34results about How to "Solve the sparse problem" patented technology

The invention provides a predictive maintenance management method and system for an industrial robot, and relates to the technical field of equipment maintenance management, and the method comprises the steps: obtaining multi-modal heterogeneous data of the industrial robot, carrying out the intelligent preprocessing, obtaining first preprocessing data, carrying out the cross-modal correlation fusion of the first preprocessing data, generating fusion data, and carrying out the prediction of the industrial robot; extracting complementary redundant information and constructing a semantic feature map; constructing a health state prediction model, adding the semantic feature map into the health state prediction model, mapping the semantic feature map to an embedding space through linear transformation, extracting time sequence information, carrying out feature interaction to obtain interaction features, and carrying out transformation and feature enhancement through a feedforward neural network layer to obtain health state probability distribution; comprehensive health state distribution is generated through Bayesian reasoning fusion, an optimal maintenance strategy is generated in combination with a detection model, an abnormal mode and a key turning point, collaborative maintenance is conducted on the industrial robot, a maintenance result is uploaded to a server, and the current state is stored.

The invention discloses a multi-level track disease recognition system based on vehicle body vibration data, and particularly relates to the field of blasting auxiliary devices. Comprising the steps that a data acquisition module acquires heterogeneity data of a track environment; the environment adaptive module analyzes the correlation between the vehicle body vibration data and the heterogeneity data, constructs a dual-view similarity matrix, and realizes adaptive classification of the track environment through a spectral clustering method; the label enhancement module generates high-quality pseudo labels and forms a completely labeled sample set; the disease recognition module adopts an attention-guided multi-task cascade CNN architecture, and outputs a multi-level recognition result containing disease positions and severity. By the adoption of the technical scheme, the problem that an existing track state evaluation method cannot evaluate the real state of track diseases in a complex environment is solved, and good performance is achieved in the aspect of geometric and structural disease recognition.

The invention discloses a large language model knowledge graph completion method based on meta-cognition and multi-angle prompt, relates to the technical field of knowledge graph completion, and solves the problem of magical property caused when knowledge graph completion is solved by adopting prompt engineering in the prior art. The knowledge graph completion method comprises the following steps of performing data extraction on a knowledge graph data set in a medical database, constructing an original knowledge graph structure of medical data according to the extracted data, performing initial embedding on the original knowledge graph structure, and enhancing embedding generation; optimizing the large language model by adopting a meta-cognition prompting method and a multi-angle prompting method; and enriching the enhanced and embedded original mapping knowledge domain structure by adopting the optimized large language model, wherein the enriched mapping knowledge domain contains new relationships and entities. The method is suitable for completing the knowledge graph in the medical database in the medical field.

The invention discloses a depression detection method based on image-text social media analysis, relates to the technical field of automatic detection of depression, and solves the technical problem in the prior art that whether a user is a depression patient or not is detected by analyzing character and picture data published by the user on a social media platform. Specifically, the method comprises four steps of data capture, post screening, image-text representation and fusion classification. Generally speaking, the method comprises the following steps: firstly, capturing a posted document or posted post of a user in a certain time period from a social media platform, including characters and picture data in the posted document; then, performing depression related symptom prediction on the posted document, and screening out the most related K posts; then, respectively carrying out text and picture representation on the screened posts; and finally, fusing the characterization vectors of the two modes for depression detection.

The invention provides a handheld bank personalized search recommendation method and system, electronic equipment and a storage medium, and is suitable for a handheld bank APP to obtain a pre-constructed collaborative knowledge graph; obtaining a historical user query, and extracting a historical query suggestion set related to the historical user query from the collaborative knowledge graph; forming a plurality of training triads according to each user in the user set, the historical user query and a historical query suggestion set related to the historical user query, and training a to-be-trained personalized search recommendation model according to a scoring function learned by a training set constructed by each training triad to obtain the personalized search recommendation model; and when a query currently input by a target user in the search session of the handheld bank APP is detected, calculating the recommendation degree between the query and each entity in the handheld bank APP by using the personalized search recommendation model, screening out to-be-recommended entities from the entities according to the recommendation degree of each entity, and outputting each to-be-recommended entity.

