41 results about "Iteration loop" patented technology

The invention discloses a large model and knowledge graph dual-drive-based guided reasoning system and method, belongs to the technical field of artificial intelligence reasoning, and solves the problem of how to improve the process reasoning ability of a large language model for engineering subject courses and the reliability of solving complex engineering problems. According to the method, metadata is extracted from teaching materials, structured problem representation is constructed, and a knowledge graph is constructed to form a complete knowledge system; the method comprises the following steps: decomposing a complex engineering problem into a structured solving plan with knowledge marks, and carrying out iterative loop based on a Monte Carlo tree search algorithm to generate a search tree comprising a plurality of high-quality candidate problem solving paths; then determining an optimal answer and a corresponding reasoning path in all simulated paths in a voting mode, performing confidence evaluation on each node of each path in the candidate path set, and further selecting a path of an optimal solution; and the process reasoning capability of the large language model on engineering subject courses and the reliability of solving complex engineering problems are effectively improved.

The invention provides an SQL statement generation method and system based on execution of feedback iterative correction, and relates to the technical field of large language models.The method comprises the steps that a query request of a user is received, and metadata information of a target database is obtained; generating a current SQL statement based on the query request and the metadata information; pre-executing the current SQL statement in a database sandbox environment; obtaining a pre-execution result, directly outputting the pre-execution result if the pre-execution result succeeds, and performing iterative loop if the pre-execution result fails; and analyzing error information in the pre-execution result, and determining an error type. According to the method, the illusion problem of a large language model is fundamentally solved, and the semantic correctness of SQL generation is ensured. And meanwhile, the iterative correction process replaces low-efficiency manual cue word adjustment, so that the use experience and the problem solving efficiency can be remarkably improved. In addition, the whole pre-execution process is completed in a sandbox environment, so that the influence of potential misoperation on production data is effectively isolated, and the safety of the system is enhanced.

A method of detecting superimposed SCMA signals from multiple user equipment in a receiver of an OTFS communication system includes initializing an executing an iteration loop, in which a mean of all a posteriori estimates of the transmitted signal determined so far is calculated, which a posteriori estimates are based on the OTFS-demodulated received signal, the corresponding channel matrix, and the mean vectors and variances determined for each UE. The mean of all a posteriori estimates determined so far is used for determining vectors and variances for each UE, further using the probabilities of the non-zero elements of the respective UEs codebooks. The iteration is repeated until a termination criterion is met.

The invention discloses a quick video auditing method and system based on time sequence prediction, and mainly relates to the technical field of computer vision. Comprising the following steps: constructing a time sequence state prediction model, performing global scanning on an initial frame of a video stream, and calling a target detection algorithm to identify and position a target; extracting state information of the target in the initial frame to form an initial state vector # imgabs0 #, and initializing a time sequence state prediction model; and performing iterative loop processing on each frame after the initial frame by using the initialized time sequence state prediction model to complete video auditing. The method has the advantages that computing resources are concentrated in a predicted region of interest (ROI) to be scanned and recognized by predicting the possible position and state of a target in a subsequent frame instead of global scanning, so that the computing resources are greatly saved, and the auditing efficiency is improved.

The invention discloses a dynamic convergence control method and device for an iterative task in intelligent application, and belongs to the technical field of computers. Task data are acquired and initialized to obtain a plurality of data units, and a corresponding high-dimensional state vector is constructed for each data unit; the high-dimensional state vector is used for representing the change of content embedding, and performing iterative loop processing on the task data; after execution of one iteration loop processing period is completed, the high-dimensional state vectors of the related data units are updated, and updated high-dimensional state vectors are obtained; and based on a preset convergence condition, judging the updated high-dimensional state vector by using a semantic convergence controller to obtain a judgment result. According to the method, the high-dimensional state vector is judged and processed, so that resources are saved, and a high-quality result is generated; by introducing the semantic convergence controller, whether semantic oscillation occurs or not is judged, meanwhile, invalid optimization can be recognized, and the iterative task processing efficiency is improved.

