107results about How to "Good explainability" patented technology

The invention discloses a shear capacity analysis method for a reinforced concrete beam with stirrups based on machine learning, and the method comprises the steps: constructing a training set and a test set based on a shear capacity database of the beam with stirrups; establishing an interpretable prediction model by utilizing machine learning, setting hyper-parameters for the interpretable prediction model, and performing training; obtaining a shear bearing capacity prediction value of the samples in the test set by using the trained prediction model; and quantifying the contribution of each key anti-shearing characteristic to the anti-shearing bearing capacity predicted value, and analyzing and judging the reasonability of the anti-shearing bearing capacity predicted value output by the prediction model according to the anti-shearing theory. According to the method, the problems of high prediction difficulty, high discreteness, lack of reliable physical models, large calculation amount of high-precision numerical simulation and the like in machine learning are solved for the first time; the prediction model can accurately reflect the complex nonlinear relation between the shear bearing capacity and the influence factors, and meanwhile, the model reasonability is verified and improved in combination with a prediction result and priori knowledge, so that the man-machine interaction trust in practical application is improved.

The invention relates to the technical field of computer vision and deep learning, and relates to a deep forgery detection method based on mask supervision. According to the technical scheme, an image is input to a preprocessing module for preprocessing; the processed image is respectively input into a mask supervision module for extracting spatial domain features and a frequency domain feature extraction module for extracting frequency domain features; and the spatial domain features and the frequency domain features are connected and input to a classification module to obtain a final prediction result. The method has the beneficial effects that the attention of the model to the irrelevant region is supervised and corrected through the mask, and a pixel-level label is used for guiding the model to face the synthesis region of the face, so that more accurate extraction of spatial features is ensured; the depth forgery detection method based on mask supervision has remarkable advantages in the aspects of detection accuracy, robustness, generalization ability, calculation efficiency, interpretability, overall performance optimization and the like, and a new solution is provided for the field of depth forgery detection.

The invention relates to an unmanned aerial vehicle cluster formation decision making and hunting method based on reinforcement learning, and belongs to the technical field of multi-agent formation control and decision making. The expected position corresponding to each agent and the control instruction of the dynamic model of the multi-agent are obtained, the central node does not need to quickly realize the expected formation configuration given by the decision-making system in a distributed mode, and the formation control problem of the multi-agent is solved; secondly, the formation of the multiple agents can be visually manipulated to adapt to dynamically changing tasks, and the formation manipulation problem of the multiple agents is solved; according to an actual confrontation task, learning and evolutionary online autonomous decision-making of the multi-agent system is realized, and the formation autonomous decision-making problem of the multi-agent system is solved; and finally, the interpretability and safety of learning decision are enhanced by utilizing a control model mechanism, and the interpretability and safety problems in the formation strategy learning process are considered.

The invention relates to the technical field of tumor image analysis systems, in particular to a multi-modal tumor image feature extraction and fusion analysis system and a method thereof.The multi-modal tumor image feature extraction and fusion analysis system comprises a data acquisition module, a multi-modal feature extraction and fusion module, a tumor prediction module and an optimization module; according to the method, the quality and robustness of feature extraction are effectively improved, and the problems of noise and artifacts in medical images can be better solved. In the feature fusion stage, an innovative weighted fusion strategy is adopted, complementary information between different modals is fully utilized, the problem of information redundancy is effectively avoided, the convolutional neural network and the bidirectional LSTM network are combined, spatial features can be captured, time sequence information can be effectively utilized, and the method has the advantages of being high in robustness and high in robustness. Therefore, the prediction accuracy of the tumor development trend is improved, the genetic algorithm is introduced to optimize network hyper-parameters, and the innovative optimization method enables the model to better adapt to specific data sets and tasks.

The invention relates to a transformer comprehensive safety assessment method and system based on symbol regression. The method comprises the following steps: acquiring monitoring variables of an electrical signal dimension, an equipment state dimension and an operation environment dimension of a transformer; based on the maximum mutual information coefficient, screening the dimension monitoring variables to obtain a plurality of monitoring variable sets with closely associated variables; performing sign regression on the plurality of monitoring variable sets to obtain a nonlinear function expression between any reference monitoring variable and other monitoring variables in the monitoring variable sets, namely a mutual check function, and realizing sign regression by using a genetic algorithm; establishing a single-variable fault probability function according to the influence of the multi-dimensional monitoring variable on the transformer equipment; and building a system dynamics model based on the mutual verification function and the fault probability function to realize comprehensive safety evaluation of the transformer. Compared with the prior art, the method has the advantages of improving the safety assessment level of the transformer, being high in interpretability and the like.