The invention discloses a three-dimensional target detection method, device and equipment based on an improved Pilllar structure and a medium, and relates to the field of three-dimensional point cloud processing and target detection. Dividing the original LiDAR point cloud data into a plurality of Pill units through a height perception double-branch coding module of the three-dimensional target detection model, and carrying out parallel branch feature coding to obtain a pseudo image feature map; global maximum and average pooling is carried out through a space attention feature extraction module to generate a space attention weight map, then the space attention weight map is multiplied by a pseudo image feature map point by point to obtain an enhanced weighted feature map, and the robustness of the model in a complex background and other scenes is enhanced; multi-scale context feature extraction fusion and receptive field expansion are carried out through the multi-scale feature extraction module, details can be seen clearly, the global context can be understood, and the problem that a long-distance target is difficult to detect due to sparse point clouds is solved; target prediction is carried out through the SSD detection head module, and the real-time requirements of applications such as automatic driving are met.

The invention discloses a single-view point cloud reconstruction method for tree branches in a real scene. The method comprises the following steps: acquiring a single RGB image of the tree branches in the real scene; performing single-view depth estimation based on loop prediction on the single RGB image of the tree branch to obtain a plurality of depth images; performing depth image-to-point cloud conversion on the plurality of depth images to obtain sparse point clouds; carrying out sparse point cloud densification on the sparse point cloud to obtain a dense point cloud; performing point cloud coordinate fine correction on the dense point cloud to obtain a reconstruction result; and reconstructing the tree branch point cloud in the real scene based on the reconstruction result. The invention further discloses a corresponding system, electronic equipment and a computer readable storage medium. The problem that the depth information of the shielded branches cannot be effectively estimated is solved through single-view depth estimation based on loop prediction; the tree morphological feature analysis-based point cloud denseness algorithm and the point cloud correction method based on point-by-point correction solve the problem that the preliminarily obtained tree branch point cloud is sparse and inaccurate.

The invention discloses a characterization alignment space-based drug-target interaction prediction method and device. The method comprises the following steps of: initializing parameters of a protein characterization dimension, a drug characterization dimension, a public embedding characterization dimension, the number of layers of a multi-layer sensor and a batch dimension; acquiring a plurality of single-source network data through a plurality of pharmaceutical databases, and performing comparison processing to construct a biological medicine heterogeneous network; according to a comparative learning strategy, integrating a protein target and drug characterization into a public embedding characterization space, and carrying out DTI association training to set a model; and inputting the protein targets and the corresponding drug representations into the trained setting model for DTI prediction to obtain a plurality of most relevant drugs. According to the method, by constructing the biological medicine heterogeneous network, the DTI association related to unknown medicines can be predicted; an embedded characterization alignment frame is built, a high-quality characterization method can be utilized to characterize a biological medicine network and a biological entity, and the problem of training data sparsity can be solved by utilizing contrast loss.

The invention provides a semi-supervised document classification method and system based on generative model GPT enhancement, and relates to the field of statistical classification, and the method comprises the steps: obtaining an initial first document set, guiding a preset generative artificial intelligence model through a prompt sentence to generate an enhanced sample for the first document set, and obtaining a second document set; inputting the first document set and the second document set into a self-supervised learning model for training to obtain an initial classifier; and performing fine adjustment on the initial classifier through a few-sample learning method to obtain a target classifier. According to the method, the initial classifier is finely adjusted by using the few-sample learning method, few-sample learning is performed through the training set, so that the target classifier can predict new-class samples, the few-sample learning method is applied to document classification research, dependence on label documents is reduced, and the classification efficiency is improved. A new class data classification mode can be mined from existing basic class data, and an effective technical scheme is provided for solving the problem of sparse label data in enterprise documents.

The invention provides a route recommendation method based on a double-tower latent semantic model, and belongs to the technical field of route recommendation. The technical problems of data sparsity, difficulty in determination of traveler preference and scene dependence in personalized route recommendation are solved. According to the technical scheme, the method comprises the steps that S1, attributes, travel tracks and scores of travelers and road section attribute data are obtained; s2, segmenting a road network according to the track route; s3, using a double-tower latent semantic model to respectively recommend road sections in peak, flat and low peak periods; and S4, recommending an optimal travel route according to the travel starting point, the travel ending point and the travel time given by the traveler. The method has the advantages that traffic characteristics of different time periods are combined, hidden characteristics of travelers, category characteristics of the travelers, hidden characteristics of road sections and category characteristics of the road sections are comprehensively considered, and therefore personalized route recommendation service is provided.

