioinformatics, and may be utilized for various graph-based tasks such as anomaly detection and link prediction.

The invention discloses an active learning training method and system for a three-dimensional point cloud segmentation model, and a storage medium. The method comprises the following steps: providing a basic segmentation model, a basic probability model and a training data set; performing supervised pre-training to obtain a pre-training model; extracting embedded features for fitting training to obtain a joint probability model; calculating prototype characteristics; interpolation fusion is carried out to obtain fusion features, and joint probability representation is generated; calculating a stability score of the unlabeled data based on the fluctuation degree; performing redundancy removal and stability screening on the unlabeled data to obtain to-be-labeled data; receiving a marking operation on the to-be-marked data to form new marked data; performing semi-supervised training on the pre-training model by using the updated training data set to obtain an updated segmentation model; and iteration training is completed. According to the method, high-value samples are screened for labeling by constructing a joint probability modeling mechanism, and the discrimination capability and generalization performance of the model can be efficiently improved under limited labeling budget.

The invention discloses a physical information neural network video flow measurement method and device based on label-free data in the technical field of hydrological information monitoring, and aims to solve the technical problem that an existing flow measurement method depends on label data and lacks physical basis. The method comprises the following steps: carrying out frame cutting on a river video, and converting world coordinates of a velocity measurement point into image pixel coordinates through projection transformation; selecting an image area and converting the image area into an aerial view to obtain aerial view coordinates of a speed measurement point; calculating an image gray scale partial derivative based on the aerial view sequence, inputting the image gray scale partial derivative to a physical information neural network model, and outputting an inter-frame speed of each speed measurement point; converting the speed into the actual flow speed, further calculating the vertical line average flow speed, and combining section information to obtain the total flow. According to the method, label data are not needed, the network is constrained by a physical equation, non-contact flow measurement is realized, and the method has high efficiency, safety and physical interpretability.

The application provides a concrete crack identification method based on a synthetic data set and semi-supervised learning, 1) an original data set is collected, and a certain mainstream neural network model is selected for supervised learning; 2) a benchmark identification model is updated based on a semi-supervised learning method; 3) unlabelled data in a personal data set is classified based on the updated benchmark identification model; 4) an image with higher confidence is selected, pixel-level mask annotation is performed on the image, and a crack segmentation model is determined; 5) the crack segmentation model is used for pixel-level annotation of the image with higher confidence; 6) based on the mask annotation result of the crack, a skeleton line of the crack is obtained, the maximum width and the average width are calculated, and error analysis of the crack width is performed; 7) the concrete crack identification model is verified; 8) the shooting result is imported into the concrete crack identification model, and the skeleton line, the maximum width and the average width of each crack are obtained.

The application discloses a kind of mask image modeling guide domain adaptation cross-domain remote sensing scene classification method, comprising: constructing domain adaptation network, the self-encoding of unlabelled self-supervised pre-training is carried out, and the model parameter of pre-training self-encoding is obtained;Self-encoding model parameters are loaded into self-encoder, data are input into domain adaptation network, and the mask image modeling of source domain image and target domain image is carried out using self-encoder;High-level semantic feature distribution of source domain and target domain is aligned using feature adaptation module;And based on data, overall target loss function is constructed, and the iterative training of domain adaptation network is carried out by optimizing overall target loss function, the decoder part of self-encoder is removed, the encoder of self-encoder and feature adaptation module are used to test target domain image, and good scene classification result is obtained.The application preserves domain-specific features in the process of extracting domain-invariant features, and further improves the classification generalization ability for unlabelled data target domain.

The application discloses a kind of semi-supervised semantic segmentation method, equipment and its storage medium.Establish picture data set by obtaining picture data in real world data, build depth model and semi-supervised semantic segmentation model, predict to obtain deep information and semantic segmentation information, preset threshold, filter out pixel area in any two categories in single picture by preset threshold, obtain depth set and logarithm difference set, obtain intra-class logarithm difference loss according to depth set and logarithm difference set, calculate the regularization loss of logarithm difference;Obtain unlabelled data, establish strong enhanced view set and weak enhanced view set, obtain robust supervision signal;Exponential normalization is used to suppress large abnormal fluctuations for intra-class depth difference, and weight is adaptively assigned based on entropy, so as to promote robust feature learning and learn rich discriminative information.The application promotes prediction consistency to maximize the use of unlabelled data and further improve model performance.

The application discloses a kind of based on pseudo label empowerment's elastic memory drift detection and hot updating method and system, method includes: obtaining new sample in data stream, to unlabelled sample in new sample, predict its class as pseudo label using current prediction model parameter, simultaneously calculate the confidence of prediction model prediction, and as the original confidence weight of this pseudo label;The new sample of empowerment weight is stored to elastic memory module;Based on the difference between historical data distribution and new received data distribution in elastic memory module, determine drift detection index;When drift detection index exceeds the drift threshold of adaptive change, determine that concept drift occurs;When detecting concept drift, trigger online prediction model update;After training convergence, seamlessly replace the prediction model of current online operation with updated prediction model.The application can efficiently utilize unlabelled data and timely detect and adapt to concept drift model update.

