178 results about "Neural network classifier" patented technology

The embodiment of the invention provides a charging service real-time monitoring method and device based on a service probe, and the method and device achieve the precise construction of a service event chain through constructing a multi-layer data analysis mechanism and integrating the data of an application layer, a session layer and a network layer. And designing an exception recognition strategy based on multi-dimensional feature fusion, and establishing a neural network classifier to perform exception event analysis in combination with hardware state features and scene features. A real-time monitoring engine is introduced, and dynamic early warning is carried out on business abnormity through hierarchical analysis and correlation analysis. According to the method, the defects of the traditional technology in the aspects of data analysis, feature analysis, real-time monitoring and the like are effectively overcome, and the intelligent level and the early warning effect of charging service monitoring are remarkably improved.

The invention discloses a layered energy management method of a fuel cell hybrid tramcar with self-adaptive working condition perception. In the recognition layer, a sliding window mechanism is adopted to extract time domain and frequency domain features of load conditions, feature data are clustered based on a spectral clustering algorithm driven by a deep auto-encoder, a data set with category labels is obtained, and a deep dynamic learning vector quantization neural network classifier is trained; in the strategy layer, a double-delay depth deterministic strategy gradient reinforcement learning algorithm is adopted, a reward function is constructed, and lithium battery SOC fluctuation penalty term limit parameters in the reward function are adaptively adjusted according to the real-time load working condition category output by the recognition layer; training the reinforcement learning agent to obtain an optimal power distribution scheme between the multi-stack fuel cell power generation system and the lithium battery; and according to the performance degradation degrees of different fuel cell stacks, a distributed cooperative control strategy considering performance difference is adopted to distribute the output power of each stack, so that the coordinated control of the running state of the multi-stack fuel cell power generation system is realized.

Systems and techniques are provided for processing one or more data samples. For example, a neural network classifier can be trained to perform small sample open set recognition (FSOSR) based on a task agnostic open set prototype. The process can include determining one or more prototype representations for each class included in the plurality of support samples. A task-agnostic open set prototype representation can be determined in the same learning metric space as the one or more prototype representations. Based on the one or more prototype representations and the task-agnostic open set prototype representation, one or more distance metrics can be determined for each of the one or more query samples. Based on the one or more distance metrics, each query sample can be classified into one of the classes associated with the one or more prototype representations or into an open-set class associated with the task-agnostic open-set prototype representation.

The invention discloses a high-strength steel welding defect nondestructive detection and identification method based on multi-modal data fusion. The method comprises the following steps: welding data acquisition: acquiring a two-dimensional image and three-dimensional point cloud data of a high-strength steel welding part to form a data pair; performing data space alignment: generating a space-aligned image-depth map data pair; feature extraction and fusion: obtaining fusion features with spatial geometric information and two-dimensional visual information through feature extraction and fusion; defect identification and classification: identifying defect pixels, decoding and recovering space information of a defect area, calculating the three-dimensional size of the defect, taking the feature information associated with a connected domain of each pixel with the defect and the three-dimensional size as input, and automatically classifying defect categories through a pre-trained full-connection neural network classifier. According to the method, the welding defects of the high-strength steel can be accurately identified and accurately and quantitatively analyzed.

Disclosed is a neural network enabled interface server and blockchain interface establishing a blockchain network implementing event detection, tracking and management for rule based compliance, with significant implications for anomaly detection, resolution and safety and compliance reporting.

The invention discloses an efficient intelligent network attack classification tool based on a federated learning framework, which comprises a central server and a plurality of end-side nodes, local models are deployed in each end-side node, a central model is deployed in the central server, and the models comprise multi-task logistic regression, random forest, XGBoost and long and short-term memory neural network classifiers; the local model collects local network flow data, performs model training and abnormal data detection and classification in combination with a D-S evidence theory optimization algorithm and a four-class complementary classifier, and transmits encrypted abnormal data and related key update parameters to the central server; the central model receives and aggregates the updated parameters, performs attack type classification based on abnormal data in combination with a D-S evidence theory optimization algorithm after updating the global model, and re-allocates a classification result and the updated global model to each end side node; the method has high robustness and high efficiency, the adaptability is enhanced, and the recognition capability of attack scenes is improved.