The invention discloses an unmanned ship positioning and mapping method based on triple factor graph optimization. The method comprises the following steps: inputting laser radar data, inertial measurement unit data and global satellite navigation system measurement data into a first factor graph for optimization to obtain the pose of an unmanned ship; taking a key frame marginalized by a speedometer as the input of a second factor graph, and obtaining a local map, wherein the key frame comprises the pose of the unmanned ship; and taking the plurality of local maps as input of a third factor graph to obtain a global map. The invention provides a registration factor fused with light intensity information. The weighted sum of three weighted Gaussian distributions corresponding to the mean value of the light intensity is used for description, and a light intensity kernel function is established. According to the invention, by adding extra light intensity information, the problem of insufficient environment information is solved. In addition, a triple factor graph optimization framework is arranged, so that the accuracy and robustness of positioning and mapping of the unmanned ship can be effectively improved.

The invention provides a multi-modal relation extraction method and system based on meta-learning, and belongs to the technical field of text entity recognition, and the method comprises the steps: obtaining a to-be-predicted text image pair; and processing the obtained text image pair by using a pre-trained relationship prediction model to obtain a multi-modal relationship between the entities. According to the method, a multi-modal relation extraction framework based on meta-learning is provided and is suitable for any relation classification model based on gradient descent; initial parameters are learned from a head relation class through meta learning, then model parameters in a tail item feature learning process are finely adjusted, and task learning from easy to difficult is realized by utilizing priori knowledge learned from head data; the feature representation of the tail class is enhanced, the feature gap between the head class and the tail class is bridged, and the relation identification performance is improved; during rare category training, the weight of the positive gradient is dynamically improved, the weight of the negative gradient is dynamically reduced, and the problem of tail data sparsity is solved.

The application provides a high-quality large scene reconstruction method based on 3D Gaussian sputtering, comprising the following steps: dividing a large scene into multiple grid units; using a point cloud adaptive interpolation module to refine point cloud points within the boundary of each grid unit; using a global-local Gaussian decoder to calculate 3D Gaussian parameters within and outside the boundary of each grid unit; using a Gaussian refinement module to perform refinement processing on the 3D Gaussian distribution; and using a weighted average method to combine the 3D Gaussian distributions of all grid units into a complete large scene 3D Gaussian distribution, and finally generating a complete large scene three-dimensional model. The high-quality large scene reconstruction method can improve the consistency and quality of large scene reconstruction.

The invention provides a blasting vibration peak velocity prediction method fusing parameter uncertainty and data-driven optimization, which comprises the following steps of: firstly, acquiring data such as blasting parameters, lithologic indexes, geological conditions and actually measured vibration peak velocity (PPV), and establishing a basic database; a prior probability model is constructed, a joint uncertainty model is established in combination with probability disturbance and a fuzzy triangular number, a multi-dimensional disturbance sample is generated through a joint central value and a joint standard deviation and is fused with original data, and an extended database is formed. Feature analysis is carried out on the fused data, and input variables which have obvious influence on the vibration peak velocity (PPV) are screened; and constructing a BP neural network model based on the screened features, carrying out global optimization on a network weight and a threshold by adopting a PSO algorithm, and then carrying out local fine tuning by utilizing Adam. Finally, through training and verification, prediction of the vibration peak velocity (PPV) is realized, and model precision is evaluated through RMSE, MAE, MAPE, Rand other indexes. The influence of rock and soil parameter uncertainty on prediction precision can be effectively processed, and the reliability and applicability of blasting vibration prediction are improved.

The present application discloses a sentence recognition method, device and computer-readable medium, which obtains an original combination of multiple words arranged in word order; inputs the original combination into a target N-gram grammar model, and if the original combination contains a name in a list of names of people and places, the target N-gram grammar model replaces the name contained in the original combination with a corresponding category label in the list of names of people and places, obtains the replaced combination, and calculates the probability associated with the replaced category label, and obtains the probability of the original combination as a sentence according to the probability associated with each word in the replaced combination, the probability associated with the category label, and the proportion of all names under the corresponding category label in the list of names of people and places. Compared with the existing N-gram grammar model, the target N-gram grammar model of the present application makes it easier to correctly recognize sentences containing names of people or places.

