70 results about "Meta-optimization" patented technology

The invention belongs to the technical field of spatial optimization of geographic information science, and particularly discloses a geographic site selection method based on quantum computing, which comprises the following steps: defining a geographic maximum coverage site selection model, and determining an optimization objective function, constraint conditions and a decision method; converting the geographic maximum coverage location model into a quantum annealing calculation solvable secondary unconstrained binary optimization model; a slack variable is introduced, the high-dimensional inequality constraint is converted into a secondary penalty term constraint, and the slack variable is added into a secondary unconstrained binary optimization model; based on quantum annealing calculation, setting coherence time and sampling frequency, and solving the secondary unconstrained binary optimization model to obtain an optimal solution of service facility configuration; and carrying out space coverage consistency check between the service facility and the demand point on the optimal solution configured by the service facility, and verifying the solving precision and effectiveness of the quantum calculation method. According to the method, the problem of space optimization in geographic information science is solved.

The invention discloses a matrix calculation unit optimization-based sparse LU decomposition solving method and a related device, and relates to the technical field of numerical algorithm optimization, and the method comprises the steps: calling an improved PRCM algorithm to carry out the reordering of a sparse matrix, improving the local density of data, obtaining a sparse matrix after bandwidth compression, and obtaining a sparse matrix; then converting the sparse matrix into a regular and dense Tile block structure, storing the Tile block structure in a pre-allocated continuous memory space, performing numerical decomposition on the sparse matrix after bandwidth compression by adopting a right view-based LU decomposition algorithm, and accelerating by using a matrix calculation unit in the numerical decomposition process to obtain an LU matrix; therefore, efficient solving of the to-be-solved linear equation can be completed. According to the scheme, the spatial locality of the sparse matrix is fully utilized, the strong computing power of the matrix computing unit is exerted, the computing efficiency is improved while the computing amount is reduced, and the time of sparse LU decomposition in symbol decomposition and numerical decomposition stages is remarkably shortened.

The invention provides a meta-learning-based large language model training method and device and a data classification method and device, and relates to the technical field of data processing. The meta-learning-based large language model training method comprises the steps of obtaining training data, and constructing a task pool containing a plurality of natural language processing tasks based on the training data; dynamically adjusting task weights of the natural language processing tasks based on an adaptive mechanism, and determining a selection probability corresponding to each natural language processing task based on the task weights; sampling the task pool based on the selection probability to obtain a meta-task set; and based on a mixed parameter freezing strategy, performing global meta-optimization, local meta-optimization and attention mechanism optimization on the pre-training model by utilizing the meta-task set to obtain a meta-learning-based large language model. According to the technical scheme, the adaptability and training efficiency of the large language model under multiple tasks and different data distributions can be improved.

The invention discloses a tunnel pipe roofing construction dynamic deviation correction method based on laser ranging and optical fiber sensing, and relates to the technical field of measurement and sensing, and the method comprises the steps: S1, collecting laser and optical fiber data, obtaining real strain through temperature coupling compensation, and calculating global pose parameters; s2, generating an eigenmode component based on the real strain and the global pose parameter, and screening and constructing an intelligent feature vector according to the contribution degree of the physical coupling feature; s3, constructing a physical information driving state estimation model in combination with pipe joint mechanical constraints, and outputting a pipe joint state estimation value in real time; s4, performing construction deviation prediction by using the pipe joint state estimation value, and generating an optimal control sequence and instruction; and S5, applying the control instruction to the construction device, and adaptively updating the key parameters of the system through a meta-optimization algorithm based on feedback. And through multi-modal data fusion and physical constraint intelligent prediction, high-precision deviation correction, closed-loop robust control and system self-adaptive optimization of the pipe joints are realized, and the intelligence and long-term precision of tunnel pipe curtain construction are improved.

