106 results about "Feature sharing" patented technology

The invention provides a method and a system for monitoring and predicting the health state of a storage battery based on multi-modal feature fusion, and relates to the technical field of storage battery monitoring, and the method comprises the steps: obtaining multi-modal data through periodically collecting static data of the storage battery and collecting dynamic data of the storage battery in real time; and then the method is realized through the steps of data monitoring, shared bottom layer feature extraction, SOH prediction, RUL prediction, multi-target joint optimization and the like. According to the method, multi-modal feature data generated by the storage battery can be monitored and analyzed on line, a multi-modal feature fusion and multi-target joint prediction and optimization mechanism is introduced, and SOH and RUL are predicted by adopting independent branches based on an SOH prediction model and an RUL prediction model, so that the health state of the storage battery is predicted; according to the method, automatic analysis on multi-modal mass data of the battery can be realized, the workload is reduced, the analysis efficiency is improved, the accuracy of a prediction result can be improved, and meanwhile, a constructed feature sharing mechanism can reduce the calculation complexity and improve the prediction efficiency.

The invention relates to a multi-element load combined prediction method fusing a decomposition reconstruction technology and multi-task learning, and the method comprises the steps: firstly obtaining a historical load sequence of an electric load, a cold load and a heat load and a data sequence of influence factor characteristics, and forming a historical multi-element load sequence library and an influence factor characteristic library; performing preliminary decomposition on data in the historical multivariate load sequence library through a variational mode decomposition method, performing reconstruction by using a mixed characteristic evaluation method, and updating the historical multivariate load sequence library; the historical multivariate load sequence library and the influence factor feature library are remodeled into three-dimensional feature tensors, and a CNN-ECA-MMoE multi-task learning model is constructed and used for outputting feature sharing and feature extraction results; and constructing a CNN-ECA-MMoE-LSTM multi-task learning model, carrying out model training by using the three-dimensional feature tensor, and outputting prediction results of an electric load, a cold load and a thermal load through the trained model. According to the method, the prediction precision of the multi-element load is improved.

The invention discloses a high-precision image detection method for a micro-drill blade surface based on deep learning, and the method comprises the following steps: S1, collecting a visible light image and a structured light image of the micro-drill blade surface, and completing the image preprocessing; s2, performing spatial alignment on the image, executing cross-modal fusion, and generating a feature fusion tensor; s3, inputting the feature fusion tensor into a multi-scale residual backbone network, and extracting a hierarchical semantic feature set; s4, inputting the semantic feature set into three task branches of defect detection, region segmentation and type classification, and outputting a corresponding prediction result; s5, calculating a multi-task loss function, dynamically adjusting task branch weights, and optimizing a feature sharing structure; and S6, generating a detection report according to a prediction result, and outputting defect coordinates, a boundary contour, a type label and a confidence value. According to the method, multi-modal fusion, high-precision identification and structured output of the micro-drill blade surface are realized, and the accuracy, efficiency and automation level of defect detection are remarkably improved.

The invention relates to a multi-modal data fusion-based multi-target tracking method for a legged robot. The method comprises the following steps of S1, data acquisition and preprocessing; s2, extracting data features; s3, multi-channel feature fusion extraction: based on the high-dimensional feature representation obtained in the S2, fusing the data features of the data of different sensor channels to obtain a multi-modal feature map, and processing the multi-modal feature map through a multi-task detection head to obtain a target detection result; s4, multi-target tracking: fusing data features of different sensor channels by using a feature sharing module, and outputting the fused data features to a multi-target tracking module; and the multi-target tracking module predicts the motion trail of the target in combination with the state of the target in the previous frame. According to the multi-modal data fusion-based multi-target tracking method of the legged robot, the accuracy and robustness of subsequent target detection and tracking can be improved, and continuous and stable identification and tracking of a sensitive target in a complex high-dynamic scene are realized.

The invention discloses a real-time auricle identification and auricular point positioning method and system for multi-task feature sharing. An auricle identification and auricular point positioning combined model used in the method comprises an encoder and a decoder. The encoder includes a backbone network and a neck network. The decoder comprises a plurality of detection heads. A plurality of detection heads receive outputs of different layers of the neck network. The detection head comprises a target regression branch, a target classification branch and a feature point regression branch which are parallel. A plurality of detection heads share output features of the same encoder, and parallel target regression branches, target classification branches and feature point regression branches in the detection heads are respectively used for calculating loss functions of different tasks; therefore, auricle identification, frame selection and auricular point feature point marking tasks can be efficiently completed; on the basis, high-frame-rate real-time prediction is realized in the field of auricular point positioning for the first time, and the limitation of the prior art is broken through.

