196 results about "Random seed" patented technology

The invention discloses a small sample modeling stability evaluation method and system based on Bootstrap resampling, and particularly relates to the technical field of computers and data intellectualization, the method comprises the following steps: completing data preprocessing and stable stage identification under a unified time base, and forming a segment set; an average block length is used as a decision quantity, an SBB is optimized to construct a sample space, and an optimal block length is determined in combination with variance consistency and nominal coverage rate consistency criteria; carrying out re-sampling training around short / medium / long scales, and collecting layered indexes such as prediction, parameters and features; stability calculation and empirical coverage rate calibration are completed based on intra-scale statistics and inter-scale weighted mixing, and a pseudo-stationary diagnosis score is output; and finally, engineering judgment and backspacing optimization are carried out according to the coverage rate, the correlation maintenance and the risk threshold. The system records random seeds, block metadata and model version generation evidence pointers, has the characteristics of traceability and reverifiability, and is suitable for scenes of production lines, network traffic, finance and the like.

The invention discloses a digital archive multi-protection-based data security transmission method, and relates to the field of data security transmission, and the method comprises the steps: generating a random seed in real time based on a quantum state, carrying out the iterative blocking of the quantum seed through combining with Logistic chaotic mapping, outputting variable-length secret key fragmentation, carrying out the symmetric encryption of archive data through employing a lightweight SM4 algorithm, and carrying out the encryption of the archive data. Obtaining encrypted data of the fragments; a behavior anomaly probability is calculated through a hidden Markov model, a risk score is calculated through a safety interval, and when a score value exceeds a safety threshold value, a grading response is triggered; key fragments are generated by adopting n nodes, at least k fragments can reconstruct signatures to monitor the node state of the block chain, and when the node anomaly rate exceeds a node threshold value, a three-layer architecture is automatically switched. The method has the advantages that an active multi-protection system is constructed through a quantum dynamic key, risk-driven hierarchical response and block chain disaster recovery switching, and quantum attack resistance and dynamic adaptation to transmission of network risks are realized.

The invention discloses a mobile payment encryption system based on a data random algorithm, and the system comprises a collection module which is used for collecting multi-modal data, and constructing a cross-modal entropy source pool; the random seed generation module is used for extracting feature vectors from the cross-modal entropy source pool and generating initial random seeds; the time sequence self-adaptive random chain module is used for generating a random factor sequence based on the initial random seed recursion and forming a random chain structure; the encryption module is used for jointly encoding the random factor sequence and transaction data to be encrypted, generating a dynamic encryption parameter set and encrypting the transaction data; and the chain type updating module is used for collecting network environment parameters, updating the time sequence self-adaptive random chain and generating a new dynamic encryption parameter set so as to form a chain type evolved encrypted data sequence. According to the method, the cross-modal entropy source and the time sequence random chain are utilized to realize dynamic encryption of the mobile payment data, and the method has the advantages of high entropy randomness, strong attack resistance and high transmission security.

The invention provides a parallel multi-target random parameter optimization method and device.The parallel multi-target random parameter optimization method comprises the steps that the number of nodes of a parallel computing cluster is configured, and a random seed sequence is generated; carrying out structure loading and compiling on the to-be-optimized model, and obtaining global configuration of the to-be-optimized model; sending the global configuration to all processes, generating an initial parameter population meeting the constraint of the global configuration by each process based on the random seed, and performing serial evaluation to obtain a target function value; independently iteratively executing shuffling complex evolution by each process until a convergence condition is met; each process outputs the respective optimal parameter vector and the corresponding objective function value to a root process, and a global optimal solution is determined through the root process. By means of the method and device, the accuracy and efficiency of parameter optimization can be remarkably improved, and the problem that the model optimization effect is poor in the prior art is solved.

