333 results about "Task network" patented technology

The invention discloses an industrial equipment communication method and system based on the industrial internet, and the method comprises the following steps: S1, collecting the multi-modal data of industrial equipment, and generating a preliminary multi-modal fusion feature; s2, performing event-driven analysis on the preliminary multi-modal fusion features to generate event sensing data; s3, constructing a semantic-driven communication protocol, and generating a semantic data queue with a priority identifier; s4, compressing the semantic data queue and carrying out time consistency calibration; s5, optimizing a communication frequency and bandwidth resource allocation strategy in combination with a mixed firefly optimization algorithm and a dynamic differential evolution algorithm; s6, constructing an industrial equipment communication topology model, and predicting path delay and bandwidth occupation indexes of the communication network in combination with the space-time multi-task network; and S7, constructing a zero-sum game model to analyze the potential confrontation relationship in the communication process, and outputting a secure communication scheme. According to the invention, semantic-driven communication, a hybrid optimization algorithm and the like are adopted, and intelligent communication of industrial equipment is realized.

The invention discloses a task scheduling optimization and feedback control method and system based on a causal graph structure, and relates to the technical field of task scheduling, and the method comprises the steps: integrating multi-source information, identifying an entity, constructing a directed edge connection entity, calculating an edge weight, and forming a task causal graph; performing deep analysis on the causal graph, and identifying risk nodes and propagation paths thereof in the current project; combining the edge weight and the node attribute, quantifying the risk propagation possibility and influence range, and generating a risk thermodynamic diagram; on the basis of the identified risk nodes and propagation paths thereof, mining controllable variables in a task scheduling process, constructing a scheduling optimization objective function, searching a scheduling solution space under constraint conditions by using a search method, and evaluating the influence of different strategies on a task network through multiple rounds of simulation, so as to improve the task scheduling efficiency. A current optimal task adjustment strategy is screened out and practically applied; and collecting data generated in a project execution process, and updating the causal diagram. The method is suitable for intelligent scheduling and fine management scenes of projects.

The invention discloses a dynamic data processing and recognition method and device for a mixed font text, and belongs to the technical field of optical character recognition, and the method comprises the following steps: collecting a mixed font text image to be processed, and inputting the mixed font text image into an enhanced DBNet detection network to position a text region; integrating a lightweight MobileNetV3 font classification sub-network, sharing bottom layer features with a detection trunk, training a multi-task network through a joint loss function, and outputting a text region with a font type label; then selecting identification branches according to the labels, and identifying the printed form by using the traditional CRNN; according to the handwritten form, traditional convolution is replaced by linear variable convolution, kernel dynamic scaling is guided by a thermodynamic diagram, and multi-head attention mechanism recognition is embedded in a BiLSTM layer; and finally, results are fused, and detection and identification results with labels are output. According to the method, the mixed font text data processing and recognition precision is improved, the adaptability and robustness of handwritten text recognition are enhanced, and various text recognition requirements are met.

The invention discloses a medical risk intelligent prediction system based on time series data mining. The system comprises a multi-dimensional time sequence data acquisition and preprocessing module, a time sequence mode deep mining engine, a multi-dimensional risk assessment engine, an intelligent intervention decision support system and a real-time monitoring feedback module. A time sequence mode mining engine adopts a layered architecture, and short, medium and long-term time sequence modes are respectively analyzed through a bidirectional LSTM-attention network, a wavelet transform-convolutional network and a seasonal decomposition-gating circulation network. The risk assessment engine integrates an isolated forest, an auto-encoder, a Transform multi-task network and knowledge graph reasoning, and realizes all-around risk quantification. The decision support system generates a personalized intervention strategy based on deep Q network reinforcement learning and case reasoning. According to the system, early prediction and accurate intervention of medical risks are realized, and the prediction accuracy and the medical safety level are remarkably improved.

An X-centric network architecture and associated methods are provided. The network architecture provides comprehensive control and processing functions that support heterogenous XaaS services. For different XaaS services, the network architecture supports the activation / selection of different functions / interfaces. The network architecture has a control / management (C / M) plane that includes a mission manager (MM) that manages a mission comprising tasks associated with heterogenous services. The network architecture also has a mission data plane (MDP) coupled to the C / M plane. The MDP plane includes processing functions (PFs) required for the tasks and also includes data plane functions (DPFs) that are configured to interface a PF of a XaaS service to a PF of a different XaaS service. The network architecture also has one or multiple XaaS module(s) which may be provided by third parties as plug-in module(s). Each XaaS module is associated with a XaaS service, one of multiple XaaS service controller(s) (XC), and one or multiple PF(s) / DPF(s) conducting corresponding tasks. Detail procedures, interfaces, and signals for mission execution and device accessing a mission (mission access) are described in the associated methods.