The invention discloses a multi-task adaptive learning method based on dynamic strategy switching. The method comprises the following steps: S1, constructing a multi-task learning model, and initializing a task encoder, a feature extraction network and a decoding module; s2, encoding an input sample into a task encoding vector; s3, extracting task feature representation; s4, inputting an improved Cross-pitch structure, and carrying out the feature cross fusion of the improved Cross-pitch structure; s5, inputting a strategy scheduling controller, and generating fusion weight and path configuration based on loss change, gradient difference and feature distance; s6, executing nonlinear cross fusion to generate shared feature representation; s7, generating a task prediction result; s8, calculating loss and updating parameters; and S9, circularly training until convergence. According to the method, adaptive regulation and control of a fusion strategy and dynamic optimization of a feature sharing structure are realized, and the method is suitable for a multi-task neural network modeling scene.

The invention discloses a milling force prediction method based on a transfer learning enhanced WM-GRU neural network, and the method comprises the steps: considering a tool wear mechanism, building a WM-GRU prediction model through the influence relation of the tool wear to the milling force in a machining process, predicting the tool wear through GRUW, transmitting a prediction result to GRUM, and carrying out the milling force prediction through the GRUM according to the tool wear and cutting parameters; a migration enhancement training strategy is formulated, fusion with a WM-GRU prediction model is carried out through parameter migration, feature sharing and field adaptation, high-precision adaptation of a target domain is realized based on progressive migration fine tuning, construction of a milling force prediction model based on migration learning and WM-GRU neural network deep fusion is completed, and finally milling force prediction is realized. According to the method, the defects that a milling force prediction model established based on a mechanism is incomplete and a milling force prediction model established based on pure data driving is poor in interpretability are overcome, and the prediction performance of the model on a target domain is improved by introducing transfer learning.

The invention discloses a multi-target prediction method, system and device for a financial time sequence and a storage medium, and relates to the technical field of artificial intelligence, and the method comprises the steps: firstly obtaining financial time sequence data and auxiliary information, and carrying out the vector conversion and fusion to form a unified input representation; extracting features through a shared encoder, and generating a global feature vector; and inputting the data to a multi-task prediction output head to realize parallel prediction of the prediction tasks. According to the scheme, cross-task feature sharing is realized by using the shared encoder, the modeling collaboration is improved, and resource waste and prediction conflicts are avoided; through data fusion, the depiction capability of complex financial dynamics is enhanced; the generalization and robustness of the model are improved by adopting a lightweight output head, the method is suitable for diversified financial prediction scenes, and the prediction efficiency and accuracy are improved while the model complexity is reduced.

The invention provides a federal learning method, system and device based on global feature sharing and a medium, and the method comprises the steps: initializing a bottom feature extraction network of a global model through employing a server, and sharing the parameters of the bottom feature extraction network to all clients; the server dynamically calculates contribution utility scores of the clients according to differences between underlying model parameters uploaded by the clients and current global underlying network parameters and local data scales of the clients, and constructs probability distribution according to the utility scores, then selecting a client set participating in the training in a weighted random sampling mode; and the selected client recombines the updated global underlying network parameters with locally reserved high-level network parameters to form a new composite model, and the training process is repeatedly executed until the model reaches a preset convergence condition, so that the data transmission quantity between each client and the server can be remarkably reduced, and the data transmission efficiency is improved. Bandwidth resources are greatly saved, and the efficiency of large-scale distributed training is improved.

The invention provides a campus bullying event detection method, and the method comprises the steps: obtaining to-be-detected monitoring data in a campus, the to-be-detected monitoring data comprising a plurality of target persons; respectively extracting a first video feature and a first audio feature of the to-be-detected monitoring data; based on the first video feature and the first audio feature, determining multiple pairs of target features related to the campus bullying event and a weight combination corresponding to each pair of target features in a pre-constructed feature sharing subspace; weighted summation processing is carried out on the target features based on the weight combination, fusion features are obtained, and each fusion feature corresponds to a pair of target features; and performing classification processing on the plurality of fusion features, and determining whether a campus bullying event exists in the to-be-detected monitoring data. According to the invention, the detection accuracy of the campus bullying event is improved, the campus bullying event can be early warned in time, and the occurrence of the campus bullying condition is effectively prevented.

