30 results about "Agent behavior" patented technology

The invention relates to the technical field of computer application, and particularly discloses a chemical industry park safety risk assessment method based on an agent model. The method comprises the following steps: constructing a multi-agent collaborative perception framework, and fusing multi-source heterogeneous data of equipment, environment, materials and personnel to form a park panoramic view under a unified space-time reference; establishing an agent behavior model library embedded with the chemical process mechanism and the safety interlocking logic; an intelligent agent decision strategy is optimized through deep reinforcement learning, and multi-target training is carried out by taking a safety index as a reward function; executing dynamic risk deduction and accident chain simulation in a digital twin environment, and quantifying risk probability and consequences in combination with Monte Carlo sampling; and finally, a visual thermodynamic diagram and an interpretable risk assessment report are generated, weak links are identified, and intervention suggestions are provided. According to the technical scheme, dynamic, accurate and prospective evaluation of the safety risk of the chemical industry park can be realized, and the early warning timeliness and the evaluation reliability are remarkably improved.

The invention relates to the technical field of knowledge graph construction, and discloses a knowledge graph-based agent behavior description method, which comprises the following steps of: firstly, analyzing an unstructured agent operation log data stream through a processor, and mapping the unstructured agent operation log data stream into a discretized module access sequence by utilizing a topological structure of a code knowledge graph so as to remove text redundancy; performing multi-dimensional time sequence analysis on the sequence, and calculating resident distribution data representing operation continuity, frequency domain fluctuation data representing switching rhythm and multi-scale coverage extension data representing a traversal range; and then aggregating the data into a fixed-length multi-dimensional index vector, constructing an index node associated with a session entity in a graph database, and writing the vector as a binary structured attribute into a storage field. According to the method, massive unstructured logs are converted into compact graph structured indexes, and quick positioning and direct retrieval of a complex agent operation mode are supported while the storage space is remarkably saved.

The invention discloses an agent training method, a data processing method and a related device, and relates to the field of artificial intelligence. In an agent training process, key information in target interaction data generated by interaction between an initial agent and a target user is automatically extracted to generate a structured memory; and a context basis for supporting training can be constructed without manually annotating data, so that the training cost of the intelligent agent is reduced. Traceable historical context support is provided for the intelligent agent by processing the target interaction data and updating the target database based on the processing result; a comprehensive basis is provided for agent behavior evaluation by acquiring whole-process trajectory data generated when an initial agent executes a task in a target task environment; the scoring model constructed by the user for the feedback information of the full interaction trajectory scores the trajectory data, and the initial agent parameters are updated, so that the agent can continuously optimize the strategy, and the accuracy of the agent is effectively improved.

The invention relates to the field of artificial intelligence, the technical scheme can be applied to the field of industrial automation, the field of intelligent transportation and the field of intelligent power grids, and discloses an abnormal behavior recognition method, device and equipment of a multi-agent system and a medium. Meanwhile, the abnormal behavior score is calculated in parallel by the abnormal detection module; and after abnormity is judged through a dynamic threshold value, a closed-loop correction mechanism is triggered, and a reward function and an experience sampling weight of a strategy module are adjusted in real time, so that online compensation and optimization are performed on the behaviors of the intelligent agent. According to the method, the anomaly recognition accuracy, the real-time response capability and the overall system robustness of the multi-agent system in a complex dynamic environment are remarkably improved.

The application relates to the technical field of agent testing, and particularly provides an agent dynamic behavior safety testing method and system based on a risk transmission quantitative model, which comprises the following steps: collecting agent behavior data, containing a plurality of risk operation information and risk correlation relations; constructing a behavior risk correlation graph taking risk operations as nodes and correlation relations as edges, marking risk attributes and time limit factors on the nodes, and marking risk transmission strength on the edges; constructing a risk rule library based on the risk marking results and a preset risk judgment logic; inputting the behavior data, the behavior risk correlation graph and the rule library into a trained quantitative model to calculate risk transmission probability, influence range and critical nodes; real-time monitoring of behavior trajectories, task suspension, backtracking reasoning chain and intervention record generation when a multi-dimensional risk threshold is triggered; and test report generation by fusing multi-dimensional data. The application solves the problems of missed judgment and response lag in traditional testing, and improves the pertinence and reliability of safety testing.