The invention relates to a risk identification method and system. The method comprises the following steps: acquiring communication behaviors, equipment fingerprints and service interaction data in real time through a privacy compliance interface; carrying out parallel preprocessing and safe desensitization on the data; inputting a feature construction engine to extract a communication mode, an equipment behavior and an interactive semantic feature vector in parallel, constructing a dynamic weighted hypergraph communication map, and outputting a social risk feature vector by using a time sequence hypergraph neural network; inputting the four types of features into a privacy perception multi-mode gating attention fusion module to output fusion features; generating a risk score through a deep neural network classifier; and monitoring an abnormal event based on Apache Flink, adaptively adjusting a decision boundary in combination with the dynamic risk entropy, and triggering secondary verification. According to the method, the problems of multi-modal data conflict, privacy disclosure and decision stiffness are solved, and the recognition precision and the real-time performance are improved.

The invention discloses an expandable category guide anomaly detection method and device for multiple categories of targets, and relates to the field of computer vision. The method comprises the following steps: carrying out abnormal region guided adaptive enhancement preprocessing on an industrial image to obtain a target image; judging the category of the target image according to a pre-constructed lightweight neural network classifier; activating at least one anomaly detection model according to the category of the target image; and performing anomaly prediction on the target image by using the activated anomaly detection model, and fusing with the confidence corresponding to the anomaly detection model to realize adaptive anomaly discrimination of the industrial image. According to the method, the sample category is quickly judged through the lightweight classifier, the pre-screening and path guidance of the anomaly detection model are realized, the category guidance weight is generated by using a confidence coefficient mechanism, the subsequent model fusion strategy is endowed with higher adaptability, and the structural clarity, the model selection accuracy and the overall calculation efficiency of the system are effectively improved.

The present invention belongs to the technical field of mechanical fault diagnosis and signal processing, and specifically relates to a bearing fault diagnosis method, system, equipment and medium based on a multi-channel analog filter feature network. The method constructs a multi-channel analog filter feature extraction network, uses multiple parallel bandpass filter channels to extract bearing vibration signal characteristics, and combines an analog neural network classifier to perform fault classification. Signal preprocessing includes removing DC components, amplitude normalization and segmented processing; the filter passband parameters and classifier parameters are jointly optimized through a particle swarm algorithm, with classification accuracy and decision confidence as fitness functions. The system includes signal acquisition, preprocessing, multi-channel filtering, feature extraction, analog neural network classification and result display modules, achieving low-power, high-real-time bearing fault diagnosis. The present invention solves the problems of high power consumption and large delay of traditional digital processing methods, and is suitable for long-term monitoring and large-scale deployment in industrial sites.

According to various embodiments, a method for encrypting image data for a neural network are disclosed. The method includes mixing the image data with other datapoints to form mixed data; and applying a pixel-wise random mask to the mixed data to form encrypted data. According to various embodiments, a method for encrypting text data for a neural network for natural language processing is disclosed. The method includes encoding each text datapoint via a pretrained text encoder to form encoded datapoints; mixing the encoded datapoints with other encoded datapoints to form mixed data; applying a random mask to the mixed data to form encrypted data; and incorporating the encrypted data into training a classifier of the neural network and fine-tuning the text encoder.

Among other techniques, techniques for machine learning-based entity resolution are described. An example method includes receiving an entity resolution request, the entity resolution request indicating a first entity and a second entity; identifying a plurality of first attributes in a data model; identifying a plurality of second attributes in the data model; creating a first string based on the plurality of first attributes of the data model; creating a second string based on the plurality of second attributes of the data model; generating a first prompt based on the first string; generating a second prompt based on the second string; providing the first prompt to a domain-agnostic large language model; generating, by the domain-agnostic large language model using the first prompt, a first domain-agnostic large language model result; clipping the first domain-agnostic large language model result; providing the second prompt to the domain-agnostic large language model; generating, by the domain-agnostic large language model using the second prompt, a second domain-agnostic large language model result; clipping the second domain-agnostic large language model result; generating, by a downstream neural network classifier, a machine learning final result based on the clipped first and second domain-agnostic large language model result; and merging, based on the machine learning final result, the first entity and the second entity.

The invention relates to the technical field of power equipment monitoring, and discloses an abnormal discharge electric field feature extraction method. The method comprises the following steps: standardizing and normalizing data acquired from a multi-channel electric field sensor array of a transformer substation to generate a standardized data stream; performing wavelet decomposition, adaptive threshold noise reduction and reconstruction on the basis of the stream to obtain a denoised signal; time domain statistics and frequency domain energy spectrum analysis are carried out on the signals, features are fused, and a joint feature vector is constructed; training a convolutional neural network classifier based on the vector; and updating a dynamic threshold rule according to the classifier output probability, and finally realizing real-time early warning and model feedback through GPU parallel computing. According to the method, the signal-to-noise ratio and the feature expression capability of the abnormal discharge signal can be effectively improved, and the detection accuracy, the real-time performance and the self-adaptive capability of the system are enhanced.

