6230 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.

The invention discloses a computing power scheduling method and system based on dynamic load prediction and resource priority ranking. The computing power scheduling method comprises the following steps: collecting historical load data, task submission data and resource state data of each node in a computing power cluster; on the basis of the preprocessed multi-dimensional load feature data set, constructing an improved hybrid prediction model, optimizing model parameters through training, and predicting the load change trend of each computing power node in a future preset time period by using the trained model to obtain a node load prediction result; extracting a service level protocol parameter, a resource demand type and historical execution efficiency data of a to-be-scheduled task, and establishing a multi-dimensional resource priority evaluation index system; according to the computing power scheduling method, the problems of low resource utilization rate and high task response delay caused by low load prediction precision and mismatching of resource allocation and task priority in a traditional computing power scheduling method are solved, and the overall operation efficiency and service quality of a computing power cluster are improved.

The invention relates to the technical field of hard disk testing and verification, in particular to a multi-thread-based solid state disk reliability automatic testing method and system.The method comprises the steps that firstly, SMART information is deeply analyzed through microsecond-level high-granularity continuous performance monitoring, and multi-thread parallel processing is assisted; according to the method, fine performance fluctuation of the solid state disk under the concurrent load can be quickly captured, a fault mode can be identified, then early warning is realized by utilizing the extracted multi-dimensional features and a machine learning model, and a detailed fault diagnosis report is generated; and through dynamic error correction code strength verification and data integrity verification under pressure, an internal error correction mechanism of the solid state disk is actively detected and optimized. And finally, in combination with prediction reliability modeling, the system can estimate the remaining service life and predict faults, and provides product optimization suggestions for design, manufacturing and firmware optimization of the solid state disk, so that automation, intelligence and full life cycle management of the fault detection reliability of the solid state disk are realized.

Systems and methods for detecting and mitigating prompt injection attacks on a generative LLM are disclosed. A deployment scenario is considered, in which the generative LLM supports a task automation function. Prompts are received and interpreted by the generative LLM, and outputs from the generative LLM are used to trigger automation actions. The prompts are constructed based on a combination of user input and external data and are, therefore, vulnerable to prompt injection attacks though manipulation of the external data. To mitigate this risk, a separate discriminative classification, decoupled from the generative LLM, engine is configured to identify malicious prompts, and filter out any malicious prompts before they reach the generative LLM.

The invention relates to an industrial control system security policy adaptive configuration method and device and a readable storage medium. According to the method, a control instruction stream, a process variable sampling value and link layer message metadata are synchronously collected, an association hypergraph fusing network semantics, control semantics and physical semantics is constructed to represent a system state, and a sampling period is divided into a plurality of time slots with fixed lengths. In each time slot, self-supervised learning is utilized to extract a topology embedded vector group, and a system operation rule is represented. A robustness envelope interval is generated by calculating a statistical distance between a current time slot and a historical baseline, and is used for driving a reinforcement learning model to generate a security policy parameter set. And executing corresponding access control, traffic shaping and integrity verification operations in the next time slot, generating receipt vector feedback model update according to an execution result, outputting a micro-policy patch with a hash signature and a timestamp, and sending the micro-policy patch to the edge security gateway to realize deployment, thereby realizing closed-loop adaptive optimization of the security policy in the dynamic environment.

Methods, systems, and non-transitory computer readable storage media are disclosed for controlling access to artificial intelligence systems based on determined intent of queries. The disclosed system utilizes one or more digital content analysis models to determine an intent of one or more queries to an artificial intelligence system. The disclosed system utilizes the one or more digital content analysis models to determine an intended use of the artificial intelligence system. Additionally, the disclosed system determines whether the intent of the one or more queries aligns with the intended use of the artificial intelligence system by generating a similarity score and comparing the similarity score to a similarity threshold. Based on whether the intent aligns with the intended use, the disclosed system executes computing instructions to enable or block the one or more queries from being processed by the artificial intelligence system.