The invention provides an output lexical element prediction method oriented to a multi-agent large language model application. Comprising a multi-agent behavior acquisition module for executing light intrusion behavior acquisition, a feature construction module for executing feature construction, a single-agent high-quantile output length prediction module for executing single-agent high-quantile output length prediction, a model online self-adaptive adjustment and optimization module for executing online self-adaptive updating, and a model online self-adaptive adjustment and optimization module for executing online self-adaptive updating. The multi-agent execution portrait construction module is used for executing the application-level execution portrait; a two-stage hybrid prediction architecture (intention classification + quantile regression) adopted by the method can effectively adapt to behavior differences of different roles (such as translation, writing and reasoning) in a multi-agent system. Meanwhile, the prediction overhead is extremely low, and the low-delay requirement of a real-time scheduling system is met.

This application provides a method, system, device, and storage medium for dynamic evaluation of agent behavior consistency, belonging to the field of agent technology. The method includes: testing the agent using a multi-turn dialogue state machine based on an adversarial evaluation dataset to obtain interaction behavior data; scoring the interaction behavior data using multiple reviewer models to obtain multi-source reviewer scoring data; calculating the scoring divergence degree based on the multi-source reviewer scoring data; if the scoring divergence degree is greater than an empirical threshold, obtaining the structured evidence chain output by each reviewer model; obtaining the divergence logic attribution result based on each structured evidence chain through a meta-reviewer model; adjusting the weights of each reviewer model based on the divergence logic attribution result and an evaluation priority matrix; and weighting the multi-source reviewer scoring data based on the adjusted weights of each reviewer model to obtain a weighted comprehensive score. This application can improve the accuracy and reliability of agent evaluation results.

This invention belongs to the field of artificial intelligence and computer simulation technology, and discloses a knowledge-driven multi-agent strategic game deduction method. Existing strategic game deduction methods generally suffer from insufficient agent consistency and stability, inadequate representation of knowledge differences, and a lack of interpretability in strategy generation. The knowledge-driven multi-agent strategic game intelligent deduction method and system proposed in this invention constructs a knowledge base for each agent, sets the game environment and context for each agent, conducts multiple rounds of game deduction based on the context of each agent, the knowledge base, and the game environment, and finally outputs a complete set of game deduction states based on the state of each round of game deduction. This invention effectively solves the problems of unstable agent behavior and convergent role characteristics by embedding the agent's domain knowledge into its decision loop in a structured manner. It can generate logically consistent, distinctive, and traceable strategy sequences, improving the reliability and analytical value of strategic game deduction.

The invention relates to an agent learning training method and device, computer equipment and a storage medium, and the method comprises the steps: converting a high-level intention into an executable code through employing an LLM model, obtaining a clone data set, constructing an intelligent driving scene into a Markov decision process, and constructing an instruction space; constructing an intelligent agent strategy model; a training target is decoupled into a behavior cloning and reinforcement learning target and weighted summation is carried out, the intelligent agent is trained according to the final training target, sampling is carried out from a cloning data set, intelligent agent strategy model parameters are updated, then sampling is carried out from an environment interaction track, and the intelligent agent strategy model parameters are updated in a PPO mode. Due to the fact that a high-level intention-instruction-strategy link is closer to a decision-making level which can be understood by human beings, intelligent agent behaviors can be debugged, audited and backtracked conveniently, maintainability is improved from the perspective of engineering landing, and a clearer interface is provided for follow-up safety fence access, formalized constraint or rule verification.