The invention relates to the technical field of automatic driving, in particular to an automatic driving task processing method and device, equipment and a medium. According to the method, firstly, the navigation guidance information is acquired and directly incorporated into beyond visual range contents (such as remote steering and distance), so that the limitation of the sensing range of a vehicle-mounted sensor is broken through, and the problem that an existing system lacks beyond visual range information is solved. And secondly, semantic analysis is performed based on navigation information and multi-view visual data to obtain text probability distribution, so that association of global navigation context and real-time scenes is realized, and a semantic basis is provided for humanization-like planning. Then, performing cross-modal fusion on the aerial view angle features and the text probability features, so that global navigation logic and real-time environment perception are deeply coupled, and decision is ensured to consider both a short-term environment and long-term planning. And finally, fusing the features and inputting the features into a task network, so that sensing, prediction and planning are more suitable for an actual scene, the utilization capability of beyond-visual-range information in a complex scene is improved, and the adaptability and reasoning capability of the system are enhanced.

The invention provides a farmland plot extraction method based on a multi-type double-branch convolutional neural network, and the method comprises the steps: creating a double-branch multi-task network DBMNet based on a multi-type convolution module, and designing a double-branch feature extraction module DBFE to replace a convolution layer of an original network, a dynamic linear constraint convolution module LCDC and a multi-scale deformable convolution feature extraction module MDFE are used to comprehensively extract edge features of an image and field block region features of different sizes and shapes, the perception ability of a network to information of different scales is improved, and thus the segmentation precision of the network is improved. The network uses boundary detection and distance regression tasks to improve the precision extraction of the farmland plot masks, so that the network has a better generalization effect. According to the method, the multi-type convolutional network is applied to the farmland plot extraction task, the experimental result shows that compared with other most advanced algorithms, the method has the best segmentation performance, discriminative features of the farmland plot can be effectively extracted, the segmentation precision is remarkably improved, the farmland plot is accurately extracted, and the method has great application potential and prospect.

The invention discloses a geological monitoring data processing method and system based on multiple neural networks, and relates to the technical field of geological monitoring data processing, and the method comprises the steps: employing a dynamic heterogeneous neural network fusion architecture, and carrying out the classification routing processing of various types of geological monitoring input data, obtaining a corresponding first local space feature, a second fault topological structure feature and a third time sequence feature vector; performing optimization training on the multi-task prediction network by using a joint loss function with geological physical constraints, the main task network being used for predicting an earth crust displacement field, and the auxiliary task network being used for predicting an earth crust stress field to obtain an optimized fusion model; based on the optimized fusion model, performing reasoning prediction on the new geological monitoring input data, and outputting a comprehensive geological risk assessment result; through the dynamic heterogeneous neural network fusion architecture, various geological monitoring input data are classified and routed, and the problem that multi-source heterogeneous geological data cannot be effectively processed through a traditional method is solved.

The invention provides an intelligent task scheduling and optimizing method and system, a medium and equipment, and the method comprises the steps: analyzing task metadata through a database storage process, generating a task dependency graph according to an obtained task dependency matrix, and converting the task dependency graph into a visual interaction interface; identifying a key link with longest time consumption and a bottleneck task on the key link through the created time prediction model, and generating a multi-dimensional optimization suggestion based on an identification result; automatically adjusting a task scheduling script according to the multi-dimensional optimization suggestion and generating a standardized configuration file, dynamically allocating resources according to the standardized configuration file, and monitoring an execution state of a task through a fault-tolerant mechanism; and performing intelligent early warning according to the task log data collected in real time and the system performance index, and generating a multi-dimensional analysis report. The system is an intelligent scheduling system integrating dependency analysis, link optimization and execution monitoring, and automatic analysis and optimization of a complex task network are achieved.

The invention belongs to the technical field of remote sensing image processing. The invention provides a multi-source remote sensing image class incremental learning method based on prompt fine tuning. According to the embodiment of the invention, a mode-specific prompt pool is constructed, an instance-based prompt query mechanism is designed, and proper prompts are dynamically selected for different input characteristics; a learnable modal sharing global prompt is added to each attention layer of the frozen Vision Transform network, and modal public information specific to a task is extracted; carrying out cross-modal prompt conversion and fusion; freezing a part of the modal mapping network to reserve a modal conversion relation of the old task; establishing a prompt-guided knowledge aggregator module, freezing a knowledge aggregation token in an old task network, and jointly guiding the knowledge aggregation token, a knowledge aggregation token of a new task and a modal specific prompt to guide the aggregator module to learn image features; and training the established incremental learning network model.

