19 results about "Decision area" patented technology

The invention discloses an anti-fact generation method for processing class imbalance based on a real sample. The method comprises the following steps: preprocessing input data; performing causal feature selection by adopting a causal discovery algorithm; calculating causal feature tendency scores, and performing matching; carrying out anti-fact generation, forming a synthesized minority class set, and integrating the synthesized minority class set with original data to obtain an enhanced data set; and performing data cleaning on the enhanced data set to obtain a balanced data set. According to the method, a data set is effectively balanced by generating a high-quality and close-to-reality anti-fact sample, so that the performance of a downstream classifier on key indexes is remarkably improved; the feature values of the real instances are combined to ensure that the generated samples are located in a reasonable area of data distribution, so that the credibility and availability of the enhanced data are improved; the generated anti-fact sample is located in a boundary region between the majority class and the minority class, the decision region of the minority class is effectively expanded, and the unique post-cleaning avoids the influence of noise accumulation on model training.

The invention belongs to the technical field of multi-quad-rotor collaborative hunting, and discloses a multi-unmanned aerial vehicle collaborative hunting method in an obstacle environment, which comprises the following steps: step 1, three-dimensional decision space representation based on a Voronoi diagram: simplified representation of a three-dimensional obstacle decision space is realized by constructing a safe priority reachable area; and step 2, constructing a distributed hunting decision based on area minimization: on the basis of a security priority reachable region constructed by a three-dimensional decision space representation module based on a Voronoi diagram, compressing a motion region of the escape unmanned aerial vehicle by adopting a collaborative hunting strategy based on area minimization, and finally realizing effective capture. According to the three-dimensional decision space characterization method based on the Voronoi diagram, the convex bounded decision area can be extracted from the three-dimensional obstacle space which is a typical non-convex area, effective characterization of the decision space is achieved, and a foundation is laid for generation of a subsequent encircle strategy.

In the field of image recognition, small sample class incremental learning of deep learning model is a challenging technical problem, which iteratively updates parameters based on pre-trained model, so that the model can continuously learn the knowledge of new classes only with a limited number of labeled samples, while retaining the knowledge of original classes. Parameter updating under extremely limited samples will cause the model to deviate, and a large number of parameter updates will cause the model to forget existing knowledge. In order to solve the above problems, on the basis of nearest neighbor class mean classification method, the application proposes a small sample class incremental learning method based on robustness and few parameters fine-tuning. The method can realize efficient fine-tuning of key parameters, and produce flexible decision region by controlling the number and amplitude of parameter adjustment, thereby avoiding the above problems.

The invention relates to the technical field of building information management, in particular to a data-driven artificial intelligence aid decision-making method and system, and the method comprises the steps: recording project data with conflicts based on a division mode of a building life cycle; based on the date attribute, the distribution attribute and the site attribute of each building component in the project data in each stage, defining a conflict feature matrix of each building component according to the conflict type of each building component; taking the conflict feature matrix of each building component as a node, performing spatial clustering on each node, and constructing a decision work flow corresponding to each building component; performing time sequence association on the decision work flow, deducing decision intervals, and constructing a multi-level decision tree by project data in each decision interval; performing collaborative analysis by using a multi-level decision tree, determining a decision path output by the decision tree, and obtaining a currently executed construction processing strategy; the efficiency and accuracy of conflict data processing are realized.

The present application relates to the technical field of building information management, in particular to a data-driven artificial intelligence assisted decision-making method and system, comprising: recording project data with conflicts based on the division mode of building life cycle; defining the conflict feature matrix of each building component based on the date attribute, allocation attribute and site attribute of each building component in each stage in the project data, and the conflict type of each building component; taking the conflict feature matrix of each building component as a node, spatially clustering each node, and constructing the corresponding decision-making workflow of each building component; time-sequentially associating the decision-making workflow, deducing the decision-making interval, and constructing a multi-level decision tree with the project data in each decision-making interval; using the multi-level decision tree for collaborative analysis, determining the decision path output by the decision tree, and obtaining the current construction processing strategy; the efficiency and accuracy of conflict data processing are realized.