The invention discloses a PCB defect detection method and system based on adaptive multiple submodels, and the method comprises the steps: collecting an image of the surface of a PCB through a high-resolution camera, and carrying out the preprocessing of the image; performing dynamic priority decision through reinforcement learning, and reasonably selecting the priority of the sub-model according to the equipment state and the load of the computing resource; yOLOv8 is adopted as a backbone network, a plurality of sub-models work in parallel, and the sub-models are trained in combination with gradient masks; realizing multi-stage cascade detection by adopting a confidence threshold screening and variance analysis mode, and reversely optimizing parameters of the model according to a batch statistical result; and according to the recall rate and the false detection rate of each detection, optimizing the sub-model through an incremental training mechanism. According to the method, the adaptive multiple sub-models are provided, and through a dynamic priority control and feature sharing mechanism, the high accuracy is ensured, the detection strategy is dynamically adjusted, the calculation burden is reduced, and efficient and accurate PCB defect detection is realized.

The invention relates to an access control strategy automatic generation system and method based on multi-task learning, and the method comprises the steps: carrying out the word segmentation, cleaning and embedded vector conversion of an original access control text through a data preprocessing module, and constructing a normative input format; the feature sharing layer module is used for extracting deep semantic features of a text through multi-layer bidirectional coding and an attention mechanism and providing unified representation for downstream tasks; the access control statement identification module is used for judging whether each sentence in the text is an access control statement or not and realizing automatic identification of strategy related contents; and the attribute extraction and annotation module is used for annotating words in the access control statements and extracting subject, object and operation access control attributes. A word coding layer and a sentence coding layer are shared, local and global attention mechanisms are combined, key information of a text is extracted, the semantic understanding ability is enhanced, and cooperative training of statement recognition and attribute extraction is achieved; a conditional random field CRF structure is used for sequence labeling, and the structural rationality of attribute labels is ensured.

The invention discloses a Conformer-based multi-task wireless communication signal classification method, and belongs to the technical field of crossing of signal processing and artificial intelligence. In order to solve the problems of task modeling isolation, insufficient feature sharing mechanism and weak model generalization ability in existing wireless communication signal classification, the method comprises the following steps: acquiring an IQ signal, performing dimension raising through a front-end convolution module, inputting the IQ signal into a Conformer network fusing local convolution and a global attention mechanism, and performing depth time sequence feature extraction; and synchronously realizing discrimination of a signal-to-noise ratio grade, a channel type, a modulation mode and a communication system by using a parallel multi-task classification head. The method can improve the classification accuracy and the model generalization ability, reduces the consumption of computing resources, and is suitable for signal recognition in wireless communication, radar and Internet of Things systems.

The invention relates to a deep feature fusion-based fabricated component surface defect multi-task detection method. The method comprises the following steps of: constructing a multi-view camera array and an illumination feedback regulation and control module at a data acquisition end; on the algorithm level, feature cross-layer propagation and information compensation are realized through an improved lightweight convolutional network and a multi-scale residual diffusion module; a dynamic feature fusion module is introduced, and a self-adaptive fusion weight is generated based on channel statistical features, so that feature sharing and differential expression are realized among different tasks; meanwhile, a defect perception attention mechanism and cross-task consistency constraint are adopted, and the problem that semantic space distribution is inconsistent in the multi-task detection process is solved; in the detection post-processing stage, a three-dimensional quantitative evaluation system based on the geometric dimension, the texture roughness and the depth volume is constructed, and unified grade evaluation of the surface defects of the component is achieved through the multi-feature fusion quality index. The problems of low detection efficiency, unstable precision, difficulty in collaborative recognition of multiple types of defects and the like in existing component delivery detection are solved.