The invention relates to a power distribution network frame optimization method based on a quantum annealing algorithm. The method comprises the following steps: constructing an objective function of power distribution network frame optimization based on actual power distribution network data and setting constraints; obtaining a quantum Hamiltonian function based on the objective function and the constraint; based on the quantum Hamiltonian function, obtaining a secondary unconstrained binary optimization objective function of power distribution network frame optimization; based on the quantum Hamiltonian function or the quadratic unconstrained binary optimization objective function, solving a power distribution network frame optimization optimal solution; and generating a power distribution network frame optimization strategy based on the optimal solution. Compared with the prior art, the method has the advantages of higher calculation efficiency and precision and the like.

The invention provides a query optimization method, electronic equipment, a storage medium and a program product. The method includes. Based on the left deep tree, the optimization problem of the table connection sequence in the database is converted into a secondary unconstrained binary optimization QUBO problem; the relationship in the left deep tree is a table in the database; solving a QUBO problem through a quantum solving method to obtain a connection relation of the tables in the left deep tree; and determining an optimized table connection sequence based on the connection relationship of the tables in the left deep tree.

Methods, systems, and apparatus for solving cost optimization problems. In one aspect, a method includes receiving data representing a cost optimization problem in a network, wherein i) the network is represented as a graph of nodes and edges, and ii) each edge comprises an associated cost; mapping the data representing the cost optimization problem in a network to a quadratic unconstrained binary optimization (QUBO) formulation of the cost optimization problem, the QUBO formulation comprising multiple variables with values determined by states of respective qubits, wherein each qubit corresponds to a respective edge of the graph of nodes and edges; obtaining data representing a solution to the cost optimization problem from a quantum computing resource; and initiating an action based on the obtained data representing a solution to the cost optimization problem.

The invention provides a target detection model robustness test method based on cross-task learning, features extracted by a semantic segmentation model are teacher features, features extracted by a target detection model are student features, the similarity between the teacher features and the student features is evaluated by using a query-key attention mechanism, and the robustness of the target detection model is tested. The optimization process of stabilizing the gradient by adopting a historical gradient stabilization strategy is adopted, the conflict gradient direction is adjusted through conflict gradient projection, better gradient compatibility is obtained in combination with alternate joint task element optimization, and the test performance is improved. According to the method, diversified gradient information with strong correlation required by the target detection task is distilled from the semantic segmentation task, so that the target detection model can learn more universal and robust characteristics, the gradient direction is further stabilized and optimized, and the performance of the model robustness test is improved.

The invention provides a construction method of a network intrusion model based on meta learning, comprising the following steps: a global server constructs a plurality of client nodes based on a federated learning framework, and constructs a node detection model in each client node; a client node collects local multivariate heterogeneous data at a fixed time interval, generates a simulated attack sample by using a GAN network, and then forms training data by using the simulated attack sample and a normal sample; performing iterative training on the node detection model by using the training data; performing meta-optimization on the trained node detection model by using an irrelevant meta-algorithm; and the global server dynamically allocates aggregation weights for each client node, and performs model aggregation in a weighted average manner to generate a global detection model. According to the method, the problems that the detection capability of a detection model on minority class attacks is insufficient and the detection model is difficult to quickly adapt to newly occurring novel network attacks due to the fact that the classes of model training samples are unbalanced in the prior art are solved.

The invention belongs to the related technical field of dual-scale structure design, and discloses a dual-scale structure optimization design method and system based on substructure data driving, and the method comprises the steps: an offline stage: for cell elements of a dual-scale structure, carrying out the cutting operation of a level set function, obtaining a virtual microstructure, and carrying out the cutting operation of the level set function; determining the density of the actual microstructure according to the density of the virtual microstructure; performing static condensation through a substructure method to obtain a stiffness matrix of the super-unit, and constructing a mapping model of the cutting height parameter, the stiffness matrix of the super-unit and the density of the actual microstructure; in the online stage, a data driving model with the cutting height parameter as the input quantity and the stiffness matrix of the super-unit and the density of the actual microstructure as the output quantity is obtained; and applying the data driving model to a finite element optimization process of the dual-scale structure, and finally obtaining an optimized dual-scale structure. According to the invention, an off-line data driving method is adopted to carry out dual-scale structure optimization, and the efficiency can be greatly improved under the condition that the precision is ensured.