The present invention relates to adaptive multi-state encoding and processing in virtualized computing environments. The system includes a virtualized state selection module that dynamically transitions between binary, ternary, quaternary, and higher-order encoding states based on workload, bandwidth, security posture, and compliance requirements. A virtual encoding engine utilizes hardware-accelerated components, such as vFPGAs, vGPUs, and cTPUs, to enhance encoding throughput. A compliance-driven feedback controller continuously monitors encoding efficiency, threat levels, and adherence to mandates such as GDPR, HIPAA, and FIPS 140-3. Additional features include AI-based anomaly detection, federated model refinement, distributed ledger-backed audit trails, and quantum-resistant encoding techniques. By integrating intelligent encoding decisions with scalable compliance enforcement, the system enables high-performance, secure, and regulation-ready data processing across distributed cloud and edge environments, delivering measurable improvements in system responsiveness, data integrity, and operational trust.

The invention discloses an intelligent agent automatic arrangement method and system based on a large language model, and relates to the technical field of artificial intelligence. The method comprises the following steps: decomposing a natural language instruction of a user into a structured subtask sequence by utilizing a first large language model; based on the agent portrait library, matching and allocating agents for each sub-task to generate an initial execution plan; task execution is scheduled and monitored in real time through an event-driven architecture; when abnormity is monitored, a self-adaptive adjustment mechanism is triggered, the affected plan part is re-planned, and an updating instruction is issued. According to the invention, efficient, flexible and robust multi-agent automatic arrangement is realized.

The invention relates to the technical field of computing power scheduling, provides a distributed computing power scheduling method and system, breaks through the limitation that a traditional computing power scheduling system depends on a single index or a static strategy by constructing an integrated architecture of data perception, load prediction and intelligent scheduling, and constructs a self-adaptive and closed-loop optimized intelligent scheduling system. The scheduling system obtains multi-source data in real time and fuses the multi-source data into a perception vector to accurately describe the running state of the system, accurate pre-judgment of load changes is achieved through hierarchical load prediction, and resources can be efficiently allocated in different scenes in combination with double scheduling paths and execution feedback optimization; according to the scheduling system, the resource utilization rate of the distributed system is effectively improved, the energy consumption cost is reduced, the task execution delay is reduced, the system stability and the fault-tolerant capability are enhanced, and an innovative computing power scheduling solution is provided for a large-scale distributed computing scene.

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 provides a PCIe resource dynamic configuration method and device, and relates to the technical field of computer hardware resource management. The method comprises the following steps of: establishing communication connection with the pluggable equipment through an expansion interface, and reading equipment identification information pre-stored in a characteristic value storage module of the pluggable equipment by utilizing an I2C (Inter-Integrated Circuit) interface; identifying mainboard PCIe resource allocation parameters and the equipment identification information; displaying configurable PCIe resource parameters on a system configuration interface, and providing configuration options which can be operated by a user; according to the device identification information and a configuration option selected by a user, a pre-stored PCIe configuration mapping table is inquired, a corresponding PCIe resource configuration instruction is generated, and the PCIe configuration mapping table comprises the device identification information and the corresponding relation between the configuration option and the PCIe resource configuration instruction; and configuring PCIe resource allocation of the mainboard according to the PCIe resource allocation instruction.

This invention pertains to a system optimized for reutilizing allocated underutilized or unused allocated DRAM, comprising a first host, a second host, and a resource composer, interconnected via CXL. Both hosts run packaged computing environments (PCEs), which may be containers or virtual machines, and are equipped to handle respective processes, P1 and P2. The resource composer is tasked with receiving data related to P1's memory usage from a kernel module on the first host, identifying underutilized DRAM mapped to P1, and subsequently remapping it to P2's address space on the second host. This process involves the use of CXL.mem commands, which are then translated into appropriate CXL.cache or CXL.io commands for DRAM access based on the mapping.

The invention discloses a solid state disk management system and a data processing method, and relates to the technical field of solid state disk management. Flexible task scheduling and resource management are provided through a lightweight operating system module, and a computing task program defined by a user is supported to be dynamically loaded; edge computing or machine learning operators are efficiently executed in combination with a programmable hardware processing unit and a DMA channel of the computing acceleration engine module, the data preloading module is utilized to predict and preload data to DDR based on LBA access history, access delay is reduced, an NVMe protocol is expanded by means of the task unloading interface module, host task issuing and result returning are achieved, and the data processing efficiency is improved. And hardware-level memory protection is ensured through the security isolation module, so that the data calculation processing capacity of the solid state disk is remarkably improved, localized calculation tasks such as edge calculation and machine learning are supported, mass data transmission is effectively reduced, and bus and network loads are relieved.