The invention discloses a traffic sign defect detection method and system based on SAM and Transform and a medium, and the method comprises the steps: collecting an image of a traffic sign, carrying out the preprocessing, recognizing the traffic sign in the image through a YOLO target detection model, generating a detection frame of the traffic sign, and framing a region where the traffic sign is located; inputting the detection frame and the corresponding image into an SAM model, and extracting the contour of the traffic sign; and simultaneously inputting the extracted contours of the traffic signs and contours of the traffic signs in a normal state in the database into a neural network classifier based on Transform for comparative analysis, finally outputting a defect detection result, and giving a specific defect type and a confidence score. According to the method, the YOLO target detection, the SAM model and the neural network classifier based on the Transform are combined, and the task of each stage can be optimized on the basis of the result of the previous stage, so that the final detection precision and efficiency are improved.

A method and system of image hashing object detection for image processing are provided. The method comprises the following steps: obtaining image head class input data and image tail class input data differentiated from the head class input data and respectively of two images each of an object to be classified; respectively inputting the head and tail class input data into two separate parallel representation neural networks being trained to respectively generate head and tail features, wherein the representation neural networks share at least some representation weights used to form the head and tail features; inputting the head and tail features into at least one classifier neural network to generate class-related data; generating a class-balanced loss of at least one of the classes of the class-related data comprising factoring an effective number of samples of individual classes; and rebalancing an output sample distribution among the classes at the representation neural networks, classifier neural networks, or both by using the class-balanced loss.

The invention belongs to the technical field of mechanical fault diagnosis and signal processing, and particularly relates to a bearing fault diagnosis method, system and equipment based on a multi-channel analog filtering feature network and a medium. The method comprises the steps that a multi-channel analog filtering feature extraction network is constructed, and bearing vibration signal features are extracted through multiple band-pass filter channels arranged in parallel; and fault classification is carried out in combination with a simulated neural network classifier. The signal preprocessing comprises direct current component removal, amplitude normalization and segmentation processing; filter passband parameters and classifier parameters are jointly optimized through a particle swarm algorithm, and classification accuracy and decision confidence are used as fitness functions. The system comprises a signal acquisition module, a preprocessing module, a multi-channel filtering module, a feature extraction module, a simulation neural network classification module and a result display module, and realizes bearing fault diagnosis with low power consumption and high real-time performance. The method solves the problems that a traditional digital processing method is high in power consumption and large in delay, and is suitable for long-term monitoring and large-scale deployment of an industrial site.

The invention relates to the technical field of signal target identification, in particular to a multi-mode aviation sonar buoy signal processing and target identification method, which comprises the following steps of: performing signal acquisition in a preset sea area range by using an aviation sonar buoy; the collected multi-mode signals are preprocessed; performing feature extraction on the preprocessed multi-mode signal to generate a unified feature vector; the fused feature vectors are input into a pre-trained target recognition model, and the model adopts a hybrid neural network structure, comprises a convolutional layer and a recurrent neural network layer and is used for processing time sequence signal features; and processing a multi-mode signal input in real time based on the trained target identification model to realize identification and classification of a sea area target, and optimizing an identification result by combining a support vector machine and a classifier of a neural network. According to the method, the continuous capture of the signal sequential relationship is ensured, so that the system can still keep efficient identification and tracking capabilities in a dynamic environment.

The invention provides a physical constraint and adversarial network fused equipment fault prediction method, and relates to the field of equipment fault prediction. According to the method, firstly, multi-source heterogeneous modal information of target equipment is fused, equipment operation parameters and operation and maintenance logs are obtained, and an equipment fault reasoning text corresponding to the operation and maintenance logs is generated through a large language model; secondly, introducing physical constraint information into time sequence modeling; thirdly, based on a multi-modal fusion type generator structure of residual fusion and a multi-head potential attention mechanism, multi-modal feature fusion and a fault sample generated through confrontation are obtained; and finally, predicting a target equipment fault risk of a future time window by using a convolutional neural network (CNN) classifier. According to the method, multi-modal information is fused, a physical constraint and sample generation mechanism is introduced, the problems of fault sample scarcity and multi-factor coupling modeling are effectively relieved, and the precision and robustness of equipment fault prediction are improved.