A method and system for building and implementing a meta-machine learning (meta-ML) optimization engine for a neural network (NN) or a machine learning (ML) connective model. A computer processor may iteratively simulate a backpropagation algorithm by executing a sequence of optimization steps. At each optimization step a position of a loss function may be determined that may be closer than a previously determined position of the loss function to a local minimum. A computer processor may compute and store after each iteration a detachment of the loss function, learning rate, and optimal learning rate. A computer processor may train a machine learning connective model to model the optimal learning rates of the simulated backpropagation algorithm. The meta-ML optimization engine may be implemented for a NN or ML connective model by generating a modified backpropagation algorithm in which algorithmic features of gradient descent may be replaced by the meta-ML optimization engine.

The embodiment of the invention relates to a non-expert data assisted meta reinforcement learning frequency hopping interference resource allocation method. The method comprises the following steps: constructing a frequency hopping interference resource allocation problem according to a communication confrontation scene; modeling a frequency hopping interference resource allocation problem as a Markov decision process; randomly initializing the element strategy to obtain an initial strategy network; meta-training is carried out on the initial strategy network based on an MAML algorithm, and the meta-training comprises the steps that in the specific task adaptation stage, based on non-expert data, optimization training is carried out on the initial strategy network through a reinforcement learning loss function and a behavior cloning loss function, a specific task strategy is obtained, and reinforcement learning loss of all interference tasks is calculated; in the meta-optimization stage, the reinforcement learning loss of all interference tasks is minimized, and an optimized distribution strategy is obtained; and carrying out fine adjustment on the optimized distribution strategy according to the new interference task. According to the method, useful information is extracted from non-expert data in combination with reinforcement learning and behavior cloning, and efficient strategy exploration is realized.

The invention discloses a renewable energy source data clustering method and system based on an optical quantum computer, and the method comprises the steps: building a renewable energy source data clustering optimization problem based on a quadratic unconstrained binary optimization (QUBO) model based on a discrete optimization objective function, carrying out the matrix of the objective function, carrying out the clustering optimization of the objective function through a kernel method, and carrying out the clustering optimization of the objective function. The method comprises the following steps of: establishing a target function item, converting the target function item into a matrix form, constructing a matrix of a new clustering model, finally solving a new QUBO optimization problem by using CIM, and converting the new QUBO optimization problem into an optimal clustering result. According to the method, data similarity measurement is improved by constructing the QUBO model and combining a kernel method, and a renewable energy source data clustering problem is converted into a form suitable for CIM solution, so that the calculation bottleneck of a classical calculation method in high-dimensional data processing is avoided. According to the method, the nonlinear characteristic of the data can be kept, efficient clustering optimization calculation can be realized, and the calculation efficiency and the clustering precision are remarkably improved.

The invention relates to the technical field of finite element optimization, in particular to a performance optimization method and system based on finite elements, and the method comprises the steps: obtaining a finite element model of a gear in a transplanting mechanism of a rice transplanter, and dividing the finite element model into a plurality of grids; corresponding loads are applied to the finite element model according to simulation data of gears when the rice transplanter works, simulation indexes of all the grids are obtained, the grid sizes of all the grids are optimized based on the simulation indexes, and optimized target grids are output; determining a performance index of each target grid, optimizing design parameters of the target grids according to the performance indexes, and updating the finite element model based on the optimized design parameters to serve as a final finite element model of the gear; according to the method, through deep coupling of grid optimization and parameter optimization, the analysis precision and optimization efficiency of the gear can be improved.