The invention provides a multi-process-mode intelligent switching method and system for vehicle-mounted sludge treatment, and the method comprises the steps: carrying out the time-space alignment and standardization of multi-dimensional sensing data such as sludge characteristics, environmental parameters and equipment states, carrying out the real-time reasoning of a process switching critical value through combining with a lightweight incremental regression model, generating a self-adaptive threshold value band with a confidence interval, and carrying out the real-time reasoning of the process switching critical value. Dynamic monitoring and mode switching pre-judgment of actual operation parameters are realized; in the switching process, a weighted fusion progressive transition control strategy is adopted, the transfer rate is adaptively adjusted according to real-time feedback, the operation stability is guaranteed, and the intelligence, reliability and switching smoothness of process switching of the vehicle-mounted sludge treatment system are remarkably improved.

The invention relates to an SQL (Structured Query Language) generation method and device, computer equipment, a computer readable storage medium and a computer program product. The method comprises the steps of determining supplementary information corresponding to an initial problem based on the initial problem, and obtaining a target problem; based on a pre-configured label system, determining a first fine-grained label of each entity contained in the target question, and performing association matching on the first fine-grained label and a second fine-grained label of each field in each preset database table to obtain a candidate table; according to the method, the target problem is converted based on the candidate table to obtain the conversion problem, and the conversion problem is processed through the preset large language model to obtain the SQL statement, so that knowledge enhancement can be performed on the problem directly input by the user, the definition of problem description can be improved, the efficiency of data retrieval can be optimized, and the interpretability of data can be optimized; in addition, alignment processing can be carried out on text descriptions of different semantic spaces, and the accuracy degree and the generation efficiency of SQL statements generated by a large language model are improved.

The invention discloses an enterprise knowledge graph inference completion method and inference completion system, and the method comprises the steps: obtaining a to-be-completed knowledge graph of a target enterprise, inputting the to-be-completed knowledge graph into a trained knowledge graph embedding model, and predicting an incomplete triple to achieve the inference completion of the knowledge graph; wherein the knowledge graph embedding model comprises an entity node coding module, a multilayer graph convolutional neural network module, a Softmax pooling layer and a gating neural network based on a self-attention mechanism. According to the method, the knowledge graph embedding model can provide more accurate semantic information for feature interaction, multi-level feature interaction and aggregation of the triple on sub-tensor kernel granularity and global granularity can be realized, and hidden semantics behind entities and relationships of the knowledge graph can be more efficiently mined. Therefore, the knowledge graph completion accuracy, the semantic modeling capability, the interpretability and the like are obviously improved.

The invention provides a wired table recognition method based on morphology, and the method comprises the following steps: S1, carrying out the graying processing of all read-in to-be-detected three-channel images, obtaining a gray image, carrying out the binarization processing of the obtained gray image, and obtaining a binary image of an original image; and S2, carrying out corrosion and expansion operation on the binary image by using a detection core, and forming an Excel format file by identifying a text table format and the like in the image so as to solve the problem of a special-shaped table. The problem that in the prior art, table recognition is always a research difficulty in the field of document recognition is solved, and the method is high in running speed, high in stability and good in interpretability. And the method is innovative, and can be used for well processing conditions such as special-shaped tables, cross-page disconnection, text cross-line, text wireless frame package and the like. The method can be popularized, and has good compatibility for identification of different types of wired tables. The method can be deployed in a server, a personal pc (personal computer) and embedded equipment, and is high in practical applicability.