The invention discloses an agent enterprise project progress management system and method, and the system achieves the intelligent task scheduling, the precise progress prediction and the optimal resource distribution through a unique dynamic task network and reinforcement learning, a space-time attention mechanism and variational auto-encoder, a multi-agent cooperation and game theory and other algorithms. The system comprises four layers of architectures of data acquisition and transmission, processing and analysis, intelligent decision and control and user interaction. Experiments prove that the system and the method effectively reduce the project progress deviation rate, improve the resource utilization rate and the task on-time completion rate, reduce the project cost, and assist large model technology enterprises to efficiently manage the project progress.

The invention belongs to the technical field of computing power distribution, and particularly relates to a dynamic computing power distribution method and system based on reinforcement learning, and the method comprises the following specific steps: S1, covering cloud, edge and end full-node scenes, and collecting computing power resource states, task demand features and cross-domain network condition data in real time; s2, on the basis of standardized data output by a cross-domain computing power sensing module, by constructing a state space fusing computing power, tasks and a network, defining an action space of computing power scheduling direction and proportion, and designing a multi-target reward function for balancing the resource utilization rate, the task satisfaction rate and long-term conflict avoidance; and realizing self-learning and self-iteration scheduling strategy generation based on a reinforcement learning algorithm. According to the invention, the reinforcement learning agent autonomously learns the computing power demand of the emergency scene and the new type of task, the rule does not need to be manually preset and modified, and the method has the advantage of realizing dynamic adaptation of computing power distribution to complex and changeable scenes.

The present invention relates to the technical field of steel industry load participation in power grid interaction, and provides a double-layer optimization scheduling method and system for steel industry load, the method comprising: constructing a continuous time interactive resource task network model based on double-layer optimization data information; constructing a resource task network model with production process constraints; constructing a complex order characteristic analysis model for steel industry load; setting dynamic maximum demand analysis model operation constraints for the complex order characteristic analysis model; constructing an upper-layer power grid scheduling optimization model that achieves optimal power grid economic cost, and the lower-layer steel industry load side full system model and the upper-layer power grid scheduling optimization model jointly form a double-layer optimization system; proposing a double-layer optimization algorithm for the double-layer optimization system, solving the double-layer optimization scheduling; and outputting double-layer optimization scheduling result information. The present invention significantly reduces the electricity cost of short-process steel industry loads, and allocates production orders in combination with steel order characteristics to achieve better peak shaving and valley filling effects.

The application discloses a memory reinforcement deep unfolding multi-modal image fusion method with downstream task enhancement, which comprises the following steps: collecting infrared images and visible light images, and dividing training set and test set; establishing an optimization target and solving, to obtain an iterative formula; using a neural network to replace a proximal operator in the iterative formula, to obtain a neural network structure; sending the training set into the neural network to obtain a fusion picture, and calculating a total loss according to the fusion picture, the infrared image and the visible light image; updating parameters of the neural network according to the total loss to obtain an updated neural network; inputting the infrared image and the visible light image of the test set into the updated neural network to obtain a fusion image. The application makes the fused image have characteristics easy to be distinguished by a downstream task network, and can realize the best performance on a data set, while performance and interpretability are taken into account, and the application has rationality and applicability.

The invention discloses a visual process management method and device and a storage medium, and the method comprises the steps: obtaining the input data of task state change, and carrying out the model initialization processing, and obtaining a task network model; according to the task network model, performing path search by adopting a graph traversal algorithm to obtain an influence path set; according to the influence path set, carrying out influence matrix construction operation to obtain an influence degree matrix; according to the influence degree matrix and the task network model, performing dynamic influence path mapping to obtain a dynamic propagation path diagram; key nodes are extracted according to the influence degree matrix and the dynamic propagation path diagram, and a decision point set is obtained; performing priority calculation and visual identification processing according to the decision point set to obtain a deviation node identification result; and performing model optimization processing according to the deviation node identification result to obtain an optimization task network model. According to the method, real-time dynamic layout adjustment of process management can be realized.