The invention provides a dynamic scheduling method and system for a redundant executor based on a foreground theory and a Q learning evolutionary game method. The method comprises the following steps: constructing a strategy space, a profit structure and strategy evolution dynamics of the executor and a reward and punishment mechanism of an enhanced analyzer-regulator; according to the dynamic scheduling method, an executor simulates interaction under a tripartite game framework, iterative loop is carried out, and a strategy space of a mimicry defense system is output as a redundant executor after the loop is ended, the simulation interaction process comprises the following steps: the enhanced analyzer-regulator monitors and obtains objective income of each execution body in real time, calculates subjective perception utility of each execution body according to an income structure of the execution body, and then adjusts the subjective perception utility of the mimicry defense system according to a reward and punishment mechanism of the enhanced analyzer-regulator. And each execution body dynamically updates the strategy space of each execution body according to the adjusted subjective perception utility and strategy evolution.

The invention relates to the technical field of generative artificial intelligence, and particularly discloses a diffusion model machine forgetting training method and system for preventing sensitive content generation, and the method comprises the steps: generating a pseudo sample set; calculating a significance gradient and a significance mask of generator parameters based on the pseudo sample set, and screening key parameters related to forgetting semantics; performing selective gradient updating on the pseudo-fractional network parameters based on the saliency mask; taking the updated parameters of the pseudo-fractional network as a supervision reference of a behavioral distillation stage, sampling forgetting class pseudo samples and reserved class pseudo samples, and guiding training of a generator network; executing gradient updating on the key parameters marked by the saliency mask for forgotten pseudo samples; executing all-parameter gradient updating for the reserved pseudo samples; and after iterative loop, obtaining a generator network of which the specific category is forgotten. According to the method, a selective updating mechanism of structure perception is introduced, so that the accuracy, stability and retention performance of forgetting sensitive contents are improved.

The invention provides a JAX-based high-performance discrete event simulation method and system, and the method comprises the steps: firstly completing environment configuration and initialization through an immutable data structure, and generating an EnvParas object with fixed static parameters; secondly, processing event logic by adopting a pure function steepv, and realizing side-effect-free functional event processing and state iteration by combining JAX condition control primitive and lax.scan iterative loop; then, a batch processing task is generated through job packaging, and job-level parallel acceleration is achieved on multiple GPU / TPU devices by means of jax. Jit compiling and jax. Vmap vectorization; and finally, packaging a standardized AI interactive interface, supporting external algorithm calling and finishing result post-processing. Functional programming and JAX hardware acceleration are combined, the simulation speed is increased, meanwhile, the event dynamic accuracy is guaranteed, reinforcement learning large-scale sampling can be seamlessly supported, and the method is suitable for AI application scenes in the field of operational research.

The invention belongs to the field of integrated circuit testing, and discloses a Shmoo testing method and device and a storage medium. The method comprises the steps of determining to-be-optimized Shmoo test parameters of an integrated circuit and defining a search range; an optimization objective function is established, an initial test parameter set is generated, and initial test or simulation is executed to obtain initial fitness data; executing an iterative loop of the grey wolf optimization algorithm, calculating a fitness value of a current test parameter set in each iteration, identifying an optimal test parameter, a suboptimal test parameter and a third optimal test parameter, and updating an algorithm control parameter; updating other test parameters according to the position information of the three test parameters, and carrying out boundary processing on the updated test parameters to keep the test parameters in a predefined search range; and when the termination condition is met, outputting the optimal test parameter as an optimal Shmoo test parameter combination, thereby determining the performance boundary of the integrated circuit or generating a Shmoo graph, and realizing more comprehensive and accurate Shmoo test.