The invention discloses a calculation task allocation method and device, equipment, a storage medium and a product, and relates to the technical field of artificial intelligence, and the method comprises the steps: obtaining a to-be-executed calculation task, and extracting the features of the to-be-executed task to obtain a feature label; decomposing the to-be-executed calculation task based on the feature labels to obtain sub-tasks; constructing a directed acyclic graph based on the dependency matrix of each sub-task, and performing topological sorting on each sub-task to obtain a priority and an execution sequence; according to the method, a to-be-executed calculation task is decomposed into a plurality of sub-tasks through feature labeling, a directed acyclic graph is constructed according to the dependency relationship among the sub-tasks, and the sub-tasks are distributed according to the reference execution time, the priority and the execution sequence. The priority and the execution sequence of each sub-task are determined by utilizing topological sorting, and the sub-tasks are allocated to each GPU for execution in combination with the reference execution time of each sub-task on different GPUs, so that the calculation efficiency of the calculation task is effectively improved.

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.

Disclosed herein are devices, systems, and methods for detecting, understanding, and classifying malicious actions and / or behaviors in software (e.g., malware), including hidden malicious actions. Specifically, disclosed embodiments use natural language approaches to understand malicious software and provide explanations for classification results. At least one embodiment constructs a knowledge graph that includes textual explanations from source materials (e.g., articles), collecting one or more sets of dynamic program traces from one or more instances of malware, and constructing and training a model (also referred to herein as Trace-BERT) using the one or more sets of dynamic program traces. Forced execution of sample segments of computer code can also be used to identify hidden or novel malicious actions.

The invention relates to the technical field of network operation and maintenance, in particular to an automatic operation and maintenance method and system based on artificial intelligence, and the method comprises the steps: collecting detailed log information containing a security event in real time through a security information and event management system configured at a target endpoint; performing cleaning, labeling and structured processing on the log data by utilizing a proxy artificial intelligence system to generate standardized metadata, and performing MITRE ATTamp with the standardized metadata; mapping the CK knowledge base, identifying technical features and behavior patterns of attacks, comparing enriched log data with an external threat intelligence source, analyzing TTP of a known attack group, generating a threat intelligence association report, generating a targeted response plan by using a large language model, generating an executable command sequence according to the response plan, and generating a threat intelligence association report; the proxy executor is connected with the server through a WebSocket protocol, executes a command in a POSIX shell environment, captures and returns an execution result, verifies the execution result, carries out necessary command optimization, records an optimized response to a vector database, and automatically matches a historical event and triggers a predefined response process through a vector retrieval mechanism. And continuous threat monitoring and adaptive response are realized. According to the method, the problem that a closed loop aiming at a terminal executor, data enrichment and historical event recall cannot be formed by security operation and maintenance in the prior art can be solved.

The invention discloses a quantum classical hybrid computing task dynamic scheduling system and method based on SLURM. The system comprises a quantum task feature extraction module, a dynamic priority evaluation module, a dependency analysis module, a quantum perception backfilling module and a uniform resource abstraction layer module. The method comprises the following steps: extracting quantum features of a to-be-processed task and a computing resource to form a quantum feature set and caching the quantum feature set; the feature set is obtained in real time, task priorities are output through multi-dimensional evaluation, and a real-time priority sequence is generated; based on the task type and the feature set, forming a dependency relationship between the classic task and the quantum task, and converting the dependency relationship into a dependency constraint and / or resource reservation instruction; according to the task priority, the resource reservation instruction and the real-time resource state, a future idle period is predicted, and a short-time quantum task is inserted for backfilling; computing resources are distributed according to the task priority and the dependency constraint, and the resource utilization state is fed back to the backfill and priority evaluation module in real time. According to the method, efficient scheduling of hybrid computing tasks can be realized, and the overall performance of the system is remarkably improved.

Some implementations of the disclosure provide a computer-implemented method including operations of receiving, by an orchestration agent, user input corresponding to a user question, wherein generating a response to the user question includes one of generating, editing, or refining programming code, and generating, by the orchestration agent, a prompt instructing a first sub-large language model (LLM) to perform a first task. In response to the prompt, generating, by the first sub-LLM, instructions for a second sub-LLM to perform a second task, wherein results of performing the second task by the second sub-LLM are provided to the first sub-LLM and performing, by the first sub-LLM, the first task utilizing the results of the second task generated by the second sub-LLM. An additional operation includes generating, by the orchestration agent, a GUI that displays the response to the user question, wherein the response includes or is based on the programming code.

