Accurate and efficient prediction of the mobility of autonomous ground vehicle (AGV) in soil terrain environments remains challenging, primarily due to the inherent trade-off between physical fidelity and computational cost in existing approaches. To address this challenge, we propose a novel physics-informed neural network enhanced multiscale terrain modeling approach. This approach achieves soil macro-micro mechanical coupling through homogenization principles. The innovation of this study is the introduction of a physics constrained long short term memory architecture, which embeds soil constitutive equations as hard constraints and overcomes the physical inconsistency of purely data driven models through particle swarm optimization for adaptive loss weighting optimization. Experimental results demonstrate this approach achieves high precision predictions in critical mobility metrics. Compared to the traditional approaches, it enhances computational efficiency by a factor of 6.8 and reduces data dependency by over 40% compared to artificial neural network baselines. Additionally, cross-terrain validation confirms robust generalizability, providing a reliable foundation for subsequent navigation and control strategies of AGV operating in off-road environments.

This study proposes a novel synchronization framework for memristive chaotic systems (MCSs) through an enhanced deep reinforcement learning (DRL) approach, featuring an improved proximal policy optimization (PPO) algorithm. Distinguished from traditional linear/nonlinear control paradigms that necessitate precise mathematical modeling, our DRL-based methodology operates without prior knowledge of system dynamics or analytical model requirements. The developed data-driven control strategy demonstrates significant advantages by reducing the required control forces from four to three dimensions, thereby substantially decreasing control complexity and operational costs compared to conventional item-by-item control methods. Through systematic optimization of the reward function architecture in classical PPO algorithms, we achieve accelerated synchronization convergence rates for MCSs, in which an optimal exponential parameter is obtained accordingly. Finally, the practical efficacy of our DRL-driven synchronization framework is successfully validated in image encryption applications. Comprehensive numerical simulations and comparative analyses demonstrate that the proposed methodology not only maintains robust performance under Gaussian noise perturbations but also achieves synchronization efficiency improvements.

The sensor coverage problem is a typical combinatorial optimization problem, which is favored by researchers due to its wide practical applications. This article studies the sensor coverage problem, where each sensor has communication, sensing, and computing capabilities to collaboratively cover a certain area. First, we treat each sensor as a player, and establish a potential game for the sensor coverage problem, where the global objective is presented by the potential objective, and individual's utility is designed as a wonderful life utility. Second, we prove that under the established potential game, a Nash equilibrium can guarantee at least 50% optimality. Third, we design a synchronous game learning (SGL) distributed algorithm, where each player has a memory length$m$. Fourth, we prove that our designed SGL algorithm can guarantee that strategies of all sensors converge to a Nash equilibrium, and analyze its complexity. Finally, we demonstrate the effectiveness and superiority of our designed SGL algorithm by comparing with the existing representative optimization algorithms via numerical simulations. In addition, we also find that by employing our designed SGL algorithm, a tradeoff between solution quality and runtime could be achieved via adjusting players' memory length.

Large language models (LLMs) have been applied across various domains due to their superior natural language processing and generation capabilities. Nonetheless, LLMs occasionally generate content that contradicts real-world facts, known as hallucinations, posing significant challenges for real-world applications. To enhance the reliability of LLMs, it is imperative to detect hallucinations within LLM generations. Approaches that retrieve external knowledge or inspect the internal states of the model are frequently used to detect hallucinations; however, this requires either white-box access to the LLM or reliable expert knowledge resources, raising a high barrier for end-users. To address these challenges, we propose a black-box zero-resource approach for detecting LLM hallucinations, which primarily leverages multi-perspective consistency checking. The proposed approach mitigates the LLM overconfidence phenomenon by integrating multi-perspective consistency scores from both queries and responses. In comparison to the single-perspective detection approach, our proposed approach demonstrates superior performance in detecting hallucinations across multiple datasets and LLMs. Notably, in one experiment, where the hallucination rate reaches 94.7%, our approach improves the balanced accuracy (B-ACC) by 2.3 percentage points compared with the single consistency approach and achieves an area under the curve (AUC) of 0.832, all without depending on any external resources.