The invention discloses an attention neural network-based personnel ability automatic evaluation method and system. The method comprises the following steps of: dividing an ability evaluation data set # imgabs0 # into a training set # imgabs1 #, a verification set # imgabs2 # and a test set # imgabs3 #; setting a plurality of different neural network configurations to form a configuration set # imgabs4 #; the obtained hyper-parameter # imgabs5 # is selected, and a neural network model # imgabs6 # is initialized; using a training set # imgabs7 # to train a neural network model # imgabs8 # obtained through initialization; after a # imgabs9 # round is carried out, the neural network model trained in the # imgabs10 # round is recorded as # imgabs11 #, a verification set # imgabs12 # is used for evaluating the trained model, and the neural network model and the evaluation result after each round of training are stored; the model obtaining the highest evaluation index on the verification set in N rounds is selected to be tested and stored on a test set # imgabs13 #; and until all hyper-parameters are used for initializing the neural network models and tested by a test set # imgabs14 #, deploying the neural network model with the highest model evaluation result for personnel ability grade evaluation work. And the requirement on multi-dimensional personnel ability portraits in a personnel evaluation scene is met.

The invention discloses an enhanced generation query method and system based on multi-granularity retrieval and reinforcement learning strategies. According to the method, semantic analysis is carried out on an original query by utilizing a large language model, the original query is decomposed into a plurality of sub-problems, and each sub-problem generates a query vector; in the multi-granularity vector knowledge base, respectively executing vector similarity retrieval on the query vector of each sub-problem to obtain a related document fragment vector of each sub-problem under each granularity, and further generating a multi-granularity vector set of each sub-problem; weighting the multi-granularity vector set of each sub-problem to obtain a comprehensive context information block, determining the weight of the multi-granularity vector set of each sub-problem through a multi-arm reinforcement learning strategy, and inputting the comprehensive context information block into a large language model to output a query result. According to the method, the retrieval accuracy is improved through semantic decomposition and multi-granularity knowledge collaborative retrieval, results are intelligently integrated through a self-learning mechanism, and lightweight self-adaptive optimization is achieved.

According to the interpretable intelligent flow control method, an automatic differential mechanism in a neural network is introduced into linear genetic programming, an explicit expression obtained by the linear genetic programming is automatically adjusted through a gradient descent algorithm, and the time needed by control law optimization can be greatly shortened; compared with a neural network control law, the explicit symbolic expression control law has higher interpretability, so that researchers can better understand a control mechanism and find new physical knowledge; the framework can set state sensors and different reward functions according to different control targets, and has very high universality.

The invention belongs to the technical field of diagnostic model construction, and particularly relates to a diagnostic model for sarcopenia of cardiovascular disease patients and a construction method. Comprising the steps that health related data of a tester are collected, the health related data serve as candidate variables, key factors related to SP are obtained through screening with a multi-dimensional feature selection method, and the key factors comprise continuous variables and classification variables; performing standardization or normalization on the continuous variables to obtain processed continuous variables, and performing One-Hot on the classification variables to obtain processed classification variables; constructing an extreme gradient enhancement model according to the processed continuous variable and the processed classification variable; and performing parameter tuning on the extreme gradient enhancement model by adopting a grid search method in combination with a five-time cross validation method to obtain a diagnosis model. The selected sample has representativeness, and the constructed model is high in diagnosis accuracy.

The invention discloses a concrete faced rockfill dam deformation prediction method based on CBMA and DE-MCMC algorithms, and the method comprises the steps: constructing a concrete faced rockfill dam deformation monitoring model according to the water pressure, temperature and aging component, replacing the original uniform distribution assumption of averaging a Bayesian model with the Copula function joint distribution between the settlement amount of a measurement point and an influence factor, and predicting the deformation of the concrete faced rockfill dam. A concrete faced rockfill dam monitoring model and measuring point data are substituted into an average calculation system of a Copula Bayesian model, the model is improved in combination with a differential evolution Markov chain Monte Carlo algorithm, and the method is applied to deformation monitoring of the concrete faced rockfill dam. According to the prediction method based on the CBMA and DE-MCMC algorithms, the uncertainty of the model is subjected to better quantitative analysis in the process of predicting the deformation of the rockfill dam, the deformation rule and characteristics of the rockfill dam can be analyzed more effectively, and good interpretability and actual measurement fitting of a statistical model are reserved.

