17results about How to "Increase generation speed" patented technology

The invention relates to the technical field of computer software, and discloses a bridging method and system for a cross-language interface of KaihongOS, a terminal and a storage medium. The method comprises the steps that an interface definition file is analyzed, and function declaration information of a target interface is obtained; analyzing a native interface definition file to obtain implementation information of a native interface corresponding to the target interface; and based on the function declaration information and the implementation information, according to a preset bridging rule, automatically generating a bridging function code conforming to a target operation environment, and bridging a cross-language interface through the bridging function code. According to the method, through automatic analysis and code generation, the bridging function code meeting the target operation environment is automatically generated by improving the code quality, and repeated labor and potential errors in manual writing are avoided. A preset bridging rule ensures the consistency and maintainability of codes, the technical threshold can be effectively reduced, and the project cycle is shortened.

The invention discloses a cell hologram automatic focusing and aberration compensation phase reconstruction method based on deep learning, and the method comprises the steps: obtaining a to-be-processed off-axis digital hologram of a cell sample based on an off-axis digital holographic microscopy system, and carrying out the preprocessing of an original hologram, and enhancing the data size; processing the holograms in batches by using a phase reconstruction method considering aberration compensation to generate corresponding phase diagrams, and forming a ''hologram-phase diagram'' image pair data set for model training; the image pair data set is used for training a pix2pix network model, the input of the model is a cell hologram, and the output of the model is a focusing phase diagram; according to the invention, automatic focusing and phase reconstruction can be simultaneously realized for a single hologram, a clear and distortionless in-focus phase diagram is directly output, the problem that only an object in a specific distance range can be clearly imaged is solved, and the method is suitable for cell observation, label-free pathological detection and dynamic analysis of living cells in the field of biomedicine.

The application discloses an adversarial sample generation method based on AdvDrop, relates to the technical field of machine learning security, and processes two different branches of an image input space domain and a frequency domain; for the frequency domain attack AdvDrop, first, the input image is segmented into N*N blocks, and discrete cosine transform (DCT) is used on each block to convert them to the frequency domain; a quantization matrix M is introduced to reduce some specific frequencies of the transformed image; a tangent function is introduced in the quantization process to gradually approach the quantization function, and then the quantization matrix M is accurately adjusted through the new quantization function; then, the image is converted from the frequency domain to the space domain through inverse discrete cosine transform (IDCT) operation; finally, the space domain attack and the frequency domain attack fusion module are used to iteratively update the adversarial perturbation by using the gradients from different fields, and the adversarial sample is generated; the quality of the generated adversarial sample is improved, the difference between the distribution characteristics of the adversarial sample and the distribution characteristics of the real sample is reduced, and the attack success rate is improved.

This application discloses a method, apparatus, storage medium, and program product for generating a compilation database. The method includes: starting the execution of a file compilation command corresponding to a target file; generating execution status information of the file compilation command and sending the execution status information to an interceptor; enabling the interceptor to intercept the file compilation command when the execution status information indicates that the execution status of the file compilation command meets preset interception conditions, thereby obtaining initial compilation attribute information corresponding to the file compilation command; receiving target compilation attribute information sent by an analyzer; and generating a compilation database corresponding to the file compilation command based on the target compilation attribute information. This application achieves the generation of a complete compilation database, improving the accuracy of static code analysis results corresponding to the target analysis tool.

This application discloses a data processing method, a question-answering method, an electronic device, a storage medium, and a product, belonging to the field of artificial intelligence technology. The data processing method includes: setting a first character for the reference answer of the i-th round question, wherein the first character is defined as a summary of the reference answer; predicting the predicted answer of the i-th round question and a second character corresponding to the predicted answer based on the previous i-1 rounds of dialogue, the i-th round question, the reference answer of the i-th round question, and the first character using a preset model, wherein the second character is used to instruct the preset model to generate a summary for the predicted answer of the i-th round question; determining a first loss based on the reference answer of the i-th round question and the predicted answer of the i-th round question; determining a second loss based on the second character and the reference answer of the i-th round question; and updating the parameters of the preset model based on the first loss and the second loss, thereby improving the accuracy of the answers generated by the trained first model and increasing the speed at which the model generates answers.