This invention discloses a robust 3D hair reconstruction method with adaptive Gaussian ellipsoids and complex lighting, relating to the fields of 3D vision and computer graphics. Addressing the problems of sparse Gaussian ellipsoid distribution, unreliable orientation supervision signals, and geometric and texture coupling interference in existing multi-view hair reconstruction methods under non-ideal lighting, this invention employs a two-stage approach. It utilizes anisotropic Gaussian units to implicitly encode hair orientation and directly transfers the photometric constraint gradient to hair nodes through hair-aligned Gaussian dual representation, achieving efficient and high-precision hair-level reconstruction. Simultaneously, an adaptive lighting preprocessing flow is constructed, using a visual language model to locate low-quality viewpoints and dynamically setting the upper limit of Gaussian density. A three-stage adaptive density strategy is used to fill geometric gaps. Decoupled two-stage optimization is employed to eliminate lighting bias and restore true hair color. Furthermore, confidence-weighted orientation loss and 3D interpolation-guided upsampling are used to improve the orientation field quality and constraint coverage.

The invention relates to the technical field of computer and big data management, and discloses a data interaction method of a handheld tablet personal computer, which comprises the following steps: receiving a big data query instruction, and obtaining a high-dimensional data set; a three-dimensional point cloud is generated through graph neural network dimension reduction embedding; constructing and rendering a dynamic interactive three-dimensional topological mapping model; responding to the multi-finger gesture operation to adjust the spatial layout and the clustering structure of the data entity in real time; and dynamically calculating and visually coding the association strength between the data. The system comprises a query receiving module, a data acquisition module, a dimension reduction processing module, a model construction module, a rendering module, a gesture recognition module, an association calculation module, a visual coding module and the like. According to the method, through three-dimensional space mapping and multi-finger gesture interaction, the information density and association visibility are remarkably improved, local detail expansion, spatial memory anchor points and multi-user cooperation are supported, and efficient and visual big data exploration is achieved.

The present application discloses a retrieval method, device, equipment and storage medium, wherein the method comprises: based on a target database, searching for retrieval input information to obtain a first retrieval result; determining a user's browsing behavior list for the first retrieval result; determining an expression tag corresponding to each browsing behavior in the browsing behavior list; generating a user's interest point feedback coefficient based on the browsing behavior list and each of the expression tags; and reordering the first retrieval result based on the interest point feedback coefficient to obtain a second retrieval result. In this way, personalized retrieval recommendation results can be provided to users, so that the user's expected retrieval results can be quickly and accurately located from massive information.

The present invention discloses a social user topic analysis method and system based on semantics and word expansion, which relates to the technical field of network user information assessment, and solves the problem that there is likely to be no contextual association between user posting information in the existing topic analysis method. The technical solution is: a short text topic model based on semantics and word expansion is proposed, which not only integrates the semantic information of the short text, but also expands word pairs through external corpus knowledge, while solving the short text sparsity problem, enhancing the topic tendency of the text; on top of the short text topic model based on semantics and word expansion, a topic weighting method based on the amount of interaction between users is adopted to distinguish the importance of posts, thereby calculating the topic distribution of users. Compared with the traditional topic feature analysis method, the social user topic analysis method of the present invention is more reasonable.

The invention discloses a target semantic segmentation method based on radar data and image data, and the method comprises the following steps: 1, carrying out the preprocessing of radar point cloud data: obtaining the radar point cloud data through a radar sensor, and carrying out the preprocessing of the radar point cloud data; 2, projecting the preprocessed point cloud data, and converting the point cloud data into a pseudo-depth map form; the method specifically comprises the following sub-steps: step 2-1, determining a projection plane; step 2-2, carrying out coordinate conversion; step 2-3, carrying out projection calculation; step 3, performing feature interaction: performing feature interaction on the radar data in the pseudo-depth map form and the input image data; and step 4, inputting the fused multi-modal features into a decoder network to generate a semantic segmentation result. By combining the advantages of the radar point cloud data and the image data, the invention provides a semantic segmentation method which can still keep high precision in severe environments such as low light, strong light, rain and fog and the like.