The invention discloses a secondary power modulus-based post-error-correction-free quantum lattice signature method, which comprises a counter, and comprises the following steps of: S1, sampling a random seed, and expanding the seed through a hash function to obtain a byte stream element I, a byte stream element II and a byte stream element III; s2, calculating and generating a public key and a private key based on the throttling element I, the byte stream element II and the byte stream element III; s3, assigning initial values to the counter and the signature, and calculating and generating a hash value I and a hash value II based on the private key and the signature message; s4, performing signature loop calculation based on the hash value I, the hash value II, the private key and the initial value of the counter to generate a legal signature final value, and returning the legal signature final value; s5, a signature verification variable is generated through calculation based on the public key and the signature final value, if the signature verification variable meets a verification condition, signature verification is passed, otherwise verification is not passed, the post-quantum lattice signature calculation efficiency, bandwidth optimization and stability are greatly improved, and a more efficient solution is provided for practical application.

The embodiment of the invention provides a large language model optimization method based on variance reduction and momentum acceleration. The method comprises the following steps: a gradient estimation stage of a large language model: initializing a seed list and a projection list; executing multiple independent query iterations, calling a disturbance subprogram, generating a random seed for the large language model, and storing a table; in the disturbance subprogram, the determined gradient projection value is stored in a projection list, and iteration is carried out for next query; after executing multiple independent query iterations, storing a plurality of random seeds and a plurality of gradient projection values corresponding to the random seeds; in the weight updating stage of the large language model, a gradient norm subprogram is called for each layer of the large language model, and a random seed is obtained to reset a random number generator; and determining variance-reduced gradient estimation according to the gradient projection value extracted from the projection list and the reproduced disturbance vector. According to the embodiment of the invention, gradient information queried for multiple times is aggregated to generate low-noise gradient estimation, and fine tuning of a large language model is completed.

The invention discloses a security encryption method, device, equipment and medium for a vehicle-gauge-level HSM, and relates to the technical field of vehicle network security, and the method comprises the steps: obtaining a target algorithm parameter corresponding to a target encryption task in an algorithm parameter register after receiving the target encryption task, and determining a corresponding encryption algorithm type; if the target algorithm type is a post quantum cryptography algorithm type, a true random number generation component of a vehicle gauge level HSM is called through a post quantum cryptography coprocessor to generate a true random seed, and a target encryption task is executed based on the target algorithm parameter and the true random seed; in the process of executing the target encryption task, second-order mask operation is carried out on the fast number theory transformation operation, whether error injection attacks exist or not is continuously detected in the sampling process, and operation results of the key steps are stored through a preset composite register set.

The invention discloses a fine tuning system of a large-scale pre-training model in a federated learning environment and application thereof. The system comprises a local disturbance gradient estimation module, a differential privacy protection module and a global model aggregation and update module. The local disturbance gradient estimation module is used for calculating a global model loss value by combining forward propagation with a zero-order optimization method so as to estimate a gradient and realize all-parameter fine tuning; the differential privacy protection module performs differential privacy protection processing on the estimated disturbance gradient to prevent gradient information from leaking user sensitive data; and the global model aggregation and update module reconstructs a disturbance vector and completes global model update based on a random seed and a scalar gradient uploaded by a client. Compared with the prior art, on the premise of not depending on back propagation, all-parameter fine tuning of a large-scale pre-training model is achieved, data privacy is guaranteed, meanwhile, calculation and memory expenses are remarkably reduced, and the method is suitable for a resource-limited distributed calculation environment.

The invention relates to a physical unclonable function modeling attack resisting method based on a neural network. The method comprises the following steps: generating a multi-bit random bit stream by using a random seed generated by an LFSR module, and simultaneously acting PUF excitation on a plurality of physical unclonable functions to generate a multi-bit physical unclonable function response; the method comprises the following steps: constructing an initial feature matrix by utilizing a Decoder module, and equally dividing position decoding excitation into a plurality of binary position decoding sub-excitation with a certain length; sequentially guiding multi-bit physical unclonable function responses to storage units of corresponding addresses to form a physical unclonable function response feature matrix, and inputting the physical unclonable function response feature matrix into the neural network model to obtain one-bit confusion response; and the other bit of confusion response of the random seed is reutilized until a final confusion response sequence with a preset bit is generated, and the server compares the final confusion response sequence with an expected confusion response to complete identity authentication. By adopting the method, the hardware security can be improved.