This application provides target recognition method, a multi-task network model training method, and an electronic device. The target recognition method includes: inputting video images into a multi-task network model one by one to obtain a predicted feature map; performing post-processing on the predicted feature map to obtain a target detection result; judging whether a target class confidence degree is greater than a preset confidence degree; if so, judging whether a target image quality score is greater than a preset score; if so, cropping out a target image from the video images according to a target detection box; and inputting the target image into a target recognition model corresponding to the target class to obtain a target name. In this way, this application decreases the number of calls of the recognition model, and also reduces a training duration of the model.

The invention discloses a rock slag multi-dimensional intelligent identification and real-time early warning method for tunnel boring machine construction. The method comprises the steps that continuous rock slag images are acquired and preprocessed; inputting the preprocessed rock slag image into the trained multi-task network, and outputting a multi-task result; wherein the task network adopts a unified encoder and multi-head decoder architecture, the multi-head decoder comprises a segmentation head, a detection head, a regression head and an anomaly scoring head, the segmentation head outputs a pixel mask, the detection head outputs a rock slag frame and category, the regression head outputs a particle size distribution parameter, and the anomaly scoring head outputs a frame level anomaly score; synchronously carrying out sliding window statistics based on output results of the regression head and the abnormal score head; and performing multi-level early warning judgment based on a sliding window statistical result. According to the method, multiple functions of rock slag instance segmentation, target detection, particle size distribution regression, flow estimation, anomaly scoring and the like can be realized, and multi-stage early warning and parameter suggestion are realized through time sequence statistics.

The invention relates to the technical field of engineering management, in particular to an AI-based engineering quantity list automatic compiling method and system, and the method comprises the following steps: constructing a model through analyzing a construction task dependency relationship, tracking the progress in real time, and predicting changes through a neural network to generate change analysis; and optimizing delay task adjustment time and resource configuration in combination with constraint conditions, evaluating a weather risk level and dynamically adjusting a construction period, and finally dynamically updating a project quantity list based on resource and risk results. According to the method, a task network structure with a quantifiable scoring mechanism is constructed by analyzing the dependency relationship between the tasks, so that the logic association between the tasks not only stays on the static hierarchy presentation, but also introduces the difference weight of time and resource dimensions to realize dynamic modeling on the dependency of key paths; and depth support is provided for subsequent prediction and adjustment. Tracking of task progress changes does not depend on a static progress comparison mode any more, but deep extraction of actual construction data is carried out.

The invention discloses a driving scene multi-task perception method based on a coordinated multi-scale feature enhancement network (CMFANet, and the method comprises the steps: constructing a multi-task network model based on the coordinated multi-scale feature enhancement network (CMFANet, firstly, employing a hybrid enhancement strategy for a shared backbone network layer, extracting multi-level feature representation from an input image, and carrying out the multi-level feature representation; and then the task neck layer processes and perfects the extracted features by adopting a feature fusion strategy, integrates multi-scale feature representations to obtain fused features, and finally the task head receives the fused features and generates final output according to the requirements of each task to realize comprehensive perception of the traffic scene. The CMFANet model constructed by the method can execute traffic object detection, drivable area segmentation and lane line segmentation at the same time in a resource-limited environment, the model gives consideration to efficiency and accuracy, meets the requirement of real-time application, can solve the problems of traffic object detection, drivable area segmentation, lane line detection and the like at the same time, and has a wide application prospect. And comprehensive perception of a traffic scene is realized.

The invention discloses a small sample aerodynamic modeling method based on multi-task learning. Comprising the following steps: constructing a multi-task prediction model, integrating an auxiliary task network and a target task network, and enhancing the feature extraction capability of an encoder through an SE layer and an attention layer; acquiring multi-source aerodynamic force data and carrying out standardized preprocessing on the multi-source aerodynamic force data; a dynamic weight mechanism is designed, the influence of auxiliary task prediction on target task output is adaptively adjusted according to input features, and effective fusion of low-fidelity data and high-fidelity data is achieved; two-stage training is carried out to ensure the efficiency and stability of multi-task learning; the prediction capability of the model under the limited sample condition is verified; through the synergistic effect of multi-task knowledge migration, feature enhancement and dynamic task integration, in combination with the feature optimization characteristics of an SE layer and an attention mechanism, high-precision aerodynamic prediction is realized under the condition of limited samples, and an efficient and accurate aerodynamic prediction method is provided for aircraft design.