The invention relates to a multi-machine multi-task configuration method and device, a medium and equipment, belongs to the technical field of task management optimization, and solves the problem that a large-scale multi-machine task configuration problem cannot be effectively solved in the prior art. According to the technical scheme, the method mainly comprises the steps that a task set and a machine set are obtained, and needed processing time corresponding to all tasks in the task set is obtained; constructing a task configuration binary variable to represent whether any task is configured to any machine; an optimization model is constructed according to a preset configuration target, the optimization model comprises a target function and constraint conditions, and the constraint conditions are expressed as constraint expressions through required processing time and task configuration binary variables; forming a penalty term according to the constraint formula, and adding the penalty term into the target function to form a quadratic unconstrained binary optimization model; and solving the quadratic unconstrained binary optimization model to determine the value of each task configuration binary variable, and determining a task configuration scheme according to the values of the task configuration binary variables.

Example systems perform complex optimization tasks with improved efficiency via neural meta-optimization of experts. In particular, provided is a machine learning framework in which a meta-optimization neural network can learn to fuse a collection of experts to provide a predicted solution. Specifically, the meta-optimization neural network can learn to predict the output of a complex optimization process which optimizes over outputs from the collection of experts to produce an optimized output. In such fashion, the meta-optimization neural network can, after training, be used in place of the complex optimization process to produce a synthesized solution from the experts, leading to orders of magnitude faster and computationally more efficient prediction or problem solution.

The invention relates to the technical field of data processing, in particular to a metal strip mine production scheduling optimization system based on multi-source data fusion, and the system comprises a data processing module which comprises a data collection unit for collecting metal strip mine data of a plurality of data sources; the model training module comprises a model generation unit which is used for carrying out model training according to the feature extraction unit so as to generate a genetic algorithm model; the production scheduling module is connected with the model training module and comprises an analysis unit which is connected with the model generation unit and is used for analyzing the metal strip mine data through a genetic algorithm model so as to output an analysis result; and the optimization module is respectively connected with the data processing module, the model training module and the production scheduling module, and is used for determining time alignment tolerances of different data sources according to the loss rate of the metal strip mine data. The optimization stability of metal strip mine production scheduling is improved.

An information processing device includes an optimization unit, a control unit, and a search unit. The optimization unit executes optimization calculation processing of obtaining an optimal solution by repeatedly executing solving processing of obtaining a solution to an optimization problem by using one or more set parameters. The control unit sets a calculation time of the optimization calculation processing for each of a plurality of search times obtained by dividing a specified time that is specified. The search unit repeatedly executes search processing of setting the parameter different from parameters of other search times for each of the plurality of search times, causing the optimization unit to execute the optimization calculation processing within the set calculation time by using the set parameter, and searching for an optimal value of the parameter by Bayesian optimization by using the optimal solution obtained by the optimization calculation processing.

One example method includes receiving a plurality of Quadratic Unconstrained Binary Optimization (QUBO) instances, the plurality of QUBO instances including at least first controllable features and their corresponding feature values and a first uncontrollable structure feature that defines a problem type of each of the plurality of QUBO instances. The first controllable features and the first uncontrollable structure feature of the received plurality of QUBO instances are used to train a QUBO generation machine-learning (ML) model to generate a QUBO instance. A QUBO feature set that includes second controllable features and their corresponding feature values and a second uncontrollable structure feature that defines a problem type of a new QUBO instance is received. The trained QUBO generation ML model generates the new QUBO instance that includes the second one or more controllable features and their corresponding feature values and the second uncontrollable structure feature.

The invention discloses a neural network reliability evaluation method based on self-attention and meta-learning. The method comprises the following steps: in a task specificity adaptation process, a self-attention mechanism is utilized to model association between a bit flipping fault position and a neural network reliability influence, and technologies such as sparse coding, label distribution smoothing and a sparse perception engine are combined to construct a neural network reliability evaluation model; in the meta-optimization process, initialization parameters are trained in task distribution formed by different neural network architectures, and the initialization parameters are applied to the evaluation model, so that the universality of the evaluation model among different tasks is improved; the method has high efficiency, accuracy and universality, and is suitable for various fault scenes and neural network architectures.