The present disclosure provides a low-orbit satellite communication resource allocation method, device, equipment and medium, the method comprising: receiving service request signaling sent by a ground user; obtaining satellite state information at the current time according to the service request signaling; inputting the satellite state information into a pre-deployed multi-agent deep reinforcement learning model to generate a resource allocation scheme for the next time; wherein the multi-agent deep reinforcement learning model comprises a first decision area and a second decision area, the first decision area is used to determine a beam scheduling scheme for the next time based on the satellite state information, and the second decision area is used to determine a frequency band and power allocation scheme for the next time based on the beam scheduling scheme for the next time. According to the embodiments of the present disclosure, real-time perception and accurate response to uneven user distribution and time-varying traffic characteristics are realized, and the dynamic adaptation capability and overall operation efficiency of the low-orbit satellite communication system resource allocation are effectively improved.

The invention discloses a table data multi-model multiplexing method based on dynamic path selection, which comprises the following steps of: collecting a table data set in a corresponding field on a website or a platform, preprocessing the table data set, and dividing the preprocessed table data set into a training set and a test set according to a proportion; a deep neural network model is constructed, initialized and trained, the deep neural network model comprises an input module and a plurality of mixed expert layers, a total loss function is defined for iterative training until the deep neural network model converges to obtain a multi-path model for storage, and a part of the multi-path model is stored to represent a training sample and a reasoning path library. And loading a multi-path model to process the to-be-processed table data sample, dynamically generating a reasoning path, outputting a prediction result, and establishing a multiplexing library based on a historical path. According to the method, the advantages of multiple models can be integrated without extra training cost, the calculation cost is reduced and the table data processing performance is improved while the decentralized decision-making area of the tree base model is duplicated, and the resource consumption is saved while the performance is improved.

This invention discloses a system and method for optimizing unstructured data join queries based on Siamese networks, belonging to the field of unstructured data processing. It includes: a feature extraction module, which uses a deep learning model to extract feature vectors from different types of unstructured data; a Siamese network module, which converts the feature vectors into a low-dimensional embedding space; a judgment layer module, which concatenates the feature differences and the original features, and outputs the join matching probability through a fully connected network; a threshold decision module, which determines the optimal upper and lower threshold pairs through adaptive mapping and grid search, and divides the join matching probability into an automatic decision area and a verification area, with the verification area further judging the join relationship through a dedicated inference model; and a query optimization module, which optimizes the query execution plan for unstructured data based on the join matching results. This invention enables efficient unstructured database join operations, supports automatic decision-making and precise verification mechanisms, and significantly improves the accuracy and efficiency of database query join operations.

The invention relates to the technical field of energy storage coordination, and discloses an energy storage system scheduling method and device, computer equipment and a storage medium. The method comprises the following steps: firstly, discretizing reference time to obtain a plurality of decision intervals; and then, carrying out linear programming based on the resource parameter, the constraint function, the decision variable and the energy value data corresponding to each decision interval of the energy storage system, thereby simplifying the complexity and improving the solving efficiency so as to determine the numerical value of the decision variable corresponding to each decision interval. And finally, on the basis of the numerical value of the decision variable corresponding to each decision interval, scheduling of energy storage and energy release is performed on the energy storage system, scheduling strategy optimization is realized, energy storage resources can be utilized more comprehensively, and the resource utilization rate and the economic benefit of the energy storage system are improved. Wherein the linear programming mode has good expansibility, and parameters and constraint functions can be flexibly adjusted according to actual requirements so as to be suitable for different types of energy storage systems and application scenes.