A data processing method performed on a verification node and including obtaining task event information and a signature for the task event information from a scheduling node. The task event information includes a first computing node set participating in execution of a computing task and a random seed provided by a task submission end of the computing task. The method further includes performing verification on the signature, deriving a random number set based on the random seed in response to the verification of the signature succeeds, selecting, for the computing task, a second computing node set associated with the random number set, verifying validity of the first computing node set by using the second computing node set, to obtain a verification result, and adding a block including the verification result to a blockchain of a verification node cluster including the verification node.

The invention provides a sequence decision strategy acquisition method and system for high-neuroplasm crowds. The method comprises the following steps: establishing a full-screen drawing canvas, and loading a task set comprising a decision map type, an emotion titer ratio, a reward parameter, a random seed and a sequence control parameter; collecting user gender selection and binding a corresponding emotion picture library; generating a decision map based on a Markov decision principle under the control of the parameter set, defining nodes and connecting edges, randomly scheduling and laying emotion pictures according to a titer ratio, and assigning rewards to form an emotion and reward coupling environment; based on the environment, recording a mouse position and a mouse path in an execution exploration stage, recording key pressing time, an action direction and an award in an execution stage, and establishing step sequence association according to a timestamp to form sequence decision data; a Latin square balance control task sequence is adopted, and the process is propelled under the trigger of a user; and finally, filing the data in a binary system and a text file for subsequent analysis. According to the invention, the authenticity and repeatability of data are improved.

The invention provides a Python bytecode obfuscation method and system based on dynamic screening and random replacement, and the method comprises the steps: calculating a plurality of safety indexes and a comprehensive safety index SDI of bytecode files, and screening out the bytecode files which have high safety requirements and need to be obfuscated; a linear congruence generator is constructed, a pseudo-random number sequence is provided for obfuscation operation of byte codes, and operation code dynamic replacement obfuscation operation is performed in byte code instructions by generating random seeds, constructing random dynamic operation code mapping tables and applying the independent operation code mapping tables to different byte code files needing to be obfuscated; and performing obfuscation state detection on the obfuscated bytecode file, if the detection result is the obfuscated bytecode file, modifying bytecode analysis and instruction scheduling logic, increasing a runtime de-obfuscation process, embedding an Opcode reflection mechanism, re-compiling to generate a customized Python interpreter supporting the execution of the obfuscated bytecode, and executing the obfuscated bytecode file. According to the invention, byte codes can be confused.

The invention discloses an automobile ECU and UDS security diagnosis method and system based on a dynamic session key, and the method comprises the steps: in the diagnosis process, the ECU generates challenge information containing a random seed, a unique session identifier and an initial counter, and calculates an expected key in combination with a hardware identifier and a main key; and the diagnostic instrument calculates a client key and returns the client key by using the same hardware identifier, the master key and the received challenge parameters, and the ECU completes authentication by checking the session identifier, the counter and the key. And after the authentication is passed, the two parties independently derive a temporary session key by taking the master key as a root key and combining the dynamic factor of the session. When the diagnostic instrument sends a request, the counter is increased progressively, an authentication code is generated in combination with the session key, and the request, the authentication code and the counter are sent to the ECU; and the ECU checks the continuity of the counter and the consistency of the authentication code to verify the legality of the request, and also generates the authentication code to be verified by the diagnostic apparatus when sending a response, thereby completing bidirectional verification. And the diagnosis safety level of vehicle-mounted electronic equipment and the like is improved.

The invention provides a watermark detection method and device based on Bayesian detection. The method comprises the steps of obtaining a target text to be subjected to large model watermark detection and a cue word corresponding to the target text; processing a token sequence spliced by the cue word and the target text through a language model to obtain probability distribution output by the language model for each token position; for the jth token in the T tokens of the target text, selecting the first k tokens before the jth token to be input into the hash function, and obtaining a random seed of the jth token; dividing a word list of the large model into a preferential selection set and a non-preferential selection set based on random seeds; on the basis of preset watermark bias, preferentially selecting the set and the probability distribution, and generating probability distribution after disturbance of the jth token; calculating a log-likelihood ratio by using the probability distribution before and after disturbance, and accumulating the log-likelihood ratio to a detection score of the current text; and if the detection score is higher than the threshold value, judging that the target text has the watermark of the large model.