The application relates to a processing method and device for a lane center line, electronic equipment and a storage medium, and comprises the following steps: determining a rotation radian; determining, in a preloaded cache file, a first radian smaller than the rotation radian, closest to the rotation radian, or equal to the rotation radian; obtaining coordinate points under a plurality of radian between zero and the first radian from the cache file; if the rotation radian is equal to the first radian, sequentially connecting the coordinate points to generate a target spiral; if the rotation radian is greater than the first radian, controlling a spiral generation program to generate a spiral between the first radian and the rotation radian based on the obtained first radian, obtaining a coordinate point corresponding to the rotation radian, sequentially connecting coordinate points under a plurality of radian between zero and the rotation radian, and generating a target spiral; and the target spiral is used for connecting the first lane center line and the second lane center line. The application reduces the operation amount of integration, improves the generation speed of the spiral, and improves the processing speed of the lane center line.

The invention discloses a data interaction method and related products thereof, and the method comprises the steps: carrying out the interpretation of obtained source data, and generating basic interpretation data; understanding the basic interpretation data by using the data understanding large model, and generating application understanding data expressing potential contents of the basic interpretation data; and in response to a user demand, processing the application understanding data and / or the basic interpretation data by using the data processing logic generated by the demand analysis large model to generate a data processing result. According to the scheme, the basic interpretation data can be understood by utilizing the large language model, and the application understanding data containing the potential content of the data is generated. The understanding data is used as the application link layer to establish a deepened link between the basic data layer and the demand response layer, so that accurate response to the user demand is realized, and the effective processing and generation speed of the data for the user demand is improved.

The application discloses a random test case generation method. The method comprises the following steps: generating at least one random factor, and acquiring a current test step in a plurality of type step libraries matched with a test object according to the value of each random factor; if the current test step satisfies a legality condition according to a generated test step set, the current test step is added to the generated test step set; returning to the operation of generating at least one random factor until the number of test steps in the generated test step set reaches a target number threshold; and constructing at least one random test case according to the generated test step set at the end of the cycle. Through the above technical scheme, the test case of the random scene can be automatically generated according to the test requirement, and the coverage of the test is effectively improved.

This invention relates to the fields of cryptography and generative steganography, and in particular to a generative steganography method and system based on asymmetric digital systems. In this invention, the sender and receiver agree on hyperparameters, hints, and a generation model; the hyperparameters include precision Pre and control factor sca; the sender uses a secret message m as an ANS state, and performs generative steganography on the secret message m using ANS decompression to obtain the steganographic text of the secret message m and the final ANS state; the receiver receives an embedded message and uses an ANS compression process to restore the secret message m; the embedded message includes the number of generation steps, the steganographic text, and the final ANS state. This invention, through a systematic reconstruction of ANS and the introduction of innovative mechanisms, solves the long-standing trilemma in the field of generative steganography, achieving significant results.

The application relates to a method for generating an intracranial aneurysm blood vessel model and a computer readable storage medium, the method comprising: acquiring a three-dimensional medical image containing an intracranial aneurysm; segmenting the three-dimensional medical image to obtain a first level set based on an intracranial artery vessel surface; screening an interested region containing an intracranial aneurysm from an intracranial aneurysm blood vessel model generated based on the first level set; extracting a center line and a radius along the line for the interested region, the extraction of the center line comprising extraction of an intracranial artery center line and extraction of an intracranial aneurysm center line; generating an initial pipe model according to each center line, and then obtaining a second level set based on a surface of the initial pipe model, and generating an intracranial aneurysm blood vessel model after segmentation. The application optimizes the cumbersome preprocessing operation in the traditional method, improves the generation speed of the model on the basis of ensuring the accuracy of the model.

The invention discloses a polymer formula analysis method and device based on a graph Q learning network, and the method comprises the steps: building a Markov decision process of formula generation through defining a state space, an action space, a state transition probability and a reward function; constructing a graph Q learning network model based on the Markov decision process; training the graph Q learning network model through a double-Q learning algorithm and a strategy function to obtain a trained graph Q learning network model; and performing state transition and action selection processing through the trained graph Q learning network model to generate a polymer formula. According to the method, the problems of low learning speed, high computing resource consumption, difficulty in accurately matching multi-target performance and the like in the prior art are solved.

This application provides a deep learning-based intelligent scheduling method and system for gateway data transmission. The method includes: acquiring three-dimensional feature tensors corresponding to signaling streams, media streams, and user behavior data; performing spatiotemporal feature joint prediction on the three-dimensional feature tensors based on a pre-set dual-channel LSTM neural network to output a link bandwidth demand matrix and a risk warning level for burst traffic; establishing a bandwidth allocation decision tree based on the link bandwidth demand matrix, and determining a real-time bandwidth threshold based on the bandwidth allocation decision tree and the risk warning level; constructing a reinforcement learning state vector based on the current communication quality indicators, the link bandwidth demand matrix, and the real-time bandwidth threshold; parsing the reinforcement learning state vector through a deep Q-network to obtain path decision instructions, and sending the path decision instructions to the corresponding gateway routing nodes.