The invention discloses a personnel behavior recognition method and device in a low-illumination scene and a medium, relates to the technical field of image processing, and aims to integrate image enhancement and attitude estimation into a behavior recognition model to solve the recognition difficulty in a low-illumination condition. Three tasks of image enhancement, human body posture estimation and personnel behavior recognition are integrated, multi-task neural network model training is carried out based on posture weighted global pooling consistency and posture-driven background random suppression, and through feature sharing and task cooperative training, the human body posture estimation and personnel behavior recognition are realized. According to the method, the problem of low accuracy of personnel behavior recognition in low-illumination monitoring environments such as oil fields can be solved, particularly for the conditions of dark and fuzzy images and changeable human body postures, the utilization of posture information is improved, the robustness and precision of personnel behavior recognition in complex environments are improved, and end-to-end optimization is realized.

The invention relates to the field of multi-modal computer vision, in particular to a real-time semantic segmentation and target detection joint model for multi-modal data fusion. The model is realized through the following technical means: designing a multi-branch network based on an RGB image, depth information and laser radar data, and respectively extracting different modal features; a dynamic modal selection module is introduced, an optimal modal combination is automatically selected according to environmental conditions, and the adaptive ability of the model in a complex scene is improved; realizing cross-modal information complementation by using a UAFM feature fusion module; multi-scale features are fused step by step through a step-by-step decoding structure so as to balance detail and global semantic information; a joint training mechanism is adopted, a semantic segmentation result and a target detection frame are synchronously output, and calculation redundancy is reduced. According to the method, the segmentation and detection precision in a complex scene is remarkably improved under the support of multi-modal data; through dynamic modal selection and feature sharing, efficient real-time processing is realized, and the reasoning speed is improved by more than 30% compared with that of an independent model.

The invention relates to the technical field of commodity value evaluation, in particular to a commodity value evaluation model construction method and system based on a small sample scene. Comprising the following steps: establishing a plurality of sub-data sources according to equipment category parameters, and constructing a federal collaboration layer according to all the sub-data sources; acquiring a first-level shared feature packet generated by the federal collaboration layer, correcting the first-level shared feature packet according to a preset transfer learning layer, and generating a second-level feature packet according to a correction result; an initial evaluation model is constructed according to the secondary feature package, and a commodity evaluation model is constructed by the time sequence layer according to the initial evaluation model; a plurality of sub-data sources are established on the basis of equipment categories, a feature space alignment mechanism of cross-category equipment is established by establishing a mapping transmission model of each sub-data source, local feature sharing of different equipment categories is realized through federated learning, and the number of dependencies of evaluation and cross-category evaluation errors of the model are reduced.

The invention relates to a multi-task lightweight detection method for factory appearance defects of fabricated concrete members. The method comprises the steps of image data acquisition and preprocessing and lightweight multi-task detection model construction. The lightweight multi-task detection model construction process comprises a feature extraction module, a task branch module, a feature fusion module and an attention optimization module, the model design follows the principle of lightweight parameter compression, multi-task feature sharing and deep semantic fusion, and recognition and quantification of multiple types of appearance defects can be completed in a single network at the same time. According to the method, real-time, accurate and integrated detection of multiple defects on the surface of the component is realized, so that the industrial application requirement of factory detection of a component factory is met, the detection quality and efficiency are remarkably improved while the detection cost is reduced, and a more efficient and more reliable technical means is provided for factory quality control of the fabricated concrete component.

The invention discloses a method for calculating a five-axis finish machining tool path approximation error through a multi-task neural network. The method comprises the following steps: firstly, acquiring core parameters required by calculation of a neural network model, and mapping the parameters to the same scale by applying mean variance normalization; then, according to PCA (Principal Component Analysis) dimension reduction, screening is carried out on the data characteristics; regularization processing is added to optimize a network structure, and a part of neurons are hidden in each iteration process, so that interference of non-core knife contacts on approximation error value calculation is reduced, and model overfitting is avoided; an attention mechanism is introduced, feature weights are dynamically distributed, key features are highlighted, a higher weight ratio is given, the learning ability of the model is enhanced, and the training precision is improved; and finally, dynamically adjusting the feature sharing proportion among the tasks by using a cross connection method, ensuring that the unique features of each task are reserved while sharing information, reducing interference among the tasks, refining feature sharing among the tasks and enhancing the generalization ability of the model, thereby improving the calculation precision of the approximation error.