The invention relates to the technical field of electronic security, and discloses a vehicle-mounted ECU identity authentication method and system based on a multi-factor random seed and multiple security levels, and the method comprises the steps: synthesizing a random seed at an ECU end through multiple factors such as CRC32, a random number, a timestamp and a global seed, and returning the random seed to a client; the client calculates a key and returns the key by using a corresponding key algorithm according to the target security level; the ECU calculates an expected key through the same algorithm and compares the expected key with a client key, if the expected key is consistent with the client key, the session of the level is opened and the delay timer is closed, if the session fails, the number of failure times of the level is accumulated, and if a threshold value is reached, the delay timer of the corresponding security level is started and the access of the level is locked until the timing is finished; and meanwhile, the authentication result, the security level, the timestamp, the failure count and the seed state are written into a security log. According to the method, it is guaranteed that each time of seed is unpredictable and is strongly correlated with the historical state, a non-zero recalculation mechanism is matched, hidden dangers of fixed seeds, replay attacks and prediction attacks are thoroughly eliminated, and attacks can be protected, and events can be traced.

The invention discloses a post-processing device of a true random number generator, which comprises a random seed simulation module used for generating random seeds; the first processing module is used for performing accelerated multiplication on each random seed group formed by the random seeds and each original random number sub-vector formed by the original random number vector output by the true random number generator based on fast Fourier transform to obtain a first result corresponding to each random seed group; or extracting bits in the original random number vector through a single-bit extractor according to each random seed group formed by the random seeds to obtain a first result corresponding to each random seed group; splicing all the first results to obtain an intermediate result; and the second processing module obtains an actual random number vector according to the intermediate result. According to the method, bit correlation defects can be processed, original random numbers from a plurality of correlative entropy sources can be processed, and the method has unconditional security independent of any calculation hypothesis or random model.

This application provides a method and apparatus for generating random instruction test cases, relating to the field of automated testing technology. First, target instructions or instruction sequences are encapsulated into C inline functions using inline assembly or macro definition, forming a unified function library. Next, a function parameter constraint system is constructed, defining parameter types, address ranges, instruction combination methods, and value range constraints for each encapsulated C inline function. Then, a multi-level nested structure is built based on the C inline functions to organize the instructions, where each level node is associated with a corresponding C inline function name and constraints. Finally, based on a YAML configuration file, the target C inline functions are determined from the multi-level nested structure and combined to generate random instruction test cases. The YAML configuration file includes at least the number of iterations, a random seed, a random strength factor, and test content configuration. The solution provided in this application has the advantages of improved verification efficiency and applicability to different scenarios.

The application discloses a tamper-proof watermark drawing method and system and a terminal, and relates to the technical field of information security and content protection. The method comprises the following steps: obtaining at least one basic template from a pre-established marking rule library, combining a current time stamp and random seed data to generate initial marking data, and ensuring that the initial marking data has unpredictable characteristics to obtain a preliminary dynamic marking result. The tamper-proof watermark drawing method and system and the terminal realize the safe protection, controllable distribution and accurate tracing of digital content, effectively prevent the content from being tampered with or unauthorized access, and improve the security and flexibility of digital copyright management.

The application discloses a large model watermarking method based on dynamic entropy, belonging to the field of large model watermarking. It includes: modifying the output token sampling process; dynamically adjusting the size of the sliding window and the size of the candidate token pool according to the probability distribution entropy value of the large model output; through pseudo-random scoring and multi-round competition; for each token in the sequence to be analyzed, the same watermark key and the context before the token are used to recalculate the pseudo-random seed, and based on the seed, the corresponding pseudo-random score of the token at all levels is regenerated; the average pseudo-random score of all tokens in the text is calculated as the aggregate score, and the statistical threshold is determined based on the preset false positive rate. If the statistical threshold is less than the aggregate score of the unknown text, it is determined as a generated text with watermark, otherwise it is a text without watermark. The application realizes high-precision source identification and tracing of large model output content through the dynamic entropy watermarking method while ensuring the original quality of the generated text.