The invention provides a few-sample fine-tuning frame construction method for video anomaly detection. The method comprises the following steps: constructing a few-sample scene video anomaly detection fine-tuning frame based on an auxiliary task, segmenting a target which is easy to generate an anomaly in a video, and predicting an optical flow graph at the same time. Through two auxiliary tasks of segmentation and optical flow estimation, rich spatial and temporal features are extracted to promote an anomaly detection model to be finely adjusted to a new scene; a three-stage training strategy is adopted, for three parts of an auxiliary task network, namely a segmentation branch network, an optical flow branch network and a segmentation and optical flow information fusion module, one part is trained in each of three training stages, and parameters of the other two parts are frozen, so that the instability of the training process is reduced. According to the few-sample fine tuning frame construction method and construction system for video anomaly detection, the segmentation target of the video and the optical flow information are effectively combined, and the method and system have the advantages of being high in fine tuning efficiency, low in cost, easy to implement, flexible in deployment and the like.

The invention discloses an accidental explosion concrete penetration depth prediction method based on a two-stage migration adversarial network, and the method comprises the steps: (1) an instance screening stage: resampling target domain test data to construct a plurality of sub-data sets, training a base learning device, and calculating a weight coefficient; and inputting the source domain simulation data into the base learner to obtain a weighted predicted value, screening out the simulation data conforming to a target domain rule through a threshold value, and optimizing the source domain data quality. (2) a prediction stage: constructing a dynamic adversarial network model (including a weight network, a task network and a domain discriminator), distributing dynamic weights through the weight network, and aligning cross-domain feature distribution in combination with domain adversarial learning; weighted task loss and adversarial loss are jointly optimized in back propagation, so that the model adaptively fuses effective information in a hidden space, and a cross-domain generalization prediction network is generated. According to the method, the knowledge migration problem of simulation data and test data is effectively solved, and the method has excellent performance in a concrete penetration depth prediction task.

The invention discloses a three-dimensional medical image reconstruction pre-training method based on task perception, and the method comprises the steps: carrying out the preprocessing of an original three-dimensional medical image, and obtaining a preprocessed image; constructing a mask reconstruction main task, and establishing a main task network in the mask reconstruction main task; the main task network comprises a mask module, a reconstruction main task encoder and a reconstruction main task decoder which are sequentially connected in series; wherein the masking module is used for performing selective masking based on a high semantic information masking strategy on the preprocessed image; constructing a segmentation auxiliary task, and establishing an auxiliary task network in the segmentation auxiliary task to perceive a downstream segmentation task; the auxiliary task network comprises a segmentation task encoder and a segmentation task decoder; and performing joint learning on the mask reconstruction main task and the segmentation auxiliary task so as to optimize and train the main task network and the auxiliary task network at the same time. According to the method, the pre-training efficiency is improved, and the segmentation precision is improved under the condition that the downstream task complexity is not increased.

The embodiment of the invention discloses an adjustable capacity calculation method based on long-process steel production process modeling, which comprises the following steps: analyzing a long-process steel production process, and determining the material flow, equipment start-stop and energy consumption relationship of each link; drawing a resource task network block diagram by using a resource task network method, and depicting a coupling relationship among the links; based on the resource task network method, an optimization scheduling model is established, and the maximum adjustable capacity is calculated; the method can analyze the whole long-process steel production process, fully considers the incidence relation between links based on a resource task network method, determines key elements such as material flow, equipment start and stop, energy consumption and the like, considers various requirements, establishes a long-process steel production process model, can determine the adjustable maximum capacity of steel enterprises in each time period, and improves the production efficiency. And theoretical support is provided for iron and steel enterprises to participate in power grid dispatching.

The invention discloses a pig segmentation and posture detection method, device and equipment, and the method comprises the steps: obtaining pig farm image data, and carrying out the marking of a bounding box, a segmentation mask and a posture key point for each pig instance in each pig farm image data, so as to construct training data; a multi-task network model based on the YOLOv8 framework is constructed, and the network architecture of the multi-task network model comprises a backbone network, a neck feature fusion module, a detection head, a segmentation head and a key nodding head, and the detection head, the segmentation head and the key nodding head are arranged in parallel; inputting the training data into the multi-task network model, and training based on a preset loss function to obtain a trained pig segmentation and posture detection model; and inputting an acquired to-be-detected image into the pig segmentation and posture detection model for reasoning to obtain a segmentation and posture detection result corresponding to each pig. According to the method, the processing efficiency is remarkably improved, and the requirement of a real-time monitoring scene of a large-scale pig farm for low delay can be met.