The application discloses a kind of intelligent pavement friction performance evaluation method and system, method includes multi-source data acquisition, pavement image depth feature extraction, vehicle-road interaction friction performance evaluation and pavement friction performance dynamic evaluation.The application relates to the field of data processing, specifically refers to a kind of intelligent pavement friction performance evaluation method and system, the scheme of the application innovatively proposes to carry out vehicle-road interaction friction performance evaluation first, then the vehicle-road interaction result is combined with environmental data to analyze, accurate and comprehensive pavement friction performance evaluation is realized;Active shape model is used to extract pavement macrostructure features, and improved local binary texture method is used to extract pavement microstructure features, to improve the accuracy of vehicle-road interaction friction performance evaluation results;The calculation efficiency is improved, and the accuracy of the model output result is improved by introducing probability sparse self-attention mechanism, distillation mechanism encoder and one-time generation decoder to improve the Transform algorithm.

The application provides a data check code generation method, a central processing unit, a computer device and a medium, and relates to the technical field of data processing. The data check code generation method comprises the following steps: obtaining target data by using a load storage unit, wherein the target data is data to be generated into a check code; dividing the target data into first data and second data according to the data processing capacity of a CRC calculation component and a cryptography extension component by using the load storage unit; processing the first data and the second data respectively by using the CRC calculation component and the cryptography extension component to obtain a first check code and a second check code; and merging the first check code and the second check code by using the CRC calculation component and the cryptography extension component to obtain a target check code corresponding to the target data. Two components are used to process the first data and the second data obtained by division, the performance of the central processing unit is fully utilized, and the generation speed of the target check code is improved.

This application relates to the field of power system emergency control technology, specifically to a method and system for generating rolling clearing schemes for post-disaster distribution network recovery. The method includes: acquiring real-time data of the distribution network; constructing a context feature vector for the current moment, inputting it into a pre-trained decision generation network for forward inference, and outputting a preliminary clearing decision within a rolling window; using a pre-trained physical feasibility discrimination network to perform security verification on the preliminary decision; if the confidence level is lower than a threshold, correcting it through linear projection to obtain the final clearing decision; issuing and executing the first step instruction in the final clearing decision, and then rolling the decision time to the next time step for cyclical execution. The decision generation network and the discrimination network are co-trained using a simulation dataset, which is generated by solving a single-leader-multiple-follower model reconstructed from a post-disaster recovery single-layer mixed integer linear programming model. This invention enables dynamic closed-loop adjustment and rapid generation of post-disaster recovery schemes.

This invention provides a method and system for random-delay UAV formation control based on an event-triggered mechanism, belonging to the field of UAV formation control technology. The invention establishes a global inertial coordinate system and normalizes the original motion state to a standard motion state; it defines the true value of lumped disturbance using heterogeneous dynamic constraint equations; it designs a nonlinear disturbance observer with auxiliary variables to calculate the estimated value of lumped disturbance approximating the true value; it constructs a full-state coupled dynamic event-triggered mechanism to calculate a dynamic triggering threshold that includes a minimum interval protection term and is negatively correlated with the disturbance magnitude; it corrects neighbor nodes with random communication delays based on the estimated value of lumped disturbance to obtain the predicted neighbor motion state; it uses a stable manifold neural network to infer the optimal costate vector and combines it with disturbance feedforward compensation to generate physical control commands. This invention improves the formation's anti-interference capability while reducing communication load.

The application relates to the field of FPGA debugging technology and discloses an FPGA signal observation method, system, device, equipment and storage medium, wherein the method comprises the following steps: setting a unique identifier for a to-be-observed FPGA function module where a target observation signal is located, and setting the same identifier in an FPGA top-level module; constructing a signal transmission link between the function module and the top-level module based on the identifier, so that the target observation signal is transmitted to the top-level module; and transmitting the signal to a debugging interface through the top-level module to complete online observation. The method solves the problems that the signals need to be repeatedly and manually connected and the observation target cannot be dynamically switched in the prior art through identifier matching and link construction, and the debugging efficiency is significantly improved.