The application discloses a multi-agent behavior decision-making method and system based on an interactive dynamic influence graph, and particularly relates to the technical field of artificial intelligence, and the scheme comprises the following steps: acquiring observation environment information and action information of each agent in an agent cluster, and utilizing the observation environment information and / or the action information to form a behavior sequence of a corresponding agent; performing evolution calculation on the behavior sequence to obtain a strategy tree, and controlling the behavior of each agent according to the strategy tree. According to the scheme, the evolution calculation is utilized to iteratively output a more diverse behavior model according to the mutual influence relationship of each agent in the interactive dynamic influence graph, and the obtained behavior model is embedded into the interactive dynamic influence graph model, so that more effective and rich environment information is provided for the main agent to optimize its own decision-making, and the main agent and other agents can make more optimal decisions.

A system and method dynamically manages the operational boundaries of an autonomous system or agent in a configurable and adaptable manner. Traditional approaches rely on static or fixed boundaries to constrain an agent's operation, which can be highly restrictive and inefficient. To address these issues, a virtual anchor state is established to provide a dynamic reference entity that represents a focal point based on factors such as proximity to objectives, environmental conditions, or system priorities. The agent state provides real-time information such as the agent's position, status, and behavior, while environmental data includes external factors like obstacles, nearby objects, or mission-specific variables. By combining these inputs, the system continuously updates the anchor state to reflect the most relevant context and uses it to calculate boundary conditions, which define permissible operational areas or constraints. These boundaries adapt dynamically to changing contexts, such as shifts in the environment or agent behavior.

The invention discloses an agent behavior compliance constraint method based on a multi-source data traceability chain. The method comprises the following steps: forming a multi-source data set; generating corresponding signature information; constructing a multi-source data tracing chain; agent behavior event data are received, and a behavior data association record set is formed; establishing an agent behavior compliance constraint model by using an improved CMAN algorithm; performing real-time verification on the agent behavior event data to generate a compliance verification result; and a traceable compliance audit chain is generated, so that fusion of a multi-source data traceability chain and an intelligent compliance algorithm is realized, and transparency, compliance and traceability of agent behaviors can be effectively improved.

The invention discloses an intelligent agent cooperative behavior control method and system based on a state-driven architecture. The method comprises the following steps: defining a multi-dimensional state vector model of a business object; defining a circulation rule based on a state vector; compiling the rule into an optimized discrimination network; during running, executing according to a state matching driving rule, and setting a non-bypassing multi-stage safety judgment process in an execution link; chained circulation is realized through automatic conduction of state change. According to the system, heterogeneous intelligent agent collaboration is realized by adopting a distributed architecture. According to the invention, the state is taken as a core driving force, the business logic is decoupled from a code into a configurable rule, the safe controllability of the behavior of the high-autonomy intelligent agent is ensured through a built-in safety decision, and the flexible, safe and explainable intelligent agent collaborative management is realized.

The invention discloses a BDI-based ground unmanned system multi-agent simulation behavior model construction method, and relates to the technical field of unmanned system simulation, and the method comprises the following steps: S1, constructing an internal BDI cognitive layer of an agent; s2, setting an intention-behavior converter; s3, constructing a Couzin model layer; and S4, establishing a sensing feedback loop. According to the method, the motion naturalness is remarkably improved while the task efficiency is guaranteed through a bidirectional coupling architecture, the trust and controllability of people to an unmanned cluster are enhanced by quantifying a mapping chain and establishing a transparent bridge from a human instruction to a cluster behavior, and the problem of disjunction of cognition and behaviors in a traditional model is effectively solved; according to the method, the intelligent agent cluster shows smooth and natural self-organizing motion characteristics during task execution, the established task target can be efficiently completed, the sense of reality of a simulation scene and the natural coordination of intelligent agent behaviors are remarkably improved, and a simulation result is closer to the operation performance of an actual ground unmanned cluster.