The invention provides a data link network guarantee demand processing method and system, and the method comprises the steps: setting a database, integrating a plurality of AI algorithm models based on service calling, and achieving the knowledge extraction, knowledge recommendation and knowledge query of a data link network guarantee demand; the database comprises a plurality of guarantee schemes and sample knowledge maps corresponding to the guarantee schemes; the AI algorithm model at least comprises a knowledge extraction model, a knowledge recommendation model and a knowledge query model; each AI algorithm model is packaged in a service form; creating a process pool, wherein one process in the process pool correspondingly loads one AI algorithm model; and selecting an idle process from the process pool through the service request to call the AI algorithm model, and feeding back a call result of the AI algorithm model through the service response. According to the method, extraction, recommendation and query are integrated, an intelligent auxiliary capability is provided for data link network design planners when data link network guarantee requirements are acquired and analyzed, and the working efficiency is effectively improved.

The present invention discloses a method and system for intelligently identifying a person falling down based on human skeleton points, and relates to the field of computer vision technology, including: using a human body detection model to judge the real-time acquired surveillance camera video frame, if the human body detection model judges that there is a human body in the surveillance camera video frame, then intercepting the human body area image; extracting the position coordinates of key human body joint points in the human body area image through a human body joint point detection model; based on the position coordinates of the key human body joint points, through a combined tuning algorithm of position judgment, posture judgment and body folding judgment, calculating the distance, angle and relative position relationship between the key human body joint points, and comprehensively judging whether the human body presents a falling action. The present invention provides an efficient, accurate and practical method for detecting a person falling down by combining human body detection, skeleton point extraction and multi-dimensional judgment algorithms, which provides strong support for improving personnel safety and quality of life.

The invention discloses an scRNA-seq data interpolation method based on a graph convolutional neural network and statistical model constraint. The scRNA-seq data interpolation method comprises the following steps: 1) data preprocessing; 2) graph information construction and pre-label acquisition; 3) carrying out GCN-based preliminary interpolation; and 4) carrying out statistical model constraint based on negative binomial / zero-expansion negative binomial distribution. The method is high in interpolation precision and practical value and good in downstream analysis effect, and a new thought and a new method are provided for interpolation of scRNA-seq data.

The invention provides a cognitive disorder risk prediction method and device. The method comprises the following steps: acquiring a feature value of at least one pre-selected user feature of an assessed person; inputting the feature value of the at least one pre-selected user feature of the assessed person into a cognitive disorder risk prediction model obtained by pre-training, and obtaining a cognitive disorder risk prediction result of the assessed person output by the cognitive disorder risk prediction model; wherein the pre-selected user features are screened out from a plurality of user features, and the contribution degree of the pre-selected user features to the cognitive disease risk prediction precision is greater than a preset threshold value. According to the method and the device, the cognitive disease risk prediction is carried out by utilizing the pre-screened user features with relatively high contribution degree to the cognitive disease risk prediction precision, so that a relatively accurate prediction result is realized by utilizing relatively few user features, the prediction efficiency is improved, and the prediction result has relatively good interpretability.

The invention provides a three-dimensional fluorescence spectrum analysis method based on unsupervised contribution modeling. The method comprises the following steps: acquiring an original three-dimensional fluorescence spectrum of a sample; carrying out independent component analysis on the three-dimensional fluorescence spectrum, and reconstructing the spectrum by utilizing an analysis result; and carrying out modeling on the reconstructed spectrum by using a multivariate curve resolution-alternating least square algorithm to obtain an analysis report. According to the method, interference signals such as scattering in an original three-dimensional fluorescence spectrum are automatically removed through combination of two two-dimensional unsupervised modeling algorithms, then modeling is carried out on a reconstructed spectrum, dependence of three-dimensional fluorescence spectrum analysis on data, tools, platforms and personnel is reduced, good fault tolerance and accuracy are guaranteed while flexibility is improved, and the method is suitable for popularization and application. And reliable technical support is provided for water quality monitoring.

The invention discloses an action quality evaluation method based on video and skeleton bimodal fusion, and the method comprises the steps: inputting a video frame sequence containing a complete action process into a fragmentation network, and generating continuous fragments; secondly, inputting each video clip into a video modal feature extraction network and a human body skeleton feature extraction network, respectively obtaining a video modal feature and a skeleton modal feature, fusing the video modal feature and the skeleton modal feature through a cross-modal attention mechanism, and generating a fused clip-level feature representation; and finally, inputting each segment-level feature representation into a multi-layer Transform encoder, obtaining a global time sequence feature of the action through a self-attention mechanism, inputting the global time sequence feature into a multi-layer perceptron (MLP) regression network, and generating an action quality score. According to the method, a prediction result is ensured to be more stable in numerical value and more accord with human perception in semantics, and the accuracy and consistency of action quality evaluation are remarkably improved.