The invention discloses a three-dimensional object local confrontation attack method based on a maximum aggregation area sparse strategy, and in a physical 3D environment, the existing target detection-oriented confrontation attack (3D-AE) mainly faces the following challenges: in order to achieve a maximum attack effect, relatively large and dispersed confrontation patches are often selected, and the confrontation effect is poor. Therefore, the confrontation patch is too obvious, and the visual concealment is reduced. A better strategy is how to utilize as small as possible, and the attack effect of the aggregated adversarial patches is maximized. In order to maximize aggregation of adversarial patches, an aggregation degree regular term is designed to constrain a shielding aggregation matrix obtained based on a surface element adjacency relation; in order to ensure minimization of an adversarial patch area, sparse regularization is designed, so that shielding weights tend to be distributed in a U shape, and extreme values are limited. Expanded data of visual angle changes are obtained through neural rendering, and a universal important decision-making area under the target multi-angle condition is positioned through suppression model detection.

The invention discloses an equivalent verification method and device of a quantum circuit, equipment and a storage medium. Comprising the steps of obtaining an initial decision diagram through a first quantum circuit and a second quantum circuit; carrying out one-way endpoint movement based on a decision starting interval in the initial decision diagram, and storing the obtained left decision diagram and right decision diagram into an unselected decision diagram set; selecting a decision diagram with a larger decision estimation value from the unselected decision diagram set according to a target estimation function as a target decision diagram; when the target decision interval of the target decision diagram does not cover the initial decision interval of the initial decision diagram, updating the target decision interval as a start decision interval to obtain the left decision diagram and the right decision diagram again to determine a new target decision diagram until the target decision interval covers the initial decision interval, and obtaining a minimum decision diagram; and when the node merging result of the minimum decision diagram is a unit matrix, outputting circuit equivalent information. According to the method, the calculation complexity of quantum circuit equivalent verification can be reduced, and the calculation efficiency is improved.

This invention relates to the field of information system detection technology, specifically disclosing an artificial intelligence-based method for detecting anomalies in information system operation. The method includes: dividing the historical intelligent decision-making chain of an auctioned property into decision interval categories, statistically analyzing the proportion of anomalous decision points in each interval, calculating the confidence level anomaly probability, setting different detection window lengths, and then using the detection window length as a fixed unit to slide and extract the decision chain to be judged to generate a window to be judged. Continuous anomalies within the judgment window are identified, and new anomalous decision points are marked. The window to be reviewed is identified by combining the interval confidence level anomaly probability. Subsequently, the offline due diligence results chain is linked to compare and verify the validity of the anomalies, distinguishing between misjudgments and genuine anomalies. Finally, the initial detection window length is dynamically adjusted based on the time characteristics of the anomalous decision points to obtain the final detection window length, thus achieving dynamic adjustment of the detection window length and improving the accuracy of anomaly detection and risk control capabilities.

The present invention discloses a shaping method based on the probability of sub-constellation overlap and geometric union, which lays out the constellation points in the form of regular triangle unit cells, so that the distance between each constellation point is equal, thereby making the constellation point decision area denser; the probability shaping is performed by overlapping sub-constellations, the information is divided into constellation partitions, the priority is set according to the distance from the origin, and the number of sub-constellation overlaps is determined according to the priority and the modulation order, and the information is transmitted after adding an identification mark. The distance between each symbol point in the constellation diagram designed by the present invention is equal, with a denser decision area, and the overall average power is reduced compared with the traditional square constellation, and the high-order modulation constellation can be compressed. Compared with the traditional probability shaping scheme based on Maxwell Boltzmann distribution, the probability of any symbol point in the constellation diagram can be flexibly changed, and the constellation compression from high-order modulation format to low-order modulation format can be realized.

The invention relates to the technical field of energy storage configuration, and discloses an energy storage system capacity configuration method and device, computer equipment and a storage medium. Firstly, capacity division is carried out based on the capacity of the energy storage system to obtain a plurality of candidate capacity values; and then, linear programming is performed based on resource parameters, constraint functions, decision variables and energy value data corresponding to each decision interval of the energy storage system, so that the complexity is simplified, the solving efficiency is improved, expected income data corresponding to each candidate capacity value is determined, and the decision intervals are obtained by performing discretization processing on reference time. And finally, under the condition that the change condition of the expected revenue data meets a preset condition, configuring the candidate capacity value corresponding to the expected revenue data as the target capacity of the energy storage system, thereby effectively assisting the user to perform optimal configuration calculation of the energy storage side capacity, providing support for the user to make a decision, and improving the user experience. Therefore, the optimal utilization of energy storage resources and the maximization of economic benefits are realized.