The invention discloses a knowledge distillation method and equipment for nuclear accident deduction and fault diagnosis of a reactor. The method comprises the following steps: acquiring fault data of a nuclear energy simulation device; building a multi-task network; dividing the acquired data set D into a training set, a verification set and a test set, and pre-training the multi-task network by using the training set; after pre-training is completed, parameters in the multi-task network are fixed, that is, a large-parameter pre-training model is frozen to serve as a teacher model; migrating the knowledge of the teacher model to the student model by adopting a method of multi-task proportion-by-proportion penetration knowledge distillation; and obtaining a physical quantity deduction result and a fault result of the target pipeline by utilizing student model reasoning. According to the method, the multi-task network based on the physical information neural network is established, fault diagnosis and parameter deduction in a plurality of reactor nuclear accidents can be processed at the same time, the correlation between different tasks is fully utilized, and the fault diagnosis precision and the deduction capacity for the development process of the nuclear accidents are improved.

The invention provides an information recommendation method and device, electronic equipment and a storage medium, and relates to the field of data processing, in particular to the field of artificial intelligence. According to the specific implementation scheme, feature extraction is carried out on user features of a target user, a user historical click sequence and resource features of candidate resources through a first processing network, and first fusion features are obtained; scene related information corresponding to a recommended scene of the target user is determined, feature extraction is performed on the scene related information through a second processing network to obtain a scene related vector, and the recommended scene is a new user scene or an old user scene; performing feature extraction on the first fusion feature and the scene correlation vector through a scene processing network to obtain a second fusion feature; performing multi-task prediction through a multi-task network according to the second fusion feature to obtain a prediction result of each task, the tasks including a click task and a conversion task; and determining a comprehensive score of the candidate resources according to the prediction result, and determining a resource recommendation list according to the comprehensive score.

The invention relates to a point cloud downsampling method with any sampling size, which belongs to the field of point cloud data, and comprises the following steps: S1, inputting original point cloud data, and determining a target sampling size range; s2, in the training process, each Epoch dynamically selects the sampling size; s3, generating offset through a point cloud feature extraction module, and generating a final sampling point cloud in combination with an original sampling point obtained by farthest distance sampling FPS; s4, calculating a point-to-point distance value between the sampling point cloud and the original point cloud, and taking the point-to-point distance value as similarity loss; s5, inputting the sampling point cloud into a downstream task network, and calculating task related loss; and S6, repeating the steps S1 to S5 until the model converges. Through one-time training, point clouds of any size can be sampled, limitation of fixed sampling size is avoided, and multi-scale feature learning of the point clouds is realized through different-dimension learning of point cloud features.

The invention provides a multi-agent collaborative cognitive calculation method based on point cloud feature marking, and belongs to the field of Internet of Vehicles. The method specifically comprises the following steps: firstly, inputting original point cloud data collected by a sensor of a vehicle intelligent agent into a point cloud feature mark generator, converting the original point cloud data into a one-dimensional point cloud feature mark sequence semantic perception dynamic encoder for encoding a mark sequence, and generating a feature sequence; packaging the feature sequence, the space coordinates and the pose information of the self-vehicle intelligent body into a message data packet; converting the feature mark coordinate space of the neighbor intelligent body into a coordinate system with the self-vehicle intelligent body as the center by a point cloud mark aggregation module, and generating a unified sequence; and the semantic perception dynamic fusion module carries out global context modeling and dynamic fusion on the unified sequence, corrects feature mark position deviation, generates a refined sequence, and inputs the refined sequence into a downstream task network for generating final prediction. According to the method, the perception robustness is obviously improved.

The invention discloses a multi-modal sentiment analysis method based on a comparative decomposition feature multi-task network, and belongs to the technical field of multi-modal sentiment analysis. The problems that in the prior art, information interaction between different sample modes is difficult to achieve through a traditional multi-mode emotion analysis method, and single-mode labels in a multi-mode emotion data set are generally lacked or manual labeling is inaccurate are solved. The method comprises the following steps: constructing a multi-modal sentiment analysis model, inputting a to-be-processed data set into the multi-modal sentiment analysis model for feature extraction and decomposition, and integrating obtained six decomposition features; according to the multi-mode sentiment prediction loss, the contrast learning loss and the single-mode sentiment prediction loss, completing a multi-mode sentiment analysis task, a contrast learning optimization task and a single-mode sentiment analysis task; and performing double-layer loss optimization on the multi-modal sentiment analysis model, constructing a multi-task loss function, and outputting a multi-modal sentiment analysis result. The method improves the accuracy of the multi-modal sentiment analysis model, and can be applied to multi-modal sentiment classification.