This application discloses a data hashing method, apparatus, electronic device, and storage medium, relating to the field of data processing technology. The method includes: acquiring at least one set of messages corresponding to the data to be hashed; generating new message words by expanding a portion of the message words in each set of messages; when there are multiple sets of messages, in the first iteration loop, generating a corresponding intermediate hash value based on a preset number of message words included in the first set of messages corresponding to the first iteration loop and a plurality of pre-configured initial values; generating the initial value for the first iteration round in the next iteration loop based on the intermediate hash value corresponding to the first iteration loop and the plurality of pre-configured initial values, and entering the next iteration loop, until the current iteration loop is the last, and determining the final hash value corresponding to the current iteration loop as the target hash value. This application can solve the problem that data hashing requires a large amount of logical resources, resulting in a large algorithm core area.

A method of detecting superimposed SCMA signals from multiple user equipment in a receiver of an OTFS communication system includes initializing an executing an iteration loop, in which a mean of all a posteriori estimates of the transmitted signal determined so far is calculated, which a posteriori estimates are based on the OTFS-demodulated received signal, the corresponding channel matrix, and the mean vectors and variances determined for each UE. The mean of all a posteriori estimates determined so far is used for determining vectors and variances for each UE, further using the probabilities of the non-zero elements of the respective UEs codebooks. The iteration is repeated until a termination criterion is met.

A data processing service generates for iteratively applying a geospatial function to geospatial data. The generated code includes at least a first iterative loop and a second iterative loop. The data processing service compiles the generated code to generate compiled code that vectorized at least the second iterative loop. The data processing service receives a request from a client device to perform one or more data processing operations including applying the geospatial function to a data table of geospatial cell indices. The data processing service compiles the request into one or more tasks including at least a vectorized operation based on the compiled code and executes the one or more tasks by at least invoking the vectorized operation on the set of worker nodes.

The application discloses an edge computing task offloading method and system based on soft clustering and dynamic constraints, and relates to the technical field of edge computing. The method comprises the following steps: clustering robots, constructing a membership matrix, and mapping the cluster center back to the individual decision space of the robot; an evolutionary iteration cycle is performed, in which, during the cycle, a dynamic constraint relaxation factor is determined according to the real-time proportion of feasible solutions, the dynamic constraint relaxation factor is set to 0 for safety level constraints and deadline constraints, a constraint violation degree vector is calculated for the candidate solution, and the candidate solution is screened according to the feasibility boundary box; the candidate solution is Pareto sorted; a safety level index is calculated, and when a preset safety threshold is met, a Pareto optimal solution set is extracted; and a task offloading scheme represented by the Pareto optimal solution set is executed. The application solves the problem that the traditional method is completely ineffective when the feasible solution density is less than 10 ‑15 , ensures 100% safety compliance rate, improves search efficiency, and solves the dimension disaster problem.

The invention discloses a data hashing method and device, electronic equipment and a storage medium, and relates to the technical field of data processing, and the method comprises the steps: obtaining at least one group of messages corresponding to to-be-hashed data; generating new message words according to part of the message words in the plurality of message words included in each group of messages; when the messages comprise a plurality of groups, in the first iterative loop, generating corresponding intermediate hash values according to a preset number of message words and a plurality of pre-configured initial values included in a first group of messages corresponding to the first iterative loop; and according to the intermediate hash value corresponding to the first iteration loop and a plurality of pre-configured initial values, generating an initial value of the first iteration round in the next iteration loop, entering the next iteration loop, and when the current iteration loop is the last one, determining a final hash value corresponding to the current iteration loop as a target hash value. The problem that a large number of logic resources need to be consumed during data hashing, and consequently the core area of an algorithm is large can be solved.

An embodiment of the disclosed invention is a computer-implemented method to build a schedule, which includes establishing a scheduling period and a number of tasks to be accomplished during the schedule, assigning an expected value to each of the tasks, and using an iterative RIDES loop to adjust the expected value for each task and rebuild the schedule until the expected value is optimized. Another embodiment is a computer-implemented method to optimize a schedule that includes using an iterative RIDES loop to determine whether to execute a first schedule allocation or continue to improve the schedule allocation. Another embodiment is a method for implementing a loop architecture to update an expected value for a task using data from the environment, wherein the expected task value is the maximum task value multiplied by a weighting factor.