This application provides a method and system for intelligent identification of minerals in weak information areas based on transfer learning, relating to the field of transfer learning technology. First, a source domain mineral sample set and a set of mineral samples to be identified in weak information areas are obtained. Then, geological correlation elements between the two are extracted and a cross-domain feature mapping is constructed. The original features of the minerals in the weak information areas are subjected to transfer adaptation processing to generate intermediate features of the minerals in the weak information areas that are adapted to the feature dimensions of the source domain. Next, sparse feature dimensions in the intermediate features are identified, and the corresponding feature distribution patterns of the source domain are used to complete the features, generating completed features. The completed features are input into a pre-trained transfer learning identification model for cross-domain identification, generating preliminary identification results. Finally, source domain samples with similar geological correlation elements are selected to determine the transfer deviation, correct the preliminary identification results, and generate the final identification results, effectively improving the accuracy and reliability of mineral identification in weak information areas.

The invention relates to the technical field of geophysical exploration, in particular to an open space well complex geology refined modeling method. The invention provides a brand-new open space well complex geology refined modeling method in a tunnel exploration stage. A point set can be effectively optimized by interpolation according to the perimeter proportion based on the W configuration, the problem of sparsity and winding distribution discretization is solved, meanwhile, the winding difference of adjacent profile contour feature points is remarkably weakened, and a foundation is laid for enhancing the constraint of the feature point set on the profile shape and reconstructing the similarity between triangles in a triangulation network. The three-dimensional physical property model and the geological anomaly model of the research area can be efficiently and accurately established, the crushing and water-rich anomaly interpretation of the area is realized, and data reference is provided for unfavorable geology advanced prediction and tunnel safety construction.

The invention discloses a big data intelligent recommendation method and system based on multi-scene dynamic adaptation. A dynamic user portrait is generated by constructing a three-dimensional scene feature space and fusing multi-source heterogeneous data; scene-feature weight adaptive adjustment is realized based on reinforcement learning, and a multi-scene model is trained in combination with a'shared-private 'neural network and transfer learning; and finally, dynamically allocating resources according to the scene priority to generate a recommendation result, and continuously optimizing through a feedback closed loop. The problems that a traditional recommendation system is insufficient in scene adaptation, poor in real-time performance and the like are solved, the recommendation precision and the resource utilization rate are remarkably improved, and the method is suitable for multiple fields such as e-commerce and content distribution.

The application discloses a kind of infrared interference signal equal optical path difference interval reconstruction methods based on optical path difference model, belong to Fourier infrared spectrometer signal processing technical field.Method includes: to reference signal is filtered and transformed, and the extraction index of quarter wavelength interval is obtained by peak searching;Linear region and nonlinear region are determined using the Taylor expansion of optical path difference mathematical model;Infrared interference signal is directly sampled or interpolation correction in linear region at equal interval, and high times interpolation reconstruction is carried out in nonlinear region.The present application makes full use of the optical path difference mathematical model of rotating mirror spectrometer, overcomes the defects that the existing method only relies on reference signal, resulting in less sampling points, low precision, improves the sampling density and spectral inversion accuracy of equal optical path difference interval reconstruction.

This utility model relates to a coating device for producing adhesive bandages, including a storage tank. The top and bottom of the storage tank are respectively provided with an inlet and an outlet. The bottom of the outlet is connected to a spray tube, and the bottom of the spray tube is provided with a coating cover plate. Injection parts are respectively provided at both ends of the coating cover plate. Both injection parts are connected to the spray tube. A conveyor belt is arranged below the storage tank. The surface of the conveyor belt is provided with a coating mold base. The coating cover plate is tightly fastened to the coating mold base. This application adopts a combination design of a spray plate with dual injection parts and rectangular array spray holes to replace the conventional injection part glue injection process. The reserved gap between the spray plate and the coating cover plate can form a stable air pressure field. With the precise control of the pneumatic spray valve, the glue is evenly distributed throughout the entire area through the array spray holes, effectively solving the problem of local accumulation or sparseness in traditional single injection part coating, ensuring the consistency of adhesive force of the adhesive bandage, and significantly reducing the positioning deviation rate of the coating area.