The invention relates to a sparse principal component analysis method, a quantum computing device and a medium, belongs to the technical field of data dimension reduction, and solves the problem that the sparse principal component analysis method is low in efficiency. According to the technical scheme, the method mainly comprises the steps that a target data matrix is obtained, and a covariance matrix is determined based on the target data matrix; setting a to-be-solved principal component vector, and determining an original objective function and a vector constraint condition; expressing each component of the to-be-solved principal component vector through a binary system, and defining a plurality of second binary variables to respectively express whether each component of the to-be-solved principal component vector is zero or not; determining constraint conditions of the first binary variable and the second binary variable according to the vector constraint conditions and the definition of the second binary variable; forming a secondary unconstrained binary optimization model; and solving the quadratic unconstrained binary optimization model to determine the value of each first binary variable.

The invention discloses a novel finite element optimization solution algorithm, which comprises the following steps of: preprocessing a real image scanned by CT (Computed Tomography) to obtain an image gray value; reading an image gray value to obtain an image finite element model; establishing a mapping relationship between the unit node number and the overall degree-of-freedom number; constructing a Jacobian matrix and a geometric matrix according to the unit node numbers; calculating an element stiffness matrix and a stress function matrix corresponding to each material attribute; calculating an equivalent node force according to the element stiffness matrix, and assembling a node force vector; solving node displacement under a given boundary condition; according to the displacement field, stress components corresponding to all units under the given boundary condition are calculated and averaged, and finally the homogenized constitutive tensor is obtained. The invention provides a novel finite element optimization solution algorithm, a traditional finite element preprocessing process is simplified, the calculation efficiency is remarkably improved, and calculation resources are greatly saved.

The present disclosure discloses a louver fin micro-channel heat exchanger design method based on heat transfer unit optimization, wherein a heat transfer unit is selected according to the louver fin micro-channel heat exchanger, the fins and the micro-channel of the heat transfer unit are modeled, the structural parameters of the heat transfer unit are extracted to establish sample space points, a parameter subset with the largest prognosis coefficient is selected, and a response surface model is established for each response variable; based on the response surface model, a genetic algorithm is applied to optimization within the limit interval and the constraint condition, the heat exchanger volume is reduced, the heat exchanger heat transfer capacity is increased, and the corresponding fan power is reduced; the output parameters of the response surface are used as a subset of the objective function of the genetic algorithm optimization, so as to reduce the fan power of the louver fin micro-channel heat exchanger and improve the heat transfer capacity under a certain capacity, and the corresponding Pareto frontier is obtained, and the optimal solution in the geometric parameter dimension region is given.

A method is disclosed for optimizing Quadratic Unconstrained Binary Optimization (QUBO) instances for efficient execution on a quantum computer. Initially, a first machine learning (ML) model receives a QUBO instance along with scaling and scheduling parameters. The first ML model transforms the QUBO instance into a scaled version and adjusts the scheduling parameters accordingly. Subsequently, a second ML model compares the scaled QUBO instance and parameters with those of a known efficiently executable QUBO instance on a quantum computer. Based on this comparison, the second ML model assigns a score to the scaled QUBO instance, indicating its efficiency for execution on the quantum computer. This method enables the optimization of QUBO instances for enhanced quantum computing performance.

The invention relates to the technical field of energy storage power station emergency disposal, in particular to an energy storage power station intelligent emergency disposal process automatic generation system and method.The system adopts a scheme of multi-modal fault understanding, large language model reasoning and multi-target optimization scheduling, receives multi-modal input such as texts, alarm codes, temperature / voltage / alarm marks and the like, and sends the multi-modal input to a server; the method comprises the following steps: firstly, taking GPT-4o as a scene vector, carrying out double-tower Transform coding on the GPT-4o as a scene vector, generating a candidate plan by combining the GPT-4o subjected to LoRA fine tuning with a knowledge graph, screening an optimal scheme through a safety, timeliness and cost ternary optimization function psi and a particle swarm optimization algorithm, and carrying out digital twinborn simulation verification and weekly-level closed-loop learning iteration; according to the method, a standardized plan is generated in 1.7 seconds, the compiling efficiency is improved by 103 times, the execution consistency is improved by 23%, the annual operation and maintenance accident loss is reduced by 18%, the monthly newly-added scene coverage rate is 92%, the deployment operation and maintenance confirmation rate in a 100 MWh energy storage station is 96%, and the energy storage safety operation and maintenance intelligent level is remarkably improved.

