This paper presents two complementary enhancements to the Very Fast Decision Tree algorithm for data stream mining. The first contribution introduces the fractional Hoeffding bound, a relaxed splitting criterion where the original threshold is scaled by a factor . Experimental evidence shows that for this modification, the resulting trees not only grow faster but also achieve higher accuracy compared to the original Very Fast Decision tree. The second contribution proposes a novel data structure, called extended statistics, which extends the traditional sufficient statistics used in the Very Fast Decision Tree by maintaining additional information about attribute co-occurrences. This allows child nodes to inherit richer knowledge from their parent, leading to deeper trees with accelerated convergence to the target accuracy. Numerical experiments on synthetic data streams demonstrate that the combination of fractional bounds and multi-index statistics yields significant accuracy gains, particularly in the early phases of learning. The extended statistics improvement, however, comes at the cost of increased memory and computational requirements, emphasizing the trade-off between predictive performance and resource usage.

Long Chain-of-Thought (CoT) reasoning enhances large reasoning models' performance but suffers from severe inefficiencies, as models often overthink simple problems or underthink complex ones. Current sequence-level optimizations, like length penalties, are too coarse-grained to distinguish core logic from verbose language, precluding the necessary token-level control for efficient reasoning CoT. To overcome these limitations, we introduce Time-Frequency token Advantage Clipping (TFAC), a novel training framework designed to build efficient large reasoning models via token-level interventions. Specifically, TFAC functions along two dimensions: 1) The Frequency Dimension: It discourages inefficient loops and encourages deeper exploration by dynamically reducing the advantage scores of high-entropy tokens that are repeatedly generated within a single reasoning path. 2) The Time Dimension: It reduces excessive overthinking of the system by establishing a historical baseline for the occurrence count of each critical token in previously successful trajectories, and clipping the advantages of tokens that exceed this baseline during training. Crucially, to preserve the model's exploratory capabilities on novel problems, this suppression mechanism is automatically disabled when no historical record of success is available. Experiments conducted on the Deepseek-Distill-32B and Qwen3-8B models show that TFAC outperforms leading baseline methods, improving performance by 2.3 and 3.1 percentage points, respectively, while simultaneously reducing inference costs by 35% and 28% in scenarios where correct answers are generated. These results validate the significant efficacy of TFAC in training large reasoning models that are both powerful and highly efficient.

Cross-tokenizer knowledge distillation, where the teacher and student employ different tokenizers, is becoming increasingly prevalent, yet it poses underexplored challenges: existing methods fail to capture the rich knowledge encoded in teacher logits, as evidenced by the neglect of semantic information, inaccurate and biased logit alignment, and discarding distributional structure—ultimately leading to unfavorable distillation. To address these issues, we propose SeDi, a semantics and distribution-aware knowledge transfer framework tailored for cross-tokenizer distillation. To preserve factual knowledge, SeDi employs bipartite graph-based alignment at the tokenization level and a sliding window re-encoding strategy at the vocabulary level, enabling unbiased transfer of the teacher’s next-token predictions into the student’s vocabulary space. To further retain distributional information, we align the student’s entropy with that of the teacher by incorporating the student’s own logits during training, which helps to mitigate the exposure bias problem. Experiments on ten datasets across three task domains and five different teacher-student model pairs with varying vocabulary sizes demonstrate that SeDi delivers substantial improvements, with gains of up to 19.8%.

Visual language models (VLMs) have made significant progress in image captioning tasks, yet recent studies have found they are vulnerable to backdoor attacks. Attackers can inject undetectable perturbations into the data during inference, triggering abnormal behavior and generating malicious captions. These attacks are particularly challenging to detect and defend against due to the stealthiness and cross-modal propagation of the trigger signals. In this paper, we identify two key vulnerabilities by analyzing existing attack patterns: (1) the model exhibits abnormal attention concentration on certain regions of the input image, and (2) backdoor attacks often induce semantic drift and sentence incoherence. Based on these insights, we propose Semantic Reward Defense (SRD), a reinforcement learning framework that mitigates backdoor behavior without requiring any prior knowledge of trigger patterns. SRD learns to apply discrete perturbations to sensitive contextual regions of image inputs via a deep Q-network policy, aiming to confuse attention and disrupt the activation of malicious paths. To guide policy optimization, we design a reward signal named semantic fidelity score, which jointly assesses the semantic consistency and linguistic fluency of the generated captions, encouraging the agent to achieve a robust yet faithful output. SRD offers a trigger-agnostic, policy-interpretable defense paradigm that effectively mitigates local (TrojVLM) and global (Shadowcast) backdoor attacks, reducing ASR to 3.6% and 5.6% respectively, with less than 15% average CIDEr drop on the clean inputs.