The invention discloses a model fingerprint method based on identification decision area differences, and belongs to the technical field of multimedia information security. According to the technical scheme, a positive model pool is constructed for a source model by using a plurality of model stealing methods, and a negative model pool is constructed for another independent training model by using a model extraction model stealing method and a re-independent training method; performing iterative optimization on the clean image to generate a model fingerprint query set of the source model; evaluating a fingerprint sample by using the calculated transferability score, and performing model fingerprint screening on the query set; and querying the suspicious model by using the model fingerprint query set, analyzing a result, and performing model fingerprint verification. The method has the beneficial effects that the method is excellent in the aspects of improving the model fingerprint construction speed, enhancing the copyright verification capability, protecting the model intellectual property, improving the model security, reducing the copyright verification cost and the like; the method not only provides powerful technical support for intellectual property protection of the deep learning model, but also promotes healthy development of the artificial intelligence technology.

The present invention relates to a multi-energy complementary hybrid energy storage scheduling system and a scheduling method thereof. The system uses a deep neural network model based on artificial intelligence technology to extract high-dimensional implicit features in the time dimension from the power consumption value of each electricity user, the valley power density of the power grid, and the energy storage capacity of the energy storage system, and uses this to perform feature fusion to judge the rationality of the proportion of electricity allocated to electricity users and the energy storage system. In addition, when fusion is performed, a class condition boundary constraint synthesis method of eigenvalues ​​is adopted to avoid excessive fragmentation of the decision area within the classification target domain due to out-of-distribution samples of the eigenvalue set, thereby improving the classification performance of the classification feature matrix and further improving the accuracy of rationality judgment.

1. The name of the design product: auxiliary diagnosis decision area of large model auxiliary diagnosis decision graph user interface for electronic equipment. 2. The use of the design product: for an electronic device. 3. The design points of the design product: in the graphical user interface. 4. The picture or photo that best indicates the design points: front view. 5. The use of the graphical user interface: overall use: interface for large model lesion staging intelligent auxiliary diagnosis; local use: auxiliary diagnosis decision area for large model. 6. The human-computer interaction mode of the graphical user interface: the front view is a lesion staging auxiliary judgment function interface, clicking the "start generation" button on the right side of the interface enters the lesion staging version generation interface, as shown in the change state diagram, the staging version can be switched to regenerate in the interface. 7. Other circumstances that need to be explained: omit other views. The partially protected design is not requested to be protected in the semi-transparent yellow covered part in the view. The "X" in each view represents replaceable or changeable text, numbers or punctuation marks; the part covered by the opaque gray block in each view belongs to the replaceable or changeable file content, which does not belong to the content of the design itself which is requested to be protected.

The invention provides a mixed depth large model reasoning acceleration method and device based on a unified decision. The method comprises the steps of processing a preset data set sample, calculating the similarity between input and output of a target model hidden layer, slidably selecting a continuous hidden layer based on a preset layer number, calculating the average similarity in an interval, and selecting the highest hidden layer as a decision interval. A unified dynamic decision-making module is deployed in front of a decision-making interval, and the module selects a target hidden layer in the interval to participate in operation based on the output of a previous hidden layer, skips other layers, and realizes reasoning acceleration. And the unified dynamic decision-making module synchronizes parameter updating with the target model in training. According to the method, decision interval selection and module deployment and training are optimized, the model reasoning calculation amount is reduced, the reasoning speed is increased, the downstream task performance is kept, the problem that the reasoning speed is reduced due to extra overhead in the prior art is solved, and the efficiency and performance of the model in practical application are improved.