The invention provides a business data feature sharing method and system based on machine learning, and the method comprises the steps: firstly obtaining a business link feature set of a talent service field, including a personnel core, post association and service process features, and a business demand description of a feature calling main body; then constructing a feature anchoring rule to perform association processing to obtain a feature anchoring result, inputting the feature anchoring result into a pre-training machine learning model to calculate a shared adaptation parameter, and constructing a shared execution link including feature extraction, processing and delivery nodes according to the shared adaptation parameter; and carrying out association mapping on the features output by the sharing execution link and a talent service business system of the feature calling main body, generating a feature sharing result containing a plurality of identifiers, and feeding back the feature sharing result to the business system to support business decision operation, thereby realizing efficient and accurate sharing of talent service business data features.

The invention discloses a multi-modal image loop iteration registration method based on physical driving. The method comprises the following steps: learning own characteristics of an SAR image by using an SAR imaging mechanism learning model; inputting the optical image into a multi-scale optical image feature extraction module, and capturing optical information from local to global; carrying out feature fusion by using a multi-modal feature sharing learning model, and introducing sharing parameters to carry out feature alignment and preliminary comparison so as to match the most suitable point pair; in the feature matching optimization stage, an RIFT2 method is used as an external supervision signal, global features extracted by deep learning are combined, a dual loss strategy is adopted, and two pieces of key information from the RIFT2 are integrated into a loss function; and final registration is realized through a checkerboard visualization result. According to the invention, through an efficient heterogeneous image registration technology, the capability of realizing heterogeneous image accurate registration under the influence of different differences of heterogeneous images is improved.

The invention discloses a lightweight multi-task network model for vehicle multi-attribute identification based on a roadside perception image. The lightweight multi-task network model comprises the following steps: constructing a vehicle multi-attribute identification data set in a road scene; the method comprises the following steps of: aiming at diversified requirements of road side application, constructing a multi-task vehicle multi-attribute identification model (Task-Driver Adaptive Multi-Task Network, TamNet) for a road side perception image on the basis of an adaptive feature sharing mechanism, and constructing a multi-task vehicle multi-attribute identification model (Task-Driver Adaptive Multi-Task Network, TamNet) for the road side perception image based on the Task-Driver Adaptive Multi-Task Network; according to the light weight requirement of road side application, the network further integrates a progressive multi-task cyclic pruning (PCP) to achieve model light weight, and a light weight multi-task vehicle multi-attribute recognition model TamNet-PCP used for road side sensing images is constructed. According to the invention, through a task-driven adaptive multi-task network (TamNet) and a progressive cyclic pruning (PCP) framework, efficient combined identification of vehicle brands, types and colors can be realized, so that deployment of road side sensing equipment is optimized, and a basis is provided for realizing rapid and accurate identification of vehicles in a road scene subsequently.

The invention relates to the technical field of data processing and information security, in particular to a thermal management model sharing method based on a block chain and homomorphic encryption. Creating a point account, completing user identity information registration and block account application, and applying for a corresponding block chain account for the user after basic identity information registration is completed; issuing the shared heat management model, namely completing the issuing of the user shared heat management model in the block chain and the storage function of the model in the IPFS; downloading the shared model, retrieving the thermal management model, and after the credit is paid to the model provider, downloading the model by the user; a user issues a thermal management model through a block chain system, other users retrieve the issued thermal management model in the system through a block chain intelligent contract, and the user needs to pay a certain amount of points to a sharing party, so that joint simulation of digital models in different subject fields and repeated use of the models can be realized; and the problem of multi-functional demand modeling of a complex system can be solved.

This invention relates to the field of artificial intelligence technology and discloses an obstacle detection method and system based on unmanned mining trucks. One obstacle detection method for unmanned mining trucks includes: collecting and preprocessing vehicle data; constructing a dust and fog evolution prediction network to obtain a defogging multispectral image; extracting the distribution information of different materials in the scene to form an enhanced scene feature representation; performing adaptive fusion and uncertainty evaluation to obtain a gating feature representation; using an obstacle detection model to detect obstacles; and constructing a multidimensional occupancy grid to output multidimensional obstacle data. This invention overcomes the performance bottleneck of traditional defogging algorithms in heavy dust and fog environments in mining areas by decoupling from the dust and fog environment through physical modeling. By constructing a multi-vehicle collaborative perception network with sparse feature sharing, it overcomes the problems of blind spots and insufficient long-distance perception capabilities in single-vehicle perception, enhancing the system's detection reliability in dynamic and complex environments and ensuring the safety of autonomous driving operations in mining areas.