Example embodiments monitor a dwelling or set of dwellings. In an example embodiment, a neural network classifier is applied to first audio data to determine a first confidence level that the first audio data represents a monitored apparatus in a first state. A first population spectral signal is obtained based on a first population of audio signals classified as representing the monitored apparatus in the first state. A first measured spectral signal is obtained from the first audio data. A first distance measure is calculated between the first measured spectral signal and the first population spectral signal. A determination is made of whether the monitored apparatus is in the first state based on the first confidence level and on the first distance measure. State changes of the apparatus may be monitored, and an alert may be issued if abnormal state changes are detected.

The invention discloses a personnel state detection system and method based on a wifi router, and the method comprises the following steps: S1, deploying the wifi router, collecting subcarrier CSI which changes due to the disturbance of the movement of personnel, and forming a CSI signal sequence; s2, configuring a neural network classifier and loading parameters; s3, preprocessing the CSI signal sequence to obtain a stable CSI time sequence signal; s4, constructing a graph structure taking the subcarriers as graph nodes, and generating a spectrogram; s5, performing Laplace feature transformation on the spectrogram, and extracting a frequency domain feature vector of the spectrogram; s6, constructing a multi-scale Sheng differential equation model, and generating a state trajectory under each time scale; and S7, splicing the state tracks to form a state evolution sequence, inputting the state evolution sequence into a neural network classifier, and generating a state recognition result. According to the method, CSI is used for modeling personnel disturbance, multi-scale dynamic representation and accurate recognition are achieved, and the method is suitable for a non-contact personnel state sensing scene.

Automated techniques provide for recalibrating cameras configured for monitoring an area of real space. The method includes first processing one or more selected images selected from a plurality of sequences of images received from a plurality of cameras calibrated using a set of calibration images that were used to calibrate the cameras previously. Transformation information between the selected images and the set of calibration images is obtained based on one or more features extracted from the selected images using a trained neural network classifier corresponding to points located at relatively immobile structures. The features extracted from the selected images match features in the set of calibration images. Camera calibrations can be updated when transform information between features matched meets or exceeds a threshold.

There is provided a computer-implemented method for verifying the robustness of a neural network classifier with respect to one or more parameterised transformations applied to an input, the classifier comprising one or more convolutional layers, the method comprising: encoding each layer of the classifier as one or more algebraic classifier constraints; encoding each transformation as one or more algebraic transformation constraints; encoding a change in an output classifier label from the classifier as an algebraic output constraint; determining whether a solution exists which satisfies the classifier constraints, transformation constraints and output constraints, and determining the classifier as robust to the local transformations if no such solution exists. A perception system and a computer readable medium are also provided.

The invention relates to the technical field of multi-target tracking, and particularly discloses a target track prediction method and system in a dense clutter environment, and the method comprises the steps: S1, obtaining an initial measurement value of a sensor and a preset parameter used for limiting the size of a hypothesis tree, and generating the hypothesis tree according to the initial measurement value, the preset parameter and a multi-hypothesis tracking algorithm; assuming that each leaf node of the tree corresponds to a candidate track; s2, each leaf node is preprocessed, and non-repeated candidate tracks are obtained; s3, the candidate tracks are divided into correct tracks and wrong tracks through a neural network classifier, and the number of targets and initial tracks are determined according to the correct tracks; s4, tracking the target according to the initial track, and performing adaptive dynamic adjustment on the number of hypotheses generated in each step in the target tracking process through a multi-hypothesis tracking algorithm of adaptive parameters to obtain hypothesis data associated with the target; and S5, estimating motion parameters of the target through a Kalman filter.

The invention discloses a fan starting condition intelligent diagnosis method and system based on multi-modal data fusion. The fan starting condition intelligent diagnosis method comprises the steps of S1, collecting fan starting data through multi-modal sensing data to construct time sequence original data; s2, performing pre-optimization processing on the time sequence original data by using a sound and vibration alignment mechanism, and respectively extracting a first time sequence feature and a second time sequence feature through a double-flow lightweight FNet branch; s3, dynamically aligning the time-frequency physical association between the first time sequence feature and the second time sequence feature by adopting cross attention, and outputting a deep joint feature; and S4, collecting an abnormal feature sample, splicing the combined depth combined feature and the abnormal feature sample, inputting the spliced deep combined feature and the abnormal feature sample into a lightweight neural network classifier, outputting a fan fault confidence coefficient, and judging whether the fan is started or not according to the fan fault confidence coefficient and a preset confidence coefficient threshold value.

Disclosed is a neural network enabled interface server and blockchain interface establishing a blockchain network implementing event detection, tracking and management for rule based compliance, with significant implications for anomaly detection, resolution and safety and compliance reporting.