An embodiment includes non-deterministic agent state transition behavior specification, monitoring, and correction. An embodiment establishes an agent, wherein the agent is configured to output text in response to input text. The embodiment defines an agent behavior specification for the agent. The embodiment inputs a text input to the agent and monitors the text output of the agent to detect an incorrect state transition, wherein the incorrect state transition comprises a state transition that deviates from the agent behavior specification. The embodiment applies a correction to the output text to create a corrected output text upon detecting the incorrect state transition. The embodiment reverts the agent to a previous state, the previous state preceding the state corresponding to the incorrect state transition detected. The embodiment inputs the corrected output text to the agent in the previous state to cause future behavior of the agent to align with the agent behavior specification.

The present application relates to the technical field of computer and artificial intelligence, and particularly relates to a kind of Agent generation, interfacing, integration and control method, device, computer equipment, storage medium and computer program product.The certain Agent generation, interfacing, adjustment and control method described in the present application comprises the following steps: generating ActionAgent model through integrated and split type, optimizing Agent behavior based on dynamic adjustment factor relationship, reducing energy consumption by optimizing data transmission mechanism, enhancing system security by introducing multiple control points, and verifying system performance by simulating fault detection.The present application can standardize the generation and interfacing process of ActionAgent, reduce repeated operation and data transmission, improve the efficiency of multi-ActionAgent collaboration, and enhance the stability and security of the system through factor parameter value sharing and fault detection technology, and is suitable for multi-Action Agent generation, interfacing, collaboration and self-feedback continuous operation requirements in medical, household, industrial, autonomous driving and other scenarios.

The invention belongs to the technical field of artificial intelligence, and particularly relates to a monitoring system for AI Agent risk behaviors. The system comprises a data sensing and collecting module used for collecting risk behavior signals and external data sources generated in the operation process of the AI Agent in real time; the main controller is used for receiving and distributing the risk behavior signal and an external data source and carrying out aggregation and risk assessment on a feedback analysis result; the multi-agent system comprises a plurality of sub-agents with independent functions; the sub-agents are configured to receive data from the main controller and perform parallel analysis and risk assessment on behaviors of the AI Agent from a plurality of dimensions; the safety response system is used for implementing dynamic protection and emergency response to the execution environment of the AI Agent according to the risk assessment result and the response instruction issued by the main controller; and the knowledge base is used for storing security rules, threat models and agent behavior parameters, and performing dynamic updating based on system operation feedback.

The invention provides an enterprise AI agent-oriented multi-dimensional behavior modeling and risk quantification method and system, and relates to the technical field of artificial intelligence, and the method comprises the steps: firstly collecting an agent behavior data flow which is generated by the operation of an enterprise AI agent and comprises a behavior event record and a time sequence association relation, inputting the agent behavior data flow into a multi-dimensional behavior decoupling network, and carrying out the multi-dimensional behavior modeling and risk quantification; the method comprises the following steps: extracting intention-oriented and environmental response type behavior characteristics, constructing a dynamic interaction mapping relation between the intention-oriented and environmental response type behavior characteristics, determining a driving weight and feedback correction parameters, carrying out iterative fusion according to the parameters to generate a behavior evolution path characteristic sequence, and finally carrying out behavior mode clustering on the behavior evolution path characteristic sequence. And matching with a preset risk behavior feature library to generate a behavior risk quantitative evaluation result. The enterprise AI agent behavior risk can be scientifically and accurately evaluated, and safe and stable operation of the enterprise AI agent is guaranteed.

The application relates to an agent learning training method and device, computer equipment and a storage medium, the method comprising: converting a high-level intention into executable code by using an LLM model to obtain a cloned data set, constructing an intelligent driving scene as a Markov decision process and constructing an instruction space; constructing an agent policy model; decoupling a training target into the targets of behavior cloning and reinforcement learning and performing weighted summation, training the agent according to the final training target, updating the agent policy model parameters by sampling from the cloned data set, then sampling from the environment interaction track, and updating the agent policy model parameters in a PPO manner. Since the link of high-level intention-instruction-policy is closer to the decision level that can be understood by humans, the agent behavior is convenient to debug, audit and backtrack, the maintainability is improved from the engineering landing angle, and a clearer interface is provided for subsequent access to safety fences, formal constraints or rule checking.