The invention belongs to the technical field of gear transmission systems, and particularly discloses a gear meshing optimization method and system based on random progressive iteration and regularization, and the method comprises the following steps: carrying out the data sampling and preliminary fitting based on a gear tooth surface curve model; aiming at the preliminary fitting, setting an adaptive double-layer optimization framework which comprises a first layer of control point updating strategy optimization and a second layer of parameter distribution optimization, constructing an objective function, and realizing adaptive fusion optimization in the same iterative loop to obtain an optimized involute curve and a curved surface; and for the meshing region, introducing second-order difference smoothing, carrying out multi-objective optimization, and outputting a finally fitted NURBS control point P and an optimized parameter mapping weight w. By adopting the technical scheme, on the basis of least square approximation (LSPIA), the adaptive double-layer optimization technology is integrated, and the fitting precision, the fairness and the convergence speed are remarkably improved by jointly optimizing a control point updating strategy and parameter distribution.

The present invention relates to a sampling apparatus and method robust against side-channel attacks, the method comprising the steps of: (a) receiving a random bit; (b) setting an iteration loop index i to 0 and determining whether the random bit is greater than or equal to NRCDT[i]; (c) if the random bit is greater than or equal to NRCDT[i], an adjustment value in an i-th iteration loop is set to 1, and if the random bit is smaller than NRCDT[i], the adjustment value in the iteration loop is set to 0; (d) updating a random integer by subtracting the adjustment value from 18; and (e) repeating steps (b) to (d) while increasing i by 1 until i reaches 17, thereby obtaining a finally updated random integer, wherein NRCDT[i] has a value obtained by bit-inverting RCDT[i] of a base sampler.

A data processing service generates for iteratively applying a geospatial function to geospatial data. The generated code includes at least a first iterative loop and a second iterative loop. The data processing service compiles the generated code to generate compiled code that vectorized at least the second iterative loop. The data processing service receives a request from a client device to perform one or more data processing operations including applying the geospatial function to a data table of geospatial cell indices. The data processing service compiles the request into one or more tasks including at least a vectorized operation based on the compiled code and executes the one or more tasks by at least invoking the vectorized operation on the set of worker nodes.

The invention is suitable for the technical field of information security, and provides a fixed-length message processing method, device, equipment and product based on an SHA-3 algorithm, and the method comprises the steps: carrying out the message filling processing of an original message, and obtaining a to-be-processed message; detecting whether the current working state of a pipeline structure established based on an SHA-3 algorithm satisfies a preset working state, the pipeline structure being constructed based on inserting registers among a plurality of round functions of an iterative loop, and the plurality of round functions being used for simultaneously and circularly processing a first predetermined number of to-be-processed messages in the same clock period; and if it is detected that the current working state meets the preset working state, round function loop processing is performed on the to-be-processed message based on the pipeline structure. According to the method, circular processing of the to-be-processed messages by the pipeline structure can be effectively controlled, the pipeline structure can simultaneously process the first preset number of to-be-processed messages in the same clock period, and the clock frequency of hardware design and the Hash calculation efficiency are improved.

The application discloses a flexible job shop scheduling method and device based on sub-batch division, and comprises the following steps: in the iteration cycle step, individual sorting is performed on the initial population according to fitness, a guide wolf individual is selected from the sorted population, and an individual is selected to establish an external file; the guide wolf individual is used to guide and optimize the subordinate wolf individual in the population through the grey wolf optimization algorithm; the coding of the individual in the initial population is randomly adjusted based on sub-batch division to expand the search range; the coding of the individual in the initial population is optimized and adjusted in a local structure; and the external file established by the individual selected from the population is output, and a scheduling scheme is selected from the external file. By using the above technical scheme, the calculation amount of production scheduling can be reduced, and unnecessary time consumption can be reduced through reasonable sub-batch division; the population evolution is accelerated, and the convergence speed is accelerated by designing multiple guide mechanisms; and for the decision wolf individual, a local search based on a key task is designed to find a better solution.