The application discloses a kind of main shaft cutter wear monitoring methods based on physical guidance and semi-supervised learning, comprising: collecting cutter acceleration signal, extracting multi-domain statistical characteristics and carrying out cross-cutter robustness screening, construct the cumulative characteristics of reflection historical trend and form hybrid input set;Explicit physical equation conforming to degradation mechanism is mined using parallel symbolic regression network with monotonicity penalty introduced;Predict and smooth monotonicity processing are carried out on unlabelled data using the equation, generate physical weak label, and build hybrid training set;Design physical guidance cross attention mechanism, weight fusion is carried out to statistical and cumulative characteristics with physical equation as priori;Lightweight neural network is constructed to process fusion characteristics, and differential loss function is designed based on hybrid training set Joint training is carried out.The application can break through the dependence on a large number of labeled data, ensure the physical consistency of prediction results, and meet the high-precision monitoring demand of industrial field.

The application discloses a kind of weak semi-supervised cardiac MRI segmentation methods based on pseudo-label frequency domain dynamic hybrid supervision, belong to medical image segmentation technical field, including the following steps: S1, a small amount of cardiac MRI image in training set is provided with scribble label;S2, establish main segmentation model and auxiliary segmentation model, output corresponding predicted value;S3, generate supervision pseudo-label using Haar wavelet transform and frequency domain dynamic hybrid method;S4, combined with partial cross-entropy loss constraint strategy establishes total loss function to optimize training three segmentation models, realize the segmentation of cardiac MRI image.The application generates reliable supervision signal by pseudo-label frequency domain dynamic hybrid supervision method, significantly improves the utilization efficiency of unmarked data using a small amount of scribble label and a large amount of unlabelled data;Using weak semi-supervised strategy, only a small amount of scribble label can realize high-quality segmentation, reduce the dependence on pixel-by-pixel accurate annotation, solve the problem of high medical image annotation cost.

A method of training a supervised machine learning system to detect anomalies within transaction data is described. The method includes obtaining a training set of data samples; assigning a label indicating an absence of an anomaly to unlabelled data samples in the training set; partitioning the data of the data samples in the training set into two feature sets, a first feature set representing observable features and a second feature set representing context features; generating synthetic data samples by combining features from the two feature sets that respectively relate to two different uniquely identifiable entities; assigning a label indicating a presence of an anomaly to the synthetic data samples; augmenting the training set with the synthetic data samples; and training a supervised machine learning system with the augmented training set and the assigned labels.

Providing unlabelled training data for training a computational model comprises:obtaining sets of time-aligned unlabelled data, wherein the sets correspond to different ones of a plurality of sensors;marking a first sample, of a first set of the sets, as a positive sample, in dependence on statistical separation information indicating a first statistical similarity of at least a portion of the first set to the at least a portion of the reference set and in dependence on the first sample being time-aligned relative to a reference time;marking a second sample, of a second set of the sets, as a negative sample, in dependence on statistical separation information indicating a second, lower statistical similarity, of at least a portion of the second set to the at least a portion of the reference set, and in dependence on the second sample being time-misaligned relative to the reference time.

The application discloses a kind of semi-supervised field self-adaptive wafer defect identification method and system, obtain the data set of wafer defect source field and target field;And the data set is preprocessed;Cross alignment network is constructed, and the image features of source field and target field of the data set after pre-processing are respectively extracted using feature extractor and encoder;For the extracted image feature, output probability is obtained using classifier, for unlabelled data, determine and update the pseudo-label threshold of each class by dynamic threshold strategy, then use sample-by-sample cross alignment, class-by-class cross alignment and consistency regularization method to calculate loss function and train, obtain the trained cross alignment network;Using the trained cross alignment network as prediction model, the wafer defect class of the unlabelled data of target field is output to be identified.The present application can effectively realize the knowledge transfer between different data sets, and achieve high recognition accuracy in the case that the target domain data set has only a small amount of labels, such as only one label per class.

Embodiments provide methods and systems for processing unstructured and unlabelled data. A method includes generating, by a processor, a structured and unlabelled training dataset from an unstructured and unlabelled dataset. The method includes categorizing the structured and unlabelled training dataset into a plurality of clusters by executing an unsupervised algorithm. Each cluster of a selected set of clusters from the plurality of clusters is labelled with an applicable label from a set of labels. The method includes executing a supervised algorithm to generate a trained supervised model using a labelled training dataset including the set of labels and an input dataset generated from plurality of datapoints present in each cluster of the selected set of clusters. The method includes generating a Labelled Data1 (LD1) by executing the trained supervised model configured to assign applicable label from the set of labels to each datapoint of the structured and unlabelled training dataset.