With the increasing number of vulnerabilities exposed on the Internet, autonomous penetration testing (pentesting) has emerged as a promising research area. Reinforcement learning (RL) is a natural fit for studying this topic. However, two key challenges limit the applicability of RL-based autonomous pentesting in real-world scenarios: the training environment dilemma—training agents in simulated environments is sample-efficient while ensuring that their realism remains challenging; poor generalization ability—agents’ policies often perform poorly when transferred to unseen scenarios, with even slight changes potentially causing a significant generalization gap. To address both challenges, we propose a generalizable autonomous pentesting framework termed GAP, which aims to achieve efficient policy training in realistic environments and train generalizable agents capable of drawing inferences about other cases from one instance. GAP introduces a real-to-sim-to-real pipeline that enables end-to-end policy learning in unknown real environments while constructing realistic simulations and improves agents’ generalization ability by leveraging domain randomization and meta-RL learning. We are among the first to apply domain randomization in autonomous pentesting and propose a large language model-powered domain randomization method for synthetic environment generation. We further apply meta-RL to improve agents’ generalization ability in unseen environments by leveraging synthetic environments. Combining the two methods effectively bridges the generalization gap and improves agents’ policy adaptation performance. Simulations are conducted on various vulnerable virtual machines, with results showing that GAP can enable policy learning in various realistic environments, achieve zero-shot policy transfer in similar environments, and achieve rapid policy adaptation in dissimilar environments.

3-path vertex cover of networks is a typical optimization problem in network science, which has a wide range of applications. Toward a 3-path vertex cover of networks from distributed optimization, we first established a potential game to describe the 3-path vertex cover problem. Next, we analyze the inherent relationship between potential game and 3-path vertex cover, that is, only the solution to minimum value of potential function are minimum 3-path vertex covered solutions, and strict Nash equilibriums are intermediate solutions between minimum 3-path vertex covered solutions and 3-path vertex covered solutions. Then, we propose a bounded best response and memory-based distributed algorithm, and prove that our proposed algorithm can guarantee any initial solution converge to a strict Nash equilibrium, and further analyze the complexity of this algorithm. Finally, numerical simulations verify the effectiveness and superiority of our proposed algorithm on some representative networks and benchmark by comparing with existing representative algorithms. This work paves an effective way for distributed optimization that could be modeled as distributed potential game.

Vertex cover problem is a typical nondeterministic polynomial combinatorial optimization problem with wide applications. In order to seek a near-optimal solution in finite time, this article achieves the goal from the perspective of serial potential game optimization. Specifically, we present the vertex cover problem as a potential game, where the corresponding potential function minimizers are equivalent to the minimum vertex cover state. Then, we propose a novel polynomial-based serial game distributed algorithm, and prove that the algorithm can guarantee that strategies of all vertices converge to a$\epsilon$-minimum vertex cover ($\epsilon$-MVC) state in finite time. Compared with the existing representative optimization algorithms on networks and standard benchmarks, numerical simulations demonstrate our proposed algorithm can effectively balance solution efficiency and computation time. We hope this work provides insight for decision-makers designing a reasonable algorithm when solving nondeterministic polynomial optimization problems, so as to guarantee both solution quality and computation time.Note to Practitioners—Vertex cover problem has a wide range of practical applications. Many distributed optimization algorithms for the vertex cover problem are available in the existing literatures. However, those algorithms mainly pursue high-quality solutions in infinite time. Thus, one of the main challenges is to design a distributed algorithm that guarantees both solution quality and computation time. This article design a polynomial-based serial game algorithm, and prove that it can converge to the$\epsilon$-MVC state in finite time. This article can serve as a supplement to the existing optimization algorithms for solving the vertex cover problem.

This article investigates the global predefined-time consensus (PTC) of multiagent system (MAS) via constructing a duplex communication network. Unlike the traditional finite-/fixed-time convergence, our method allows the upper-bound of settling-time to be an explicit constant, which is tunable and can be set beforehand without relating with the network information, controlling parameters, and initial conditions. In particular, our approach uses a smooth, nonchattering consensus scheme that avoids conventional discontinuous functions like signum and absolute value functions. By the Lyapunov stability analysis, the sufficient criterion is deduced for ensuring the PTC of MAS. Finally, simulations confirm the effectiveness of our proposed nonchattering scheme.