The present invention provides a task adaptive generation method, system, storage medium and electronic device based on loop feedback. The method includes: Step S1: generating an artificial intelligence operation task based on a test instance list and artificial intelligence operation feedback; Step S2: processing the artificial intelligence operation task to obtain an artificial intelligence operation result; Step S3: returning the artificial intelligence operation result as new artificial intelligence operation feedback to Step S1; Step S4: iteratively and circularly executing Step S1 to Step S3. The present invention realizes automatic task generation, dynamically adjusts according to the artificial intelligence operation result, avoids mechanically listing single-type questions, effectively couples the generated task with the artificial intelligence operation model, and truly realizes the loop optimization of automatic task generation and problem solution.

The present invention discloses a method, system and readable storage medium for optimizing video memory in distributed training of a model, including: obtaining the parameter quantity and access order of each operator during the first round of training of the model, generating an execution order graph after uniquely numbering the operators and storing it; evenly splitting each operator and recording the shard mapping relationship; during the training process, according to the execution order of the operators, obtaining the complete parameters of the initially running operator through a computing device; performing the calculation of the current operator, and prefetching the parameters of the next operator to be run through asynchronous communication according to the operator execution order graph; after the current operator is executed, only retaining the original partial operator parameters and releasing the remaining partial operator parameters; synchronizing the operator parameter communication, verifying the integrity of the shard parameters and then performing the training iteration, and looping the above process until the training is completed. By using the present invention, while reducing the video memory occupation of a single computing device, the efficient training of the model can be ensured.

The invention discloses a heterogeneous equipment maintenance task distribution system and method based on A2C reinforcement learning, and belongs to the technical field of reinforcement learning. The implementation method comprises the steps that 1, a server side establishes an A2C-GA hybrid model used for distribution of maintenance tasks of heterogeneous equipment in a workbench-task state, and hyper-parameters are configured for the model; the client side carries out parameter initialization on the working table-task state of the server side; 2, the server side updates the Actor network and the Critic network by utilizing action state interaction evaluation of the task allocation set, and performs parameter updating on the A2C model by utilizing the gradient loss of the Actor network and the Critic network; 3, executing the step 1 and the step 2 in an iterative loop mode until the fitness is improved to be less than 1%; 4, the server side visually outputs the task allocation set and the total time of task completion through the client side; compared with the prior art, the technical problem of improving the equipment maintenance task allocation accuracy in the existing heterogeneous equipment maintenance task allocation is solved.

The invention provides a task adaptive generation method and system based on loop feedback, a storage medium and electronic equipment, and the method comprises the steps: S1, generating an artificial intelligence job task based on a test instance list and artificial intelligence job feedback; s2, processing the artificial intelligence operation task to obtain an artificial intelligence operation result; s3, the artificial intelligence job result is used as a new artificial intelligence job to be fed back to the step S1; and S4, iteratively and circularly executing the steps S1 to S3. According to the method, automatic task generation is realized, dynamic adjustment is carried out according to an artificial intelligence operation result, mechanical listing of single-type questions is avoided, the generated task and the artificial intelligence operation model are effectively coupled, and automatic task generation and question answering loop optimization are truly realized.

A method, a system, and a non-transitory computer readable medium are provided. The method includes performing, by one or more computing devices, a lookahead scan of a file of a circuit design to extract information associated with a query in an iterative loop, performing an action to retrieve attribute information from one or more partitions of the circuit design before executing the iterative loop, and querying the iterative loop using the stored attribute information. The action stores the attribute information based on the extracted information.