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 a high-performance AIOT real-time data processing method and device based on Flink, Kafka and Dores, and the method comprises the steps: constructing an end-to-end hierarchical architecture, capturing multi-source database change data in real time through Flink CDC, and analyzing semi-structured data in combination with Apache NiFi, and generating a standardized data stream; the method comprises the following steps: realizing message routing based on KafkaTopic division of a service domain, ensuring the consistency of a data structure by using Schema Registry, and optimizing the transmission efficiency through a dynamic compression algorithm; an Flink real-time computing engine is adopted to implement window aggregation, asynchronous dimension table association and incremental state persistence strategies, and the resource fluctuation and state management problems in streaming computing are solved; time pre-partitioning and intelligent cold and hot layering are achieved through a Dores storage layer, the query performance is improved in combination with materialized views and multi-level indexes, and finally a full-link closed loop of collection-transmission-calculation-storage-management is formed. According to the invention, the problems that the real-time performance and the consistency are difficult to balance and the system complexity is high in the prior art are solved.

The invention relates to the technical field of computers, discloses an intelligent scheduling method and system for server cluster load balancing, and aims to solve the defects of the existing server cluster load balancing technology in response lag, non-uniform resource utilization rate, service quality guarantee, global optimization capability, fine-grained state perception and scheduling decision. The method comprises the following steps: collecting server state and request feature data, constructing a cluster state and service capability model, and predicting a load trend; and generating an optimal routing strategy by using deep reinforcement learning and multi-objective optimization, and issuing adjustment request distribution. The system comprises a data acquisition module, an application request feature acquisition module, a state sensing and modeling module, a load prediction module, an intelligent scheduling decision module and an instruction execution module. By adopting the technical scheme, the resource utilization rate can be improved, the response time can be reduced, the throughput can be improved, the system stability and elasticity can be enhanced, and the operation cost and energy consumption can be reduced.

The invention discloses a GPU computing power resource scheduling method and device based on load awareness and a medium, and relates to the technical field of computing power scheduling. The method comprises the steps that according to the video memory capacity and the number of calculation cores of the GPU, the video memory capacity is divided into a plurality of continuous fragments, the calculation cores are divided into a plurality of logic calculation groups, and sub-resource units are obtained; recording occupation states and load indexes of the sub-resource units in real time to obtain a resource pool; analyzing a job submitted by a user, and collecting an execution period of a kernel function, a video memory access mode and an instruction pipeline blocking rate when the job runs; and on the basis of a reinforcement learning algorithm, predicting a resource demand inflection point of the job according to the load fingerprint model, performing hierarchical scheduling on the sub-resource units based on virtualization isolation and multi-dimensional resource quantitative evaluation, and generating a preemptive allocation strategy of the sub-resource units. According to the method, efficient utilization and secure sharing of GPU resources are realized through collaborative design of dynamic granularity segmentation, load aware scheduling and hardware-level security isolation.

An automation testing system includes a processor and a memory storing historical data which at least comprising past test results including input parameters and testing outcomes for each test case included in the past test results. The processor is configured to identify input parameters for a first iteration of a first software application having first functionalities; execute an initial test on the first iteration to generate initial results; based on the input parameters and the initial results, collecting first historical data at least comprising first past test results for at least one second software application having second functionalities corresponding to the first functionalities; training a model employing AI or ML based on the initial results and the first historical data to generate a trained model; executing the trained model with the input parameters as input to generate a set of test cases for testing the first functionalities.

The invention discloses a server fault diagnosis method and device, a storage medium and a program product, and relates to the technical field of server fault localization, and the method comprises the steps: carrying out the preprocessing of original log data, determining a state space according to the obtained preprocessed log data, a fault classification model containing a value network and an action network is obtained through reinforcement learning in advance, the value network in the fault classification model is used for carrying out value evaluation on the state space, and the action network in the fault classification model is used for carrying out action probability distribution calculation on the state space according to a target value obtained through value evaluation; and determining the target fault type according to the calculated probability distribution result, thereby solving the technical problem that new features or anomalies cannot be effectively judged, effectively improving the quality and integrity of log data, improving the server fault diagnosis efficiency and accuracy, realizing intelligent identification of the fault type, and improving the fault diagnosis efficiency and accuracy. The method has the technical effects of good training stability and generalization ability.