The invention discloses a hyperspectral blind unmixing method, which comprises the following steps of: constructing a hyperspectral blind unmixing model, inputting hyperspectral image data into the hyperspectral blind unmixing model, encoding spatial features and spectral features of a mixed spectrum of each pixel in the hyperspectral image data through a shallow encoder, and obtaining a hyperspectral blind unmixing model; synchronously transmitting the combined features to an end member extraction branch and an abundance estimation branch through a feature sharing mechanism; the end member extraction branch receives hidden end member features extracted from the joint features through a feature extraction module, and then pure end members are extracted from the joint features by using a Colmogolov-Arnold network; according to the abundance estimation branch, an abundance graph corresponding to a pure end member is extracted from the joint features by using a windowed multi-head self-attention mechanism, and an unmixing result of a mixed spectrum is obtained. The method does not need to depend on prior end member information, the dependence of a traditional method on a known end member library is reduced, and the de-mixing efficiency, stability and accuracy of the system are improved.

This invention discloses a lightweight method and system for detecting small targets on UAVs in complex scenarios, relating to the field of target detection technology. It acquires image data captured during UAV flight and annotates the targets to construct a standardized dataset for training and validation. Based on this, the images undergo size normalization and left / right block cropping to form candidate detection regions suitable for small target detection. These candidate detection regions are then input into a small target detection model built on an improved YOLOv5s platform. Through lightweight feature extraction, multi-scale feature fusion, and detection result mapping and filtering, the model outputs detection results containing target category information, target bounding box position, and target confidence. Simultaneously, by introducing a cross-regional collaborative enhancement mechanism, the semantic constraints and feature sharing between block regions are strengthened, thereby improving the continuity, stability, and lightweight deployment adaptability of UAV small target detection in complex backgrounds.

The application discloses a system and method for synchronously detecting epilepsy attack and type identification, and the system comprises an electroencephalogram signal processing module, a multi-scale time-space-frequency feature electroencephalogram fusion module, an epilepsy attack detection task branch, a cross-task attention interaction module, an epilepsy attack type classification branch and a multi-task learning optimization module; the cross-task attention interaction module realizes bidirectional feature sharing between the task branches by establishing an information interaction mechanism at a feature level, so that the epilepsy attack detection task can utilize fine-grained structural information related to attack types to improve detection precision, and meanwhile, the attack type classification task can utilize attack time positioning and context information provided by the detection task; the application realizes collaborative modeling of epilepsy attack detection and attack type classification, improves the accuracy, robustness and clinical application value of epilepsy attack identification by sharing feature representation and cross-task information interaction, while ensuring the calculation efficiency.

Aspects relate to techniques for visual feature sharing between wireless communication devices for relative pose determination. A first wireless communication device may transmit a request for visual feature sharing to a second wireless communication device and in response receive a message from the second wireless communication device including a plurality of features (e.g., keypoints) of an image captured by the second wireless communication device. The first wireless communication device may then calculate a relative pose of the first wireless communication device with respect to the second wireless communication device based on an association between the features provided by the second wireless communication device and additional features obtained from an additional image captured by the first wireless communication device.

The invention provides a deep collaborative multi-task feature learning method based on mutual information regularization. Aiming at the technical problems of low feature sharing efficiency, serious inter-task interference, poor generalization ability and the like in the existing multi-task learning, the invention constructs a deep learning framework based on mutual information constraint, and designs a mutual information calculation module of variational inference and a KL divergence regularization optimization mechanism. And cross-task feature efficient sharing is realized through a multi-layer encoder structure and a feature adaptive selection mechanism. According to the scheme, mutual information calculation is optimized by adopting a batch estimation strategy and a parallel calculation mechanism, and a dynamic task weight distribution and cross-task knowledge migration mechanism is introduced to enhance the generalization ability of the model. Experiments show that compared with the prior art, the classification accuracy, convergence speed, calculation efficiency and the like of the method are improved by 31.2%, 65.3% and 44.8% respectively, and the method has higher cross-domain migration ability and can be widely applied to multi-task learning scenes such as computer vision and natural language processing.