As a well-known NP-hard problem, the vertex cover problem has broad applications, which has aroused the concern of many researchers. In recent years, its related optimization problems, including the weighted vertex cover problem, the $\ell \geq 3$ path vertex cover problem, and the connected vertex cover problem, and other related optimization problems have came into the view of researchers, who have designed various optimization algorithms to solve those related optimization problems. First, based on the existing works, we give detailed descriptions of the vertex cover problem and its related optimization problems and then review the current research progress. Then, we present some main representative optimization algorithms and provide numerical results and corresponding analysis. Finally, we summarize the existing works and present the future research directions.

As a typical combinatorial optimization problem, the 3-path vertex cover problem has wide applications in practice. To solve the 3-path vertex cover problem from the perspective of distributed optimization, we treat each vertex as an agent (i.e., player) with computation, and decision-making capabilities. First, we establish a 3-player symmetric game model to describe the 3-path vertex cover problem, and design the corresponding cost function for each player. Then, we prove that under the established game model, strict Nash equilibriums (SNEs) act as the basis of the connection between 3-path vertex cover states and minimum 3-path vertex cover states. Next, we propose a novel memory-based synchronous learning (MSL) algorithm, where the initial profile strategy generation of players relies on the designed degree preference rule, and each player has a memory length for recording strategies and independently update their strategies concurrently based on the accessed local information. After that, we prove that our proposed MSL algorithm can guarantee that any strategy profile converges to an SNE, and provide a theoretical analysis of the algorithm’s complexity. Finally, we present numerous numerical simulations to demonstrate the performance of our proposed algorithm on various networks. Moreover, we find that increasing the memory length and adopting the degree preference initialization can yield a better SNE.

Multiagent dynamic task assignment of forest fires is a complicated optimization problem because it requires the consideration of multiple factors, such as the spread speed of fires, firefighting speed of agents, the movement speed of agents, and the number of deployed agents. In this article, we investigate multiagent dynamic task assignment based on a forest fire point model, the objective of which is to minimize task completion time. First, we establish a model for the spread of fire and dynamic task assignments. Second, we prove that the optimal static task assignment always makes all task completion times the same under certain assumptions. Furthermore, we calculate the optimal solution to the static task assignment problem assuming no travel time for the agents, which provides the theoretical basis for the initial deployment and dynamic deployment. Third, we propose a dynamic task assignment scheme based on the global information, which ensures that every reassignment reduces the task completion time and makes all task completion times close to each other. Finally, the simulation is carried out on the MATLAB platform to verify the performance of the proposed dynamic task assignment scheme by comparing with a multistage global auction algorithm. We hope that this work provides insight for decision-makers designing reasonable assignment strategies based on the model and solving assignment optimization problem in different situations.Note to practitioners—The forest firefighting problem considered in this article is a typical multitask and multistage optimization problem. Many searching algorithms for multistage optimization problem are available in the existing literature. However, one of the main challenges is that the time of searching increases exponentially with the number of stages. This work first proves that the tasks are completed in the minimum amount of time, under the constraint of one-shot assignment. This finding helps us to evaluate the gap between the searching algorithm and the optimal solution. In addition, in practice, if the underlying dynamic process can be modeled or partially modeled, then we can predict the behavior of future stages and reduce the searching domain. If a model is available, then we can also adjust the assignment scheme dynamically based on the principle that each adjustment would reduce the total time of tasks completion. In this article, we establish a dynamical fire-spreading model and propose a model-based solution to the multistage optimization problems. The findings in this work can serve as a supplement to the existing optimization algorithms.

The vertex cover of networks is a classical combinatorial optimization problem. In this paper, we investigate the vertex cover problem solved by the memory-based best response update rule, which can not converge to a strict Nash equilibrium (SNE) with memory length m =3D 1. To overcome this shortcoming, a bounded rational behavioral (BRB) update rule is newly proposed in this paper. We prove that the BRB with m =3D 1 can guarantee that the whole vertices' state converges to a SNE. The simulation is carried out to verify that the performance of the proposed BRB update rule on representative networks. Moreover, we also find that a better SNE will be achieved by increasing the selection intensity.