The invention discloses a fleet behavior prediction method based on a potential function and a Bayesian network and a medium. The fleet behavior prediction method comprises the following steps: establishing a topological relation of an unmanned aerial vehicle fleet; constructing an unmanned aerial vehicle potential energy field function and external regulation and control force; setting an adaptive obstacle avoidance distance, and establishing collision avoidance constraints, planning track continuity constraints and dynamics feasibility constraints; dividing the cluster by using a K-means algorithm; constructing a dynamic Bayesian network, and predicting a cluster behavior; setting a behavior prior probability; preprocessing the data to obtain required information; dynamically adjusting the weight of each behavior probability; and updating and normalizing the behavior probability. According to the method, cluster division and behavior identification results can be given under discrete data input (single moment) and continuous data input (multi-moment) conditions, good convergence time and stability are achieved, the requirement for stable output of a behavior identification algorithm in a flight scene is met, and meanwhile the behavior identification efficiency is improved. And algorithm identification results in a specific flight scene can be traced.

The present invention discloses a method and electronic device for fine-grained depression recognition of interview texts based on a large language model and an emotion dictionary, and belongs to the field of natural language processing detection. The method includes analyzing the interview text using a large language model ChatGLM3‑6b in conjunction with an instruction prompt template to generate a restatement text; constructing an emotion dictionary, and using the bart‑large‑mnli pre-trained language model to extract the emotion features that match the restatement text with the emotion dictionary to obtain an emotion feature matrix at different granularities; constructing a deep neural network model, and training the deep neural network model by selecting the emotion feature matrices corresponding to different emotion vocabularies to obtain a fine-grained depression recognition result. For simpler binary depression classification tasks, more complex fine-grained depression classification tasks, and regression tasks, emotion subsets of different scales are flexibly selected to adapt to the difficulty of the task.

The invention relates to a visualization processing method and device based on an attention mechanism and a storage medium. The visual processing method based on the attention mechanism and applied to the electronic equipment is characterized in that the visual processing method based on the attention mechanism is applied to the electronic equipment and comprises the steps that an original text C and a question Q are spliced and combined, and first input information is obtained; inputting the first input information into a question and answer model to obtain answer information A, output result information T of the question and answer model and an attention matrix Attenm; giving a focusing range, and calculating an attention matrix in the focusing range; based on the attention matrix in the focusing range, a heat map is obtained through calculation, and visualization processing is carried out on output result information T of the question and answer model.

The present application relates to a multi-modal large model-based industry electricity abnormality intelligent analysis method and system, comprising the following steps: S1: obtaining power system multi-source data and preprocessing to obtain standardized time series data set; S2: based on the standardized time series data set, feature extraction is carried out, and a feature data set is constructed; S3: based on the feature data set, an industry electricity abnormality knowledge graph is constructed in combination with a multi-level correlation relationship network; S4: a multi-modal abnormality detection model is constructed by fusing time series Transformer, graph neural network and knowledge embedding module, and is trained based on the feature data set and the power knowledge graph; S5: the trained multi-modal model is deployed on a distributed stream processing platform to realize real-time abnormality detection and multi-dimensional intelligent analysis; S6: the real-time abnormality detection and multi-dimensional intelligent analysis results are visualized. The present application can realize all-round intelligent monitoring of industry electricity abnormalities facing complex industry actual demands.

The present application relates to the technical field of short-circuit current calculation, and particularly relates to a flexible DC system short-circuit current calculation method based on a physical information neural network, comprising: setting a model input feature vector, the model input feature vector being used to represent a system operating state before a fault and fault information, and performing data preprocessing on the model input feature vector; constructing a hybrid driving calculation model, the hybrid driving calculation model comprising a physical calculation module and a neural network module, and being coupled based on a preset fusion architecture; performing end-to-end training and optimization on the hybrid driving calculation model by using a preset composite loss function; and performing flexible DC system short-circuit current calculation based on the optimized hybrid driving calculation model, so that the problems of poor precision, speed and convergence, poor interpretability and weak generalization ability in the prior art are solved.