The disclosed concepts relate to employing agent behavior models to control agent behavior in applications, such as video games or simulations. For example, in some implementations, agent behavior models with relatively greater resource utilization, such as generative language models, are assigned to agents at higher levels of an agent hierarchy. Agent behavior models with relatively less resource utilization, such as reinforcement learning or hard-coded models, are assigned to agents at lower levels of the agent hierarchy.

The application provides an agent behavior alignment detection method and system based on a thinking chain audit, first obtains multi-dimensional thinking chain information of an agent, constructs an audit query of a selected action of the agent to be audited, provides the audit query and a preset small amount of audit prompt sample examples to an audit LLM, obtains an audit result of the selected action of the agent, and generates an execution decision result of the selected action of the agent according to the audit result, the method can accurately identify target deviation hidden in a complex reasoning process, malicious behavior caused by indirect prompt injection, and subtle reasoning deviation of the agent itself, and is designed to perform real-time audit and intervention before the agent executes a key action, can timely prevent potential harmful behavior or serious target misplacement. At the same time, through an adaptive intervention decision mechanism, the interference on the normal function of the agent can be maximally reduced under the premise of ensuring safety, and safety and usability are balanced.

The invention discloses a multi-agent action prediction method of a differential attention model based on intention driving, and belongs to the technical field of multi-agent behavior prediction. The method comprises the following steps: acquiring environment information around an unmanned aerial vehicle group and a previous-step action of the unmanned aerial vehicle group as input data; inputting the input data into the short-term intention encoder module, extracting the correlation between the input data to obtain the current intention feature, inputting the current intention feature into the differential attention module, and outputting the current intention feature to the short-term intention encoder module; the current intention features are subjected to weighted aggregation through differential attention weights, perception features are obtained, and a differential attention module comprises a message attention sub-module and an environment attention sub-module; and generating a current action instruction of the unmanned aerial vehicle group based on the sensing features to guide the next action of the unmanned aerial vehicle group. The method improves the accuracy of unmanned aerial vehicle group motion prediction.

A method of maintaining behavioral integrity of an artificial intelligence (AI) agent includes capturing AI agent execution signals. The AI agent is bound to a person. Moreover, the method includes calculating similarity scores from the execution signals and an agent similarity score from the similarity scores, verifying runtime attestation of an execution environment hosting the artificial intelligence agent, and calculating an attestation conformity score from the runtime attestation. Furthermore, the method includes calculating an integrity score by combining the similarity and attestation conformity scores and evaluating policy-as-code at one or more policy checkpoints to yield a decision outcome. The policy checkpoints include discover, invoke, and runtime. The method also includes comparing the integrity score against a threshold value. In response to determining the integrity score failed to satisfy the threshold value, or the decision outcome indicates allow with modification, the method includes modifying capabilities of the artificial intelligence agent.

A method for predicting external agent behavior can include: receiving decisioning data, identifying a set of environmental agents (e.g., external agents), and evaluating a set of hypotheses for each environmental agent. A system for predicting external agent behavior can include a computing system and a sensor suite (e.g., onboard an autonomous vehicle), which can function to implement any or all of the processes of the method.

The embodiment of the invention discloses an agent behavior control method, electronic equipment, a storage medium and a product. According to the main technical scheme, the method comprises the steps of obtaining multi-modal interaction information of a user; extracting interaction data of each mode from the multi-mode interaction information, and obtaining a multi-mode interaction context according to the interaction data of each mode; based on the multi-modal interaction context, generating a reply text and a multi-modal control parameter corresponding to the reply text by using a preset large language model; in the generation process of the reply text, performing short sentence division on the reply text to obtain a short sentence text, and generating an expression unit according to the short sentence text and the multi-modal control parameter corresponding to the short sentence text; and in response to completion of generation of any expression unit, multi-modal behavior data is generated based on the short sentence text corresponding to the expression unit and the multi-modal control parameter, and a multi-modal behavior is executed according to the multi-modal behavior data, so that the real-time performance and coordination of agent behavior expression can be effectively improved.