This invention discloses a test case screening method, apparatus, medium, product, and device, comprising: obtaining pre-configured command-line parameters, including a coverage report input path and a screening threshold; reading a coverage report according to the coverage report input path, and extracting the test case name, random seed, and coverage information of each test case in the coverage report; clustering test cases with the same test case name but different random seeds according to the extracted test case name, random seed, and coverage information; obtaining the total contribution of test cases to coverage in each test scenario based on the clustering results of all test cases in the coverage report; and screening test cases according to the screening threshold and the total contribution of test cases to coverage in each test scenario to obtain screening results. This invention can improve the accuracy and efficiency of test case screening results.

The application relates to the technical field of mainboards, in particular to an industrial mainboard security starting method based on a hardware root of trust, which comprises the following steps: after the industrial mainboard is powered on and reset, a security control domain runs prior to a business processing domain, and a main processor is kept waiting for starting; the security control domain reads hardware root anchor credentials and an initial boot program, and sequentially performs integrity measurement and signature verification on the initial boot program and a later-stage boot program; after the verification passes, running key materials are derived based on a random seed, device private credentials and a later-stage boot program digest, a running private key is written into a restricted key area, and a running public key is written into a public storage area; when a boot update package exists, a firmware public key certificate, a boot image signature and a replacement identifier are verified, a boot image is loaded after the certificate is valid and the signature passes, starting proof information is generated, an access path is closed and the main processor is released; any verification failure enters a fault processing flow.

Embodiments of the application disclose a database primary key generation method and device, electronic equipment and a storage medium. The method comprises: generating a primary key generation matrix according to a target character set and a target number set, the target character set comprising characters in an ASCII character set; generating a position matrix according to the number of primary key elements in the primary key generation matrix; generating a random integer in a target integer range through a pseudo-random number generator according to random seed information; obtaining a target position element corresponding to the random integer in the position matrix; determining a target position represented by the target position element and obtaining a target primary key element corresponding to the target position from the primary key generation matrix; and repeatedly performing the operations of generating the position matrix and obtaining the target primary key element until a character sequence meeting a target condition is obtained, and taking the character sequence as a database primary key, the character sequence comprising the target primary key element. The embodiments of the application can improve the primary key generation efficiency and security.

The invention discloses a method and a device for batch production of television system keys and a medium, and relates to the technical field of Android television systems. The method comprises the following steps: acquiring batch key generation materials; a GPG algorithm is used to decrypt the Mgkid.gpg, and a batch of Netflix ESNs are obtained; generating a first random seed PFID and a second random seed PFPK in batches; the method comprises the following steps: reading a Mac.txt file, generating a batch of mac.bin files, and then respectively creating corresponding folders according to the mac.bin files; the method comprises the following steps of: respectively distributing a batch of Netflix ESN (Enhanced Sensor Network), a batch of PFID (Personal Frequency Identifier), a batch of PFPK (Personal FPK), a batch of mac.bin and a batch of Widevine Key into each Mac And executing operation on the keys in each Mac folder to obtain a combined key. According to the method and the device for manufacturing the television system keys in batches and the medium, the problems that multiple groups of keys are burnt in at the same time and the binding relation of the keys is recorded are solved, and the production efficiency and the accuracy are improved.

The application provides a kind of unmanned aerial vehicle farmland boundary extraction-oriented dataset generation method and system, determine dataset initialization parameter set, obtain unmanned aerial vehicle farmland image and corresponding mask image according to the dataset initialization parameter set;Based on mask image, the number of reserved slices and random points are calculated, enough points are randomly generated inside the original mask area, to form a random seed point set;Probe boundary detection strategy is used to detect the boundary of the random seed point set, to perceive effective points, to obtain candidate sample points;The spatial distribution of candidate sample points is optimized using the minimum overlap rate sampling algorithm to obtain the final reserved sample points;The final reserved sample points are used as the center points of the slices of the generated dataset to generate the final dataset.The application uses point sampling to generate slices, probe boundary detection strategy and minimum overlap rate sampling algorithm, which truly realizes the saving of computing cost, the reduction of farmland boundary omission and the improvement of dataset quality.