Knowledge distillation (KD) transfers capabilities from large language models (LLMs) to smaller students, yet it can fail unpredictably and also underpins model leakage risks.Our analysis revealed several distillation traps: tail noise, off-policy instability, and, most fundamentally, the teacher-student gap, that distort training signals.These traps manifest as overconfident hallucinations, self-correction collapse, and local decoding degradation, causing distillation to fail.Motivated by these findings, we propose a post-hoc calibration method that, to the best of our knowledge, for the first time enables control over a teacher's distillability via reinforcement fine-tuning (RFT).Our objective combines task utility, KL anchor, and acrosstokenizer calibration reward.This makes distillability a practical safety lever for foundation models, connecting robust teacher-student transfer with deployment-aware model protection.Experiments across math, knowledge QA, and instruction-following tasks show that students distilled from distillable calibrated teachers outperform SFT and KD baselines, while undistillable calibrated teachers retain their task performance but cause distilled students to collapse, offering a practical knob for both better KD and model IP protection.

Zheng Zhang, Qi Liu, Siyuan Liang, Ning Li, Zirui Hu, Weibo Gao, Rui Li, Zhenya Huang, Leszek Rutkowski, Baosheng Yu, Dacheng Tao. Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2026.

This paper addresses the iterative learning control (ILC) problem for linear time-invariant systems operating under bit-rate constraints and denial-of-service (DoS) attacks. An adaptive encoding–decoding mechanism is proposed, where the encoder dynamically adjusts its quantization range according to the allocated bit rate of each sensor node, thereby ensuring reliable decoding under bandwidth limitations. To minimize the tracking error bound, a constrained optimization problem is formulated and solved using a particle swarm optimization (PSO)-based allocation scheme, which achieves superior resource utilization than conventional Lagrange multiplier methods. Furthermore, a novel compensation strategy based on single exponential smoothing is introduced into the ILC framework, providing accurate prediction of lost data and strong resilience against both isolated and successive DoS attacks. Theoretical analysis establishes boundedness of the tracking error, and simulation studies demonstrate that the proposed method significantly improves trajectory tracking accuracy and robustness relative to existing methodologies.

In this paper, a novel hidden semi-Markov model (HS-MM) in the continuous-time domain is proposed, and theH∞static output feedback control problem for hidden semi-Markov jump linear systems (HS-MJLSs) is investigated. Data transmission via a wireless network medium in HS-MJLSs is vulnerable to cyber attacks. Therefore, we adopt a resilient control strategy to ensure the stability of HS-MJLSs under hybrid cyber attacks, factoring in both denial-of-service attacks and false data injection attacks. In comparison with the conventional Markov jump linear systems, the HS-MM proposed by us is more universal, which is reflected in the following two aspects: (i) The system mode is considered to be undetectable. The jumping of the controller depends on the observation mode, and is asynchronous with the jumping of the system mode. (ii) Both the mode jumps of the original system and controller are considered to be related to the sojourn-time, which is more reasonable than the memoryless jumping processes. By using the established HS-MM and a Lyapunov-based approach, some sufficient conditions are obtained to ensure the stochastic stability of the HS-MJLSs with anH∞performance and hybrid cyber attacks. Finally, the proposed theoretical results are demonstrated by two simulation examples.

This paper investigates the resilient dynamic feedback control problem for fuzzy multi-rate systems (FMRS) subject to a novel important-data-based attack (IDBA). In practical engineering systems, physical limitations such as sensor aging, quantization effects, and unstable power supplies often cause discrepancies between measurement outputs and actual system states, leading to multi-rate sampling behaviors. To accurately characterize such sampling inconsistencies, a nonhomogeneous Markov sampling model is introduced, which provides a more realistic description and reduces conservatism compared with traditional homogeneous models. Motivated by the need for stealthy yet destructive cyber attacks, this paper further proposes an IDBA scheme, in which the adversary selectively identifies and disrupts only the most critical data nodes based on their instantaneous impact on system behavior. To mitigate the severe disruptions caused by IDBA, a resilient dynamic output feedback controller is designed, thereby enhancing adaptability under adverse network conditions. Simulation studies, including both numerical and practical examples, demonstrate that the proposed control framework effectively preserves system stability and achieves superior performance in the presence of IDBA.

This paper introduces a novel Balanced Binary Whale Optimization Algorithm (BB-WOA) designed specifically for dynamic feature selection in Green Cloud Computing (GCC). Traditional feature selection methods used in cloud resource forecasting often suffer from either suboptimal predictive performance or excessive computational complexity. To address this, we propose significant algorithmic enhancements over the standard Binary Whale Optimization Algorithm (B-WOA), including dynamic binary transition functions, progressive scaling, diversified population initialization via Sobol sequences, balanced exploration-exploitation strategies, and an activation-based recovery mechanism. Our comprehensive experimental evaluation using real-world cloud resource data demonstrates that BB-WOA outperforms existing methods. Specifically, BB-WOA reduces predictive error (RMSE) by up to 7.45% compared to B-WOA and 1.55% compared to Genetic Algorithm (GA