The application discloses a kind of lightweight cross-domain recommendation method and system based on user alignment Agent drive.It first obtains the historical behavior data of user in multiple fields, fuses text, image and other multi-modal content, generates fine-grained interest prototype through cross-domain semantic encoder, and constructs personalized Agent to simulate user intent.Then, in the multi-field collaborative environment, the behavior strategy of Agent is optimized using reinforcement learning and hybrid reward mechanism, and the general preference and domain-specific preference are modeled through hierarchical strategy network, and knowledge fusion is realized through gating mechanism.Subsequently, combined with preference distillation technology, transferable representation is extracted from Agent behavior, and lightweight cross-domain knowledge graph is constructed.Finally, by using behavior trajectory compression and cross-domain preference mapping, efficient and low-consumption personalized recommendation is realized.This method effectively alleviates the problems of cross-domain data sparsity and model complexity, improves recommendation accuracy and system response efficiency, and is suitable for real-time recommendation services in multiple scenarios.

The application discloses a multi-agent reinforcement learning method and device based on skill discovery and distribution, and relates to the field of multi-agent reinforcement learning. The method can solve the problem of behavior homogenization between agents caused by parameter sharing in the prior art, and can enhance the diversity of agent behavior, so that the method can better adapt to a task scene requiring complex coordination. The method comprises the following steps: obtaining skill probabilities of each skill included in a skill set according to a parameterized neural network and observation latent variables of each agent; obtaining a total value function of an agent in a current time period according to a skill to be executed by the agent in a next time period, observation latent variables of the agent in the current time period, and a skill strategy of the agent in the current time period; and obtaining a loss function of the agent according to an intrinsic reward of the agent in the current time period, the total value function of the agent in the current time period, and a total value function of the agent in the next time period.

The application provides a multi-protocol interaction analysis method and device for an AI agent, a medium and a program product. The method comprises: loading a collection probe in the operating system kernel layer of a computing node where the AI agent runs; acquiring, by using the collection probe, multi-source data generated by the AI agent in the running process and the external system interaction, the multi-source data comprising network communication flow data and local pipeline communication flow data; performing data reorganization processing on the network communication flow data and the local pipeline communication flow data, restoring the streaming or fragmented interaction data to complete application layer protocol data; calling an analysis rule corresponding to the protocol type of the application layer protocol data to analyze and process the application layer protocol data, so as to extract interaction metadata and further construct an interaction link relationship corresponding to the AI agent. The application realizes full-link restoration of the multi-protocol and multi-path interaction behavior of the AI agent, and improves the overall observability and correlation analysis capability of the agent behavior.

The application discloses an AI intelligent agent behavior credible execution method and system based on intention verification, and belongs to the technical field of artificial intelligence security. The method comprises the following steps: receiving user input, analyzing the user input, and generating a structured high-level intention object; generating a candidate action sequence according to the high-level intention object; dynamically generating a set of time sequence logic constraint formulas in combination with the high-level intention object and a pre-constructed intention action mapping atlas; verifying whether the candidate action sequence meets the time sequence logic constraint formulas; if the verification is passed, the candidate action sequence is executed; and if the verification fails, the execution is aborted and an alarm is issued. The application solves the technical problem that, in the related art, over-authorization or harmful operation caused by prompt word injection, model hallucination and the like cannot be effectively blocked at the semantic level in real time.

A collective artificial intelligence consensus and anti-herding system detects correlated agent behavior at execution time, dynamically rebalances agent influence, enforces diversity constraints prior to execution, and generates auditable consensus artifacts, enabling stable and governable multi-agent artificial intelligence systems.