The application discloses a weaving method of a back-hanging fabric and the fabric, the fabric is woven by a circular knitting machine, the back-hanging fabric comprises at least one cycle unit, the cycle unit is first woven by hooking the face yarn in a positive direction through a needle cylinder, then is woven by hooking the back-hanging yarn in a positive direction and a reverse direction in sequence through the needle cylinder, is then woven by hooking the face yarn in a reverse direction through the needle cylinder, and finally is woven by hooking the back-hanging yarn in a positive direction and a reverse direction in sequence through the needle cylinder; the above technical scheme makes each face yarn weaving row have a back-hanging yarn weaving row formed by the positive direction and the reverse direction of the back-hanging yarn, further makes the back-hanging yarn weaving row in the fabric have no interval, solves the problem of fabric sparseness, and meanwhile keeps the original weaving positions of the face yarn and the back-hanging yarn unchanged, avoids many flying lines in the cycle weaving process, and even avoids the problem of entanglement.

The invention discloses a simulation enhanced cognitive recommendation method and system based on a large language model, which are used for carrying out data support on a sparse problem of a knowledge tracking dataset, and the specific method comprises the following steps: inputting real data obtained from a real world dataset into the large language model in a static simulation data generation module; outputting the generated data; real data features corresponding to the real data and generated data features corresponding to the generated data are input into a discriminator to output the probability that whether data in the generated data is the real data or not, through the constructed static simulation loss, a large language model is fixed firstly, the discriminator is trained, then the trained discriminator is fixed, and the probability that the data in the generated data is the real data or not is output. Training the large language model; inputting the real data and the generated data into a KT dynamic agent module, and outputting an answer prediction result at the next moment; according to the simulation enhanced cognitive recommendation method and system, the data set sparse problem and the cold start problem of the knowledge tracking model are solved.

The invention discloses a causal federation element reinforcement learning intelligent decision-making method and system for people-benefiting insurance claim settlement, and relates to the field of insurance intelligent claim settlement, and the method comprises the steps: building and adaptively generating a regional causal knowledge graph based on a multi-dimensional policy text; a causal federation element reinforcement learning framework is adopted, through element strategy network initialization, and federation learning is utilized to carry out fine adjustment on small samples in an encryption mode on the premise that data does not go out of a domain, so that local intention recognition model adaptation is realized; according to the claim consultation text of the user, performing intention recognition in combination with neural symbol double-track reasoning and knowledge graph causal chain matching; and training the decision tree, and generating a visual report through anti-factual reasoning to assist in decision making. According to the method, the user intention recognition accuracy and the rule interpretability are enhanced, the response adaptation efficiency of policy dynamic updating is improved, privacy is guaranteed on the premise that data are not out of a domain, small sample learning is achieved, then the people-benefiting insurance claim settlement processing efficiency and the customer satisfaction degree are optimized, and the operation cost is reduced.

The application discloses a traditional Chinese medicine technical element importance evaluation method and system based on patent citations, and relates to the field of traditional Chinese medicine information technology. The method comprises the following steps: obtaining the composition of the target traditional Chinese medicine compound and the monarch-minister-assistant role label of each medicinal material; constructing a technical element graph containing medicinal material layer nodes and component layer nodes; assigning an initial weight value to each medicinal material node based on the monarch-minister-assistant role; obtaining patent citation relationship data, including the citation edge between patent nodes and the mapping relationship between patent nodes and technical element nodes; calculating the final weight value of each technical element node based on the initial weight value and the citation frequency information; and outputting according to the final weight value sorting. The application solves the problems of strong subjectivity and sparse objective data in traditional Chinese medicine technical element importance evaluation by combining the subjective weight of the monarch-minister-assistant theory with the objective data of the patent citation relationship, and the evaluation result can be applied to FTO retrieval strategy generation, patent navigation analysis and other scenes.

This invention discloses an immediate reward learning method based on self-supervised reinforcement learning, belonging to the field of artificial intelligence reinforcement learning technology. First, the invention calculates a first prediction error between the agent's predicted state and the actual state. Second, it reconstructs action data using an inverse dynamics model and calculates a second prediction error. Then, it constructs a causal confidence factor based on the second prediction error to generate an effective prediction error. Next, it generates a fast-channel reward and calculates the volatility of the reward signal using an oscillation index. Finally, it generates the final immediate reward. By integrating a multi-layered mechanism of prediction error correction, causal relationship modeling, reward stability assessment, and policy stochastic adjustment, the stability and reliability of the reward signal can be effectively enhanced, noise interference reduced, and training oscillations avoided. Simultaneously, it significantly improves the quality of immediate reward generation in sparse reward environments, accelerates policy convergence, and enhances the efficiency and stability of self-supervised reinforcement learning in complex tasks.