The invention relates to a shuffling machine control method, device and equipment and a storage medium, and belongs to the field of shuffling machines, and the shuffling machine control method comprises the following steps: obtaining entropy sources, and mixing the entropy sources to obtain randomized seeds; generating a control parameter by using a pseudo-random number generation module based on the randomized seed; and controlling the shuffling machine to operate based on the generated control parameters. According to the shuffling machine control method and device, the equipment and the storage medium, by introducing a randomization control algorithm and a real-time feedback mechanism, high randomization and self-adaptive control in the shuffling process are achieved, the shuffling quality is improved, card collision is avoided, and the failure rate of card sticking and one-time rubbing of multiple cards is reduced.

The invention provides a high-throughput random number generation method and generator based on a quantum entanglement source, and relates to the field of quantum computation.The high-throughput random number generation method comprises the following core steps that firstly, in the entanglement debiasing process, a quantum sequence entangled with a random seed is constructed through inference, complete extension mapping of an anti-correlation entanglement quantum sequence is achieved, and the quantum sequence is generated; and generating a random bit sequence with doubled length and unbiased statistics. Secondly, in the full-permutation expansion process, the inter-domain correlation of the full-permutation sequence is eliminated through a shuffling method while sequence length exponential growth is achieved; and thirdly, in the layered asynchronous arrangement process, the intra-domain correlation is further eliminated through two-layer segmentation, asynchronous arrangement and shuffling processing, and finally a random bit sequence which meets the statistical quality requirement and has the length. According to the method, through removal of intra-domain and inter-domain correlation, the unpredictability of output random numbers is remarkably enhanced; and exponential-level high-throughput output is realized while the output random number is ensured to pass through all test items of the NIST SP 800-22.

The application provides a mixed supervision language model distillation method based on bridge domain-park domain decomposition, comprising the following steps: constructing a prefix state sequence, and saving context and task identification. The state information is used for subsequent partition discrimination and error statistics. The probability distribution, hard label, confidence and entropy statistics of the teacher model are obtained for each state. The state risk is calculated, and the bridge domain and the park domain are divided to form a partition label. The partition label can be reused for training weighting and slice analysis. Soft supervision and hard supervision are calculated at the state level, and are dynamically weighted. The joint optimization process keeps the single model parameter space solution, and reduces the process complexity. The final model is selected according to the accuracy, stability, time delay and cost, and an auditable evaluation report is formed. In the evaluation stage, unified benchmarks and fixed random seeds are recommended to improve comparability.

The invention belongs to the technical field of artificial intelligence, and relates to an artificial intelligence-based rhythm evaluation method, which comprises the following steps of: based on a first random seed and a second random seed, processing acquired text data by using a voice generation model to obtain first voice data and second voice data; preprocessing the first voice data and the second voice data to obtain a first target voice and a second target voice; performing feature extraction on the first target voice and the second target voice based on a self-supervised learning voice model to obtain a first feature sequence and a second feature sequence; carrying out editing distance calculation processing on the first feature sequence and the second feature sequence to obtain conversion operation data; calculating the conversion operation data to obtain an evaluation score; and generating a rhythm evaluation result based on the evaluation score. In addition, the rhythm evaluation result can be stored in the block chain. The method can be applied to rhythm evaluation scenes in the financial science and technology field and the digital medical field, and the accuracy and efficiency of rhythm evaluation are improved.

This invention provides a parallel multi-objective stochastic parameter optimization method and apparatus. The method includes: configuring the number of nodes in a parallel computing cluster and generating a random seed sequence; loading and compiling the model to be optimized and obtaining its global configuration; sending the global configuration to all processes, with each process generating an initial parameter population satisfying the global configuration constraints based on the random seeds, and serially evaluating the objective function value; each process iteratively executing a shuffling complex evolution until convergence is met; and each process outputting its optimal parameter vector and corresponding objective function value to the root process, which then determines the global optimal solution. This invention significantly improves the accuracy and efficiency of parameter optimization, solving the problem of poor model optimization performance in existing related technologies.

The application discloses a large code length private key amplification data access method suitable for a quantum key distribution system. Through the combination of DDR and on-chip memory BRAM, the reading of the error-corrected key and the random seed is no longer required to be synchronous, so that the requirement for the external memory can be reduced, the subsequent hardware integration processing is facilitated, the influence of the limitation of the IO speed of the external memory on the PA processing speed is reduced, and under the condition of resource permission, multi-path synchronous calculation can be performed, and the implementation performance of the PA is effectively improved.