The invention discloses an electronic lead seal data management method based on a block chain, relates to the technical field of internet data services, and aims to solve the problems of node data asynchronization, state updating failure and potential safety hazards of an intelligent contract in the prior art. The data consistency is ensured through a distributed clock synchronization algorithm and a version number verification mechanism; the reliability and fault tolerance of state updating are improved by adopting an improved Byzantine fault-tolerant algorithm and an asynchronous two-stage transaction submission mechanism; data integrity and traceability are guaranteed through a state machine detection mechanism and transaction log records; the security and maintainability of the smart contract are improved based on an automatic code auditing algorithm and a proxy contract mode of a graph neural network; real-time data monitoring and exception repair are realized by using an under-chain auditing mechanism and a cross-chain data consistency verification protocol; according to the invention, the data consistency, the system safety and the reliability of electronic lead seal data management are obviously improved.

The invention relates to a large model data protection method and device based on a trusted environment, equipment and a medium. In the scheme, after a server side obtains original large model data, the original large model data needs to be encrypted and signed, and then the original large model data and an encrypted information table are issued to a target application side; when the target application end calls the large model data, performing exception check according to a credible strategy; if the check is passed, performing signature verification on each encrypted data slice; if the signature verification succeeds, decrypting the encrypted slice data by using the encrypted information table to obtain original large model data; wherein the server side and the target application side are both in a trusted environment. It can be seen that the trusted environment and the key system are combined, and a dual protection mechanism for large model data is achieved; moreover, before the large model data is called, the operation environment of the target application end needs to be subjected to exception check, so that an attacker is prevented from acquiring or tampering the sensitive data, and all-around data protection is provided for the large model data.

Access to secured items in a computing system is requested instead of being persistent. Access requests may be granted on a just-in-time basis. Anomalous access requests are detected using machine learning models based on historic patterns. Models utilizing conditional probability or collaborative filtering also facilitate the creation of human-understandable explanations of threat assessments. Individual machine learning models are based on historic data of users, peers, cohorts, services, or resources. Models may be weighted, and then aggregated in a subsystem to produce an access request risk score. Scoring principles and conditions utilized in the scoring subsystem may include probabilities, distribution entropies, and data item counts. A feedback loop allows incremental refinement of the subsystem. Anomalous requests that would be automatically approved under a policy may instead face human review, and low threat requests that would have been delayed by human review may instead be approved automatically.

According to the operator code generation method and system based on large model driving and a multi-agent cooperation mechanism, automatic high-performance code generation is achieved through cooperative work of four kinds of agents including strategy recognition, operator generation, compilation testing and function testing. A strategy identification agent accesses a high performance optimization knowledge base (HPOK) by using retrieval enhancement generation (RAG), analyzes operator characteristics by means of a hybrid expert model (MoE), and generates an optimization strategy protocol oriented to a specific optimization target. And the operator generation agent calls a large language model (LLM) to generate an initial code, and iteratively corrects the code according to feedback iteration of compiling and function testing. And the compilation test and function test agents execute code compilation verification and function correctness verification respectively to form a closed-loop error correction mechanism. According to the scheme, through a modularized cooperation framework and dynamic knowledge retrieval, the operator code generation efficiency and quality oriented to diversified hardware platforms such as a CPU and a GPU are remarkably improved.

The invention is suitable for the field of data processing, and provides a task execution engine generation method and system, electronic equipment and a program product.The method comprises the steps that a user task is analyzed, and at least one subtask, target function requirements corresponding to the subtasks and execution sequence rules among the subtasks are obtained; determining a tool existence state of each target tool for executing each sub-task based on the target function requirement corresponding to each sub-task; based on the tool existence state, obtaining the target tool meeting the target function requirement from a locally created model context protocol tool library, and / or generating the target tool meeting the target function requirement by using a locally deployed tool generation agent; and arranging the target tool according to the execution sequence rule, and generating an agent-based task execution engine. According to the scheme, the task execution engine based on the intelligent agent can be generated in a privatized deployment scene.