One example method includes receiving a Quadratic Unconstrained Binary Optimization (QUBO) problem that comprises a matrix that includes various cells having data. It is then determined that one or more of cells is missing data or has corrupted data. A machine learning (ML) model performs a denoising process that removes random noise from the one or more cells having the missing data or corrupted data. This results in data being imputed to the one or more cells having the missing data or the corrupted data. The imputed data approximates the missing data or approximates an expected value of the corrupted data before the corrupted data was corrupted.

The invention provides a meta-learning-based domain generalization blood pressure estimation training method, which is characterized in that a learning process is divided into a meta-training stage and a meta-testing stage through a two-stage meta-learning strategy, the meta-training stage is further divided into a training sub-stage and an optimization sub-stage, and in the learning process, the meta-testing stage is further divided into a training sub-stage and an optimization sub-stage; source domains are selected in a transformation mode to form a meta-optimization domain, the other source domains serve as meta-training domains, in each round of meta-training stage, the model is trained on the multiple source domains in the meta-training domains to simulate data distribution of different individuals, and the generalization performance of the model is improved by learning data of the different individuals; in the meta-test stage, the model is finely adjusted by using source domain data so as to improve the prediction precision on a real target domain.

This invention discloses a routing method and electronic device for heterogeneous low-Earth orbit satellite networks. The method includes: constructing a domain-specific network architecture and establishing a hierarchical architecture comprising a data plane, a control plane, and a meta-learning layer; establishing an end-to-end latency model and an energy consumption model; defining a multi-objective optimization function based on KL divergence; and executing meta-learning routing decisions, which includes a meta-optimization stage and an agent optimization stage. In the meta-optimization stage, an adaptive learning objective is generated through the meta-network; in the agent optimization stage, the routing method is updated according to the learning objective, and routing actions are executed. This invention solves the problems of poor adaptability of routing methods and difficulties in multi-objective trade-offs in heterogeneous networks, achieving adaptive generation and efficient collaboration of routing methods, and improving the overall performance of the network.

The invention discloses a meta-optimization adversarial sample generation algorithm combined with frequency domain feature fusion. The meta-optimization adversarial sample generation algorithm comprises the following steps: S1, frequency domain feature decomposition; s2, low-frequency reconstruction: reconstructing a low-frequency reconstructed image of the input image by adopting inverse discrete wavelet transform, and calculating a corresponding high-frequency image according to an original image and a low-frequency reconstructed difference value; s3, feature storage: respectively inputting the low-frequency image and the high-frequency image of the clean sample into a white box model, and extracting and storing features corresponding to each layer; s4, performing feature fusion; s5, inputting the fusion features obtained in the step S4 into a subsequent layer of the white box model to obtain model output; and S6, executing a meta-optimization strategy, calculating and combining internal and external gradient information to update the adversarial sample, and outputting a finally generated adversarial sample. According to the method, the image is decomposed and reconstructed in the frequency domain, and the multi-frequency features and the meta-optimization updating strategy are extracted and fused respectively, so that conflicts among different frequency domain features are effectively relieved, and the attack mobility of the generated confrontation sample to a conventional model and a defense model is improved.

A method for driving a quantum computational network for determining an extremal value of a cost function for solutions of a quadratic unconstrained binary optimization problem includes initializing qubits, sequentially applying layers of quantum gates to the qubits, determining an output state of the quantum computational network for obtaining a solution associated with the set of variational parameters {right arrow over (θ)}, and determining an output state for shifted variational parameters {right arrow over (θ)}* to evaluate a partial derivative with respect to the subset of the variational parameters {right arrow over (θ)} for determining a gradient of the cost function based on the output state for the shifted variational parameters {right arrow over (θ)}*, and by updating the variational parameters {right arrow over (θ)} based on an update function of a moving average over the gradient of the cost function and of a moving average over the squared gradient of the cost function.