The application discloses a machine learning algorithm for screening of metal elements in a diatomic catalyst full cycle based on product yield prediction and application thereof, and the algorithm comprises the following steps: (1) collecting attribute features of metal elements in the diatomic catalyst as a feature space 1; (2) preliminarily screening the attribute features according to comprehensive scores of feature importance to obtain a feature space 2; (3) performing feature operation processing on the feature space 2 to generate a multivariate descriptor space matrix; (4) performing dimension reduction processing on the multivariate descriptor space matrix by combining a gradient boosting regression model with a recursive feature elimination algorithm; (5) further screening to obtain an interpretable descriptor by using a least absolute shrinkage and selection operator; and (6) applying the interpretable descriptor to a random forest regression model to screen a catalyst for a propane dehydrogenation reaction to prepare propylene. The application can solve the problem that a traditional descriptor is difficult to directly predict a catalytic product yield.

The application discloses a new correlation image recognition and machine learning method based on fuzzy monotony, and belongs to the technical field of artificial intelligence of pattern recognition and machine learning; the method is defined as FMMCA, which evaluates local fuzzy monotone correlation by comparing row vectors and column vectors of an image matrix pair by pair, then the local correlations are weighted and accumulated, and finally the global fuzzy monotone correlation between images is obtained. The application directly uses fuzzy monotone correlation analysis to replace classical correlation analysis for multi-view research, so that the problems existing in classical correlation analysis do not exist, and the fuzzy monotone method feature does not need to be measured statically by distance, but can be measured dynamically by interval change, so that the influence of some noise is offset, the performance is improved, a new fuzzy monotone machine learning method is formed, the optimization of a focus loss function is not needed, the parameters are few, the robustness is good, and the computing power is small.

The application is a kind of automation testing and application fluency testing method based on control matching, relating to the field of Android end automation testing.The application method comprises: setting a test script containing behavior simulation and fluency determination in advance, reading the corresponding test script after the Android end receives a test command; acquiring interface control structure positioning target control, executing simulation operation; starting timing after each simulation operation, and acquiring all control elements of the current interface when reaching the set time threshold, judging whether the interface is loaded or not in combination with the interface structure information recorded in the fluency determination script; obtaining application fluency determination result according to the interface loading condition at different time thresholds and returning to the server end.The application does not need to acquire ROOT permission, does not need to connect PC end for testing, is safer, is more convenient to use, reduces the adaptation cost, and is higher in testing accuracy and testing speed.

The present invention relates to the technical field of abnormal decision making, and discloses a lightweight intelligent abnormal decision making method based on multi-scale feature fusion difference, comprising selecting multiple normal images as memory samples; obtaining an image to be detected; obtaining multiple stage features with different resolutions of each memory sample and the image to be detected, forming sample storage information of each memory sample and real-time storage information of the image to be detected; calculating the Euclidean distance between the real-time storage information and each sample storage information, obtaining the smallest distance as the target sample storage information and the target memory sample; obtaining the x-axis and y-axis convolution stage features of the target memory sample and the image to be detected at each resolution, and calculating the differential information between the target memory sample and the image to be detected at each resolution; obtaining the differential output of each layer in the multi-scale feature difference based on the pyramid structure, performing channel splicing, and inputting it into a decoder to obtain the defect positioning result in the image to be detected.

This invention discloses a method, apparatus, device, and storage medium for depth extraction of the focus stack weight domain. The method includes: acquiring a multi-focal plane image sequence; calculating focus sharpness pixel-by-pixel and normalizing it along the focal plane to generate an initial weight map; using the original image as a structural reference to perform edge-preserving filtering on the initial weight map to obtain an optimized weight map; constructing a weight-focal plane curve based on the optimized weight map, and sequentially performing three levels of logical judgment: the first level removes textureless regions based on the maximum weight value; the second level removes high-reflectivity false peaks based on waveform flatness; and the third level filters candidate focal plane positions based on peak focus sharpness; and performing sub-pixel fitting on the selected candidate focal plane positions to convert them into physical depth and generate a 3D point cloud. This invention effectively solves problems such as false points on textureless backgrounds, false peaks on highly reflective surfaces, edge breaks in weak textures, and limited depth accuracy through edge-preserving filtering to optimize the weight map, three-level progressive judgment, and sub-pixel fitting.