The invention belongs to the technical field of intelligent control and energy conservation of a thermal power plant, and particularly relates to a full-process optimization method for a pneumatic ash conveying system based on multi-mode sensing fusion, dynamic digital twin modeling and self-adaptive reinforcement learning control and a closed-loop verification and self-healing system. A sensing-analysis-decision-execution-self-optimization closed-loop architecture is constructed, and real-time working condition matching, energy consumption optimization and equipment service life prolonging in the ash conveying process are achieved through data communication collection, data analysis and calculation, control result output, control instruction execution and system iterative circulation. The technical problems of control lag, high energy consumption and fast equipment loss of a traditional system are solved.

The invention discloses a micro-grid optimization scheduling method based on an improved MOJAYA algorithm, and the method comprises the steps: firstly setting a micro-grid objective function and constraint conditions, and initializing a population: generating an initial population and an empty archive set; the initial population and the empty archive set are combined, the total fitness F (i) of all individuals in the combined set is calculated, the total fitness F (i) is composed of the original fitness R (i) and the congestion density D (i), and the archive set is updated; and finally, identifying an optimal individual and a worst individual through the total fitness F (i), and updating population individuals through an MH sampling method. Iteratively circulating, and finally outputting an archive set, individuals in the archive set representing the power generation output scheme of each micro power supply after optimization scheduling of the micro-grid. According to the method, the problems that in the prior art, a comprehensive micro-grid is poor in global search capability and prone to falling into local optimum are solved.

The invention discloses a model distributed training video memory optimization method and system and a readable storage medium, and the method comprises the steps: obtaining a parameter quantity and an access sequence of each operator in a first training process of a model, carrying out the unique numbering of the operators, generating an execution sequence diagram, and storing the execution sequence diagram; performing uniform segmentation on each operator, and recording a fragment mapping relation; in the training process, according to the execution sequence of the operators, obtaining complete parameters of the initially operated operators through a computing device; calculating a current operator, and pre-fetching parameters of a next operator to be operated through asynchronous communication according to the execution sequence diagram of the operator; after the current operator is executed, only part of original operator parameters are reserved, and the rest of operator parameters are released; and synchronizing operator parameter communication, verifying the integrity of the fragment parameters, then executing training iteration, and circulating the process until the training is completed. By utilizing the method, the efficient training of the model can be ensured while the video memory occupation of a single computing device is reduced.

The invention discloses a feasible pump acceleration column and constraint generation method, and relates to the technical field of energy system planning, and the method comprises the steps: obtaining basic parameters of a three-layer mixed integer optimization problem, and constructing an original problem model based on the basic parameters; a column and constraint generation framework is adopted, an original problem model is decomposed into a main problem model and sub-problem models, dual processing is conducted on the sub-problem models, and a single-layer max model is obtained; calculating through a feasible pump accelerated column and constraint generation algorithm to obtain a main problem target solution and a sub-problem target solution, and updating the lower bound and the upper bound of an original problem model; judging whether a convergence condition is met or not, and if yes, directly outputting a target solution; and if not, iterative loop calculation is carried out until the convergence condition is met, and a target solution is obtained. According to the method, the global optimality of the solution can be guaranteed, the solving efficiency can be remarkably improved, and the method has good expandability and is suitable for solving the large-scale three-layer mixed integer optimization problem.

The invention discloses a security constraint unit commitment efficient solving method and system based on neighborhood search, and relates to the technical field of power system operation optimization, and the method comprises the steps: setting an initial iteration frequency, constructing a UC model without network security constraint, and obtaining an initial UC solving result; according to a UC model solving result, safety checking is carried out on the power transmission line, and a specific unit set is screened; building a neighborhood search cutting plane based on a specific unit set, adding the neighborhood search cutting plane into the UC model, and solving the UC model again; and after the UC model is solved again, iterative circulation is carried out, and the unit operation state and the planned output are output. According to the method, the search space of the mixed integer programming problem is effectively reduced, the invalid search range is reduced, the solving speed of the security constraint unit commitment problem is increased, the strict requirement of real-time scheduling of a power system on the solving time can be met, and the method is suitable for large-scale popularization and application. The unit combination scheme obtained through efficient solution can realize optimization of the system operation cost, and the economic benefit of the power system is improved.