A physiological monitor has an auxiliary sensor and a processor, the processor configured to read the auxiliary sensor and determine when the monitor is in a stable position on a wearer. Alternatively, analog circuitry is provided to determine when the monitor is in a stable position on the wearer. When in an unstable position on the wearer, the monitor enters a low power state. Once in a stable position on the wearer, the processor goes to a full power state and detects physiological events with a primary sensor and logs these events with timestamps. A method is disclosed of determining when the monitor is in a stable position on a subject by reading signals from an auxiliary sensor and classifying the signals from the auxiliary sensor with a classifier selected from the group consisting of a decision tree classifier, a KNN classifier, and a neural network classifier.

The invention relates to the technical field of gesture recognition interaction, and discloses an intelligent interaction system and method based on gesture recognition. The method comprises the following steps: when an environment meets a condition, synchronously acquiring an image sequence of a gesture and depth distance field data; a hand joint point two-dimensional motion track is extracted from an image, surface deformation fluctuation information is separated from depth data, the two-dimensional motion track and the surface deformation fluctuation information are subjected to space-time registration fusion, a continuous motion track curved surface in a three-dimensional space is constructed, and then geometric topology features and dynamic change features of the continuous motion track curved surface are extracted. And inputting the fusion features into a neural network classifier subjected to incremental learning training, outputting corresponding semantic tags and confidence evaluation values, and mapping to generate a control instruction after verification. According to the method, an accurate three-dimensional dynamic model is constructed through deep fusion of multi-source data, and an incremental learning mechanism is utilized to enable the system to have online adaptive capability, so that the recognition precision of complex gestures and the long-term applicability of the system are improved.

The invention discloses a terrain classification method based on audio-visual information fusion. The method first controls a vehicle to travel on different terrains, uses a forward-facing camera mounted on the bottom of the vehicle to obtain a terrain image when the vehicle is traveling, and uses a sound sensor to collect sound signals of the interaction between the vehicle and the ground during the driving process; preprocesses the visual image and the sound signal; then extracts the features of the visual image and the sound signal respectively, and the extracted image features include the color histogram features and local binary features of the image; fuses the visual features with the sound features and performs feature dimension reduction; then inputs the reduced feature vector and the original image into the designed fully connected neural network classifier and residual network classifier for classification training; the final classification decision result is obtained by weighting the outputs of the two classifiers, the fully connected neural network classifier and the residual network classifier. After the model is trained, the model is used for online terrain classification.

Embodiments of the present application provide a quantum neural network classifier training method and device, electronic equipment and medium. The scheme is as follows: obtaining a training data set and a to-be-trained classifier; for each training sample data, classifying the training sample data by using the to-be-trained classifier to obtain a first predicted label; calculating a first loss value of the to-be-trained classifier according to a sample label corresponding to each training sample data and the first predicted label; when the to-be-trained classifier has not converged, adjusting the classifier parameters based on the first loss value, and returning to execute the step of classifying each training sample data by using the to-be-trained classifier to obtain the first predicted label corresponding to the training sample data until the to-be-trained classifier converges at the current time. Through the technical scheme provided by the embodiments of the present application, the optimization of the quantum neural network classifier is realized, and the classification accuracy and attack resistance of the quantum neural network classifier are improved.

A system and method utilizes masked language models in order to provide data-type detection, such as (but not limited to) prediction of columnar headings. Two masked language models are pre-trained on example columnar text. One model predicts missing data at the entity level (e.g., masked entity names that may be made up of whole words), while the other predicts missing data at the character level (e.g., masked individual characters). The table with missing column headings is fed into both models, and the output is contextual word embeddings and contextual character embeddings. These results are merged, and then fed into a neural network classifier to then predict the column names.

The invention discloses an intelligent power grid power supply optimization method and device based on big data and a storage medium, and belongs to the field of power grid informationization. According to the method, original power consumption data of a target power grid within a period of time is collected, the original power consumption data is preprocessed, missing value compensation, error value correction, normalization and the like are included, and the original power consumption data are obtained; the method comprises the steps of obtaining high-quality power consumption data, improving a whale optimization algorithm through an ant lion rule, performing feature extraction on the high-quality power consumption data by adopting the improved whale optimization algorithm, identifying the power consumption data by adopting a neural network classifier to obtain power consumption abnormal information, and giving an alarm according to the power consumption abnormal information. The improved algorithm can effectively capture related information from power consumption data, the ability of the model to identify abnormal modes related to non-technical loss is enhanced, the whale optimization algorithm is improved through the ant lion rule, and the convergence speed, stability and global search ability of the whale optimization algorithm can be effectively enhanced.