3267 results about "Exploit" patented technology

The present invention discloses a system and method for performing artificial intelligence (AI) safety red-teaming with integrated policy fuzzing and adversarial prompting to systematically identify, characterize, and mitigate unsafe or non-compliant behaviors in AI models. The disclosed invention automates the process of generating, executing, and analyzing adversarial test cases through coordinated functional units comprising a policy fuzzing unit, an adversarial prompting unit, an execution sandbox, a telemetry processing unit, a scoring and triage processor, and a cryptographic provenance processor. The system applies grammar-driven and reinforcement-based fuzzing techniques to vary policy descriptors, model configuration parameters, and instruction hierarchies, while a learned adversarial prompt generator synthesizes contextually coherent adversarial prompts optimized for maximum policy violation likelihood. The generated prompts and policy vectors are executed in an isolated, instrumented sandbox that records input-output interactions, timing characteristics, and intermediate representations.

Techniques for leveraging a large language model (LLM) in software development are described. A description of a software development task is received from a user. Data associated with the software system is obtained from a data source. An LLM is prompted to identify at least one aspect of the task which requires clarification from the user, at least partly by providing the obtained data to the LLM and asking the LLM to identify a question for the user which remains unanswered by the obtained data. The question is presented to the user. An answer to the question is received from the user. The LLM is prompted to respond to propose an implementation of the task at least partly based on the data associated with the software system and the answer received from the user. The proposed implementation is received from the LLM and caused to be displayed to the user.

The invention belongs to the technical field of equipment knowledge engineering and natural language processing, and discloses an LLM-based constraint and self-repairing uniformization equipment knowledge graph automatic construction method. The method comprises the following steps: firstly, acquiring equipment related document data through network collection, document arrangement and database query; then, utilizing an equipment domain ontology and constraints as preposed soft and hard constraints, driving LLM to generate semantic intermediate representation, and obtaining candidate triples through structured compiling; then, a self-repairing closed loop is formed through semantic unit testing, logic consistency detection and evidence binding verification, and triples which violate constraints and have conflicts or illusions are automatically recognized and repaired; and finally, entity standard identification, cross-document duplicate removal combination and conflict resolution are realized through cross-segment unification and incremental alignment. According to the method, the fragile path that the LLM directly generates the triple and blindly stores the triple is avoided, the illusion and inconsistency problems are effectively inhibited, and the correctness, interpretability and maintainability of the equipment knowledge graph are remarkably improved.

The systems and methods disclosed herein monitor application (e.g., artificial intelligence (AI) model) operations using interactions between the application and a kernel. The systems and methods disclosed herein intercept, using a kernel interface, one or more function invocations transmitted from the application (e.g., an AI model without model modification). Event record(s) are generated for one or more functions to define process identifiers, resource interaction types, timestamps, and / or resource identifiers. Observed pattern(s) for the application are identified by comparing current event record(s) with previous record(s), and the identified observed pattern(s) are evaluated against reference pattern(s) to generate score(s). Data packet(s) that indicate observed pattern(s), corresponding score(s), and / or cryptographic digital fingerprint(s) of the one or more functions are generated. The data packet(s) are transmitted to distributed ledgers for immutable storage.

The invention discloses an active defense method and system based on a large model, and relates to the technical field of security protection, and the method comprises the steps: intercepting a malicious request of an external attacker, cleaning sensitive information and adversarial samples in the malicious request, and outputting standardized data; injecting the standardized data as training data into a training confrontation sample to optimize a protection model, and ensuring the leakage traceability of the protection model by embedding a digital watermark; and trapping an attacker by deploying a honey spot interface and triggering a countering strategy, generating a dynamic defense rule by using the protection model, and updating the training confrontation sample in real time for continuous optimization of the protection model. Active attack sensing and advanced attack blocking are achieved through malicious request interception cleaning and honey spot trapping countering, dynamic defense rule generation and protection model continuous optimization are combined to adapt to attack iteration, a digital watermark tracing mechanism is matched, an'interception-protection-optimization 'closed-loop full link is constructed, and the security defense capability of the protection model is improved.

The invention provides a node task migration and scheduling system based on digital twinning, and relates to the technical field of computer system structures and data processing. Computing resource interference in a multi-tenant sharing environment is quantized by sensing a cross-tenant noise coefficient and a state synchronization complexity entropy in a node micro-architecture; extracting track features of the mobile terminal, calculating spatial discrete variance, generating a self-adaptive migration decision hysteresis factor, and converting the migration decision hysteresis factor into decision damping to inhibit invalid high-frequency reciprocating migration; constructing a digital twin sandbox before physical cutover, cooperating with a chaos scene injection engine to inject a composite fault operator into a bottom layer, and performing actuarial calculation on service continuity retention after risk adjustment by using a fidelity integrator; and finally, a bottom layer controller is linked through safety baseline comparison to execute physical flow switching. According to the method, network boundary deduction is completed on the premise that physical bandwidth is not consumed, the interruption risk caused by state hard switching is avoided, and smooth transition of stateful services is effectively guaranteed.

The invention provides a heterogeneous computing power resource pooling scheduling platform which comprises a task and resource tensor construction module, a scheduling decision module, a breakpoint state generation module and a rescheduling and recovery module. According to the platform, task requirements and node resources are modeled into a tensor structure in a unified mode, and accurate initial scheduling is achieved through a structured scoring function containing conditional penalty terms. In the task running process, interruption judgment is carried out based on the node dynamic state and the scheduling score, and a breakpoint state tensor containing an execution state is generated. When interruption occurs, a proper node is selected through compatibility screening, and task seamless migration and execution recovery are realized by utilizing the stored state. According to the method, the technical problems of non-uniform resource expression, lack of scene adaptability and incapability of realizing uninterrupted migration in a heterogeneous computing power environment are solved, and the resource utilization rate and the service continuity are remarkably improved.

The invention discloses an interface calling method, electronic equipment and a storage medium, and the method comprises the steps: responding to a received target access request, and pre-loading a target constraint file; analyzing the real intention of the user request by using the target language model, and outputting an initial parameter; mapping the initial parameter into an initial structured parameter according to a type, a structure and a parameter dependency relationship defined in the target constraint file; matching a target cloud interface from the interface list by evaluating the matching degree between the initial structured parameter and the standard parameter specification of each cloud interface in the target constraint file; and verifying the initial structured parameter according to a target parameter specification corresponding to the target cloud interface, and calling the target cloud interface through the initial structured parameter under the condition that the verification is passed. Through the method, the technical problem that the accuracy of the interface parameters output by the model in the related technology is low is solved, and the interface calling efficiency of the cloud platform is improved.

Examples of the present disclosure describe systems and methods for determining exploit prevention software settings using machine learning. In aspects, exploit prevention software may be used to identify processes executing on a computing device. Metadata for the identified processes may be determined and transmitted to a machine learning system. The machine learning system may use an exploit prevention model to determine exploit prevention configuration settings for each of the processes, and may transmit the configuration setting to the computing device. The computing device may implement the configuration settings to protect the processes and monitor the stability of the protected processes as they execute. The computing device may transmit the stability data to the machine-learning system. The machine-learning system may then modify the exploit prevention model based on the stability data.

According to the automatic software development method and device and the computer program, intelligent demand analysis is achieved through a natural language processing technology, dynamic task arrangement is conducted through a finite state automaton, collaborative development is completed by means of a multi-specialized role agent, intelligent verification is implemented based on a template matching mechanism, and the development efficiency is improved. And a complete automatic closed loop from demand to delivery is constructed. According to the scheme, full-process automation is realized, manual intervention is greatly reduced, and the development efficiency is remarkably improved; through division and cooperation of multiple agents, the professionality and decision-making quality of each link are ensured, and the capability limitation of a single agent is overcome; flexible and reliable process control is provided based on state machine management, and development state changes are dynamically adapted; and the standard consistency of the delivery result is ensured by combining template verification. According to the method, the problems of long development period, large quality fluctuation and the like caused by chain splitting, low intelligent level and excessive dependence on manpower of an existing development tool are effectively solved, and a feasible path is provided for comprehensive intelligence of software development.

The invention discloses a heterogeneous computing power resource dynamic scheduling system and method based on multi-objective optimization, and the system is deployed in a digital ecological platform complex based on multivariate consensus and embedded intelligent management. Comprising a platform access module, a data acquisition and perception module, a multi-target modeling and optimization module, a dynamic scheduling and execution module and a fault-tolerant mechanism module. The system registers as a computing power scheduling service node through an intelligent contract interface, obtains and verifies the compliance of a computing task, collects heterogeneous computing power resource node state data, and generates a scheduling decision by dynamically adjusting the weight of a target function; evaluating task suitability by using an AMCU suitability scoring model, and executing a scheduling AMCU strategy; according to the method, the technical problems of low resource scheduling efficiency and poor fault-tolerant capability in a heterogeneous computing power environment are solved, and the system throughput and the resource utilization rate are improved.

The invention relates to the technical field of computers, discloses a heterogeneous computing low-delay communication method and system, and aims to solve the problem of high delay caused by high communication protocol overhead, lack of dynamic scheduling collaboration, memory migration redundancy and non-uniform cross-node communication abstraction in existing heterogeneous computing. The method comprises the following steps: receiving a task scheduling request and analyzing a task dependency graph; tasks are dynamically allocated based on node loads and link states; rDMA, NVLink or PCIe straight-through protocols are adaptively selected according to node types to establish communication channels; zero-copy data exchange is realized through a shared memory mapping buffer area; hardware timestamps are utilized to synchronize feedback delays with PTP to optimize scheduling. The system comprises a heterogeneous computing node cluster, a unified communication scheduling controller, a low-delay communication protocol stack, a shared memory mapping buffer area and a communication delay sensing task distributor. According to the scheme, the communication delay is remarkably reduced, and the throughput and the task execution efficiency are improved.

The invention provides an automatic data exception root cause positioning and playback repair method and system and related equipment. The method comprises the following steps: constructing a dynamic heterogeneous topological graph containing data logic nodes and physical resource nodes in real time, and establishing real-time directed dependency connection between the nodes; in response to a monitored abnormal signal, generating an anti-fact intervention set assuming that an upstream node is in a reference state by using a Do operator based on the dynamic heterogeneous topological graph; the anti-fact intervention set is substituted into a causal reasoning model for simulation, and the abnormal maintenance probability of the abnormal signal still existing under the condition that the anti-fact intervention set takes effect is obtained; and calculating a causal contribution degree according to the exception maintenance probability to lock a root cause node, and transmitting state data of the root cause node to a repair module to trigger playback repair. According to the invention, the accuracy and stability of abnormal root cause positioning are improved.

Methods, systems, and computer-readable media for ranking large language models (LLMs). Input including list of LLMs, list of hardware and artificial intelligence (AI) prompt are provided by the user for ranking the LLMs. Based on the input, first estimating minimum number of hardware units needed to process AI prompt on each LLM / hardware combination and second estimating time to process the AI prompt using each LLM / hardware combination. Based on minimum number of hardware units and time to process AI prompt, third estimating amount of energy consumed by each LLM / hardware combination. Based on energy consumed, ranking LLM / hardware combinations for AI prompt. Based on ranking, selecting LLM and hardware, submitting AI prompt to LLM on hardware, and receiving response to submitted AI prompt from LLM.

The embodiment of the invention provides an information processing method and device, and the method comprises the steps: obtaining an application exception index of a target application; extracting application exception information corresponding to the target application according to the application exception index, performing information processing on the application exception information in at least one exception analysis dimension, and determining an application exception type of the target application; creating an application repair strategy for the application exception type, and determining application code information corresponding to the target application; and inputting the application repair strategy and the application code information into a code repair model for processing to obtain an application repair code, and performing exception repair on the target application by using the application repair code.

The invention relates to the technical field of computer processing, in particular to a Kafka cluster configuration parameter adjusting method and device, equipment and a medium, and is used for relieving problems caused by a fixed change means of code modification. The method comprises the following steps: according to related historical data of a Kafka cluster, predicting a first predicted load index in a future time period through a first time sequence model, and predicting a second predicted load index in the future time period through a second time sequence model; weighting the first predicted load indicator; weighting the second predicted load index; determining a predicted load index according to the weighted first predicted load index and the weighted second predicted load index; determining a plurality of first candidate adjustment actions according to the predicted load index; selecting a first candidate adjustment action from the plurality of first candidate adjustment actions as a first adjustment action; and adjusting a partition allocation index and a copy allocation index of the Kafka cluster configuration parameters by using the first adjustment action.

The invention relates to the technical field of computers, and provides an automatic data acquisition and fault processing method based on a Kylin operating system, which comprises the following steps: a Kylin-Agent acquires full-dimensional observation data in real time from a monitoring component deployed in the Kylin operating system in a data direct acquisition mode; based on the full-dimension observation data, the Kylin-Agent carries out fault detection to identify an anomaly or a fault; the Kylin-Agent is combined with a Kylin exclusive knowledge base and the reasoning ability of a large language model to carry out root cause analysis on abnormity or faults so as to locate fault causes; based on a fault reason, the Kylin-Agent retrieves related solution knowledge from the Kylin exclusive knowledge base, and utilizes the related solution knowledge based on the specific type and severity of the fault; the Kylin-Agent calls a predefined self-defined tool set, carries out security verification on the specific operation instruction, and executes a fault repair scheme after the verification is passed; and the Kylin-Agent captures and feeds back an execution state and a result in real time, and generates a fault processing report based on a complete execution process.

Systems and methods for generating predicted end-to-end cyber-security attack characteristics via bifurcated machine learning-based processing of multi-modal data are disclosed. The system accesses multi-modal data indicating a set of security information related to a computing system. The system then generates a set of extracted characteristics indicating a cyber-security attack on the computing system, via a supervised machine learning model, using the multi-modal data. Using this information, the system generates a revised set of extracted characteristics indicating the cyber-security attack, via an unsupervised machine learning model, using the extracted set of characteristics indicating the cyber-security attack, where the revised set of characteristics includes at least one new characteristic that was not included in the extracted set of characteristics indicating the cyber-security attack on the computing system. The system then generates for display a graphical representation of the revised set of extracted characteristics.

The invention discloses an AGC instruction anti-error method and system based on digital twinning and multi-risk look-ahead, and belongs to the technical field of automation and control of power systems. The method comprises the following steps: constructing a digital twinborn body which operates synchronously with a physical power system AGC control loop; acquiring operation state data; receiving an original control instruction from a superior AGC system; performing look-ahead simulation on the execution effect of the original control instruction to obtain a system state track prediction value; calculating five-dimension risk indexes by using a risk quantitative evaluator; judging whether any risk index of the five-dimension risk indexes exceeds a preset threshold value or not; if yes, triggering the instruction safety correction module, and executing the next step; if not, directly issuing the original control instruction to a station physical execution unit; an instruction safety correction module is started, rolling optimization is carried out on the original control instruction, and a safety instruction is generated; and issuing the security instruction to a physical execution unit.

A computing system and method are disclosed for converting network configurations across different platforms using a language model. This system preprocesses a source network configuration from a first network platform, generates a prompt based on the preprocessed source network configuration and additional network data for input to the language model. In response to the prompt, the language model converts the source network configuration into a target network configuration for a second platform, ensuring that common configuration elements are preserved while unsupported protocols are adapted to supported alternatives on the second network platform. The system outputs the converted configuration, facilitating seamless network migration with minimal, or no, manual intervention. Additional features include the ability to receive and process active device states and network diagrams, suggesting alternative solutions for unsupported protocols, validating the target configuration against the second platform's specifications, and interacting with users to refine the conversion based on their input.

The embodiment of the invention provides a process interception method and device and electronic equipment. The method comprises the following steps: a kernel space receives a process protection list sent by a user space by utilizing a virtual file system interface; wherein the process protection list comprises a process identifier of at least one to-be-protected process; the kernel space obtains a process termination request; wherein the process termination request is used for requesting to terminate a target process, and the target process is a process entity scheduled by an operating system; under the condition that the kernel space determines that the request type of the process termination request is a preset request, determining a matching relationship between the target process and a process protection list based on a target process identifier of the target process; and when the matching relationship indicates that the process identifier of the at least one to-be-protected process in the process protection list comprises the target process identifier, the kernel space intercepts the process termination request. By the adoption of the method, the important process can be effectively prevented from being terminated due to malicious or misoperation, and the safety and stability of the system are improved.

The invention discloses a task processing method and device, electronic equipment and a readable storage medium. The method comprises the steps that a task query request of a client side is obtained, and the task query request is used for representing a to-be-queried task of the client side; based on the task query request, identifying a task category to which the to-be-queried task belongs; an intelligent agent corresponding to the task category is called, a multi-round dialogue mode is triggered, query information of the to-be-queried task is obtained, and the query information is used for representing the query requirement of the to-be-queried task; querying the query information by using the intelligent agent to obtain a task query result; and sending the task query result to the client. The technical problem that the task processing effect is poor is solved.

According to the large file uploading method and device, the electronic equipment and the storage medium provided by the invention, the operation environment parameters of the target client are collected, and the file uploading strategy containing the data block size, the concurrent transmission quantity and the retry mechanism is dynamically generated based on the parameters; meanwhile, according to the strategy, the file to be uploaded is divided into a plurality of data blocks and integrity verification is executed, the uploading state of each data block is persistently stored in the uploading process, and uploading of the uncompleted data blocks is recovered based on the uploading state after transmission interruption is detected. The problems that due to the fact that a fixed fragment size and a static concurrent strategy cannot adapt to different device performance and network environments, the uploading state is lost after page refreshing or browser restarting is caused by depending on browser cache, the verification efficiency is low, and uploading fails, resource utilization is not good, and uploading starting is delayed can be solved. The technical effects of improving the environmental adaptability, the resource utilization efficiency and the uploading success rate of Web cloud disk large file uploading are achieved.

The invention relates to the technical field of software development, in particular to a code change influence recognition and automatic test execution method and system.The method comprises the steps that multiple types of nodes are extracted from code data, relation edges among the nodes are constructed according to code dependency relations, weights are set, and a semantic map is obtained; when the code is changed, analyzing by using a static analysis strategy, a dynamic analysis strategy and a semantic understanding strategy to obtain a change entry node; calculating the affected probability of each node according to each entry node, the semantic map and a preset propagation threshold; and obtaining and executing a test case related to each influenced node to obtain a test result. It can be understood that according to the technical scheme, three analysis strategies are combined, the direct influence and indirect influence of code change are comprehensively captured, then the test cases can completely cover the influence nodes, and the test efficiency is high.

The invention discloses a dynamic self-adaptive precision switching method for a voice recognition NPU, and aims to solve the core problems that precision and energy efficiency of an end-side voice recognition NPU in multiple scenes cannot be considered at the same time, fixed-precision hardware resources are seriously wasted, software switching efficiency is low, and errors are accumulated. A hardware architecture of a 4 * 4 reconfigurable MAC subunit array and a switch matrix is constructed, and a task-driven adaptive scheduling mechanism, an operation period synchronous control technology (single-period hardware reconfiguration is realized by utilizing idle gap latch configuration) and a direction-sensitive error compensation algorithm (historical data interpolation extension or correction truncation) are combined; hardware-level dynamic lossless switching of the MAC unit between 8bit / 16bit / 32bit precision is realized, so that different voice tasks achieve the highest necessary computing power with the lowest necessary power consumption, no data is lost in the switching process, and finally, the energy efficiency rigidity of a fixed-precision framework is oppositely converted into the precision on-demand supply capability capable of being deployed in an engineering manner.

The invention relates to the technical field of confidential computing, and discloses a data credibility-oriented vehicle-mounted integrated security computing system and credibility construction method, which comprises the following steps of: establishing a static trust chain by using a hardware trust root, shielding external interruption in a credible execution environment, and controlling a performance monitoring unit; synchronously collecting real-time hardware micro-architecture events of the key business algorithm to generate a runtime feature vector; according to the method, the instruction stream during operation is anchored by utilizing the physical characteristics of the micro-architecture, the hijacking attack of the control stream is accurately identified in a ciphertext environment, the hijacking attack of the control stream is accurately identified, the hijacking attack of the control stream is accurately identified, the hijacking attack is accurately identified, and the hijacking attack of the control stream is accurately identified. Strong coupling verification of a calculation result and an execution behavior is realized, and logic safety of a vehicle-mounted platform is ensured.

The invention provides a malicious program detection method and computer equipment, and relates to the field of computer security. Before the process runs, the computer equipment judges whether the process is triggered by the shortcut or not, and if the process is triggered by the shortcut, whether the process is a malicious process or not is judged according to the process information. The process information comprises dynamic behavior information, and the dynamic behavior information comprises information generated in a creating stage and / or a starting stage of the process. Malicious detection is carried out by using the information dynamically extracted in the creation stage and / or the starting stage of the process, so that an attacker has no opportunity to modify or mix the information, and bypassing of detection can be avoided. The dynamic behavior information is generated in the creating stage and / or the starting stage of the process, so that malicious detection on the process can be completed before the process runs, that is, malicious detection is performed on the shortcut before the shortcut really runs, and the malicious shortcut can be prevented from executing malicious behaviors in time.

A system and method for ingesting delayed logs for cybersecurity detection is presented. The method includes detecting a resource deployed in a computing environment, the resource configured to generate a local log on a disk of the resource; periodically fetching the local log from the disk, the local log including a plurality of event records written by at least a software application executed on the resource; applying a control on the fetched local log; and detecting a cybersecurity issue on the resource based on a result of applying the control.

The invention discloses a multi-language PLC configuration software component library expansion method and system based on LLVM, and relates to the technical field of configuration software, the method comprises the following steps: obtaining language source codes of various types of programming languages supporting the LLVM front end; inputting the language source code of each type of programming language into a corresponding LLVM front-end compiler, and converting the language source code into a standard intermediate representation; establishing a general language characteristic analysis framework; establishing a general abstract metadata injection framework; building a mapping rule framework; integrating the general language framework, the general abstract metadata injection framework, the mapping rule framework and the output result, and generating a unified extended intermediate representation; and generating an interface component library conforming to the IEC standard. According to the method, any programming language with an LLVM front end is supported, a developer can directly utilize a mature algorithm library in each language ecosystem to select the programming language most suitable for a specific application scene, and the development period is remarkably shortened.

The invention belongs to the technical field of cache prefetching optimization of a firmware layer of a computer storage system, and particularly provides a memory control method, a storage device, a medium and equipment.The instantaneous flow is effectively filtered and persistent hotspot data is accurately recognized through a three-layer structure of combining a first-stage hotspot filtering module with a window cache, a view queue and a protection queue; the second-stage continuous access filtering module dynamically judges a continuous access mode by means of mean-variance statistics of an annular buffer area, and accidental continuous triggering of prefetching is avoided; and the third-stage random access judgment module prohibits invalid prefetching and effectively distinguishes access modes. The pre-fetching bandwidth is limited through the token bucket algorithm, the global priority queue of the skip list is combined, it is ensured that the priority of the write task is higher than that of the pre-fetching task, pre-fetching and foreground I / O resources are isolated, delay jitter control over foreground key I / O can be achieved during full-load pre-fetching, and the problem that delay soaring is caused when pre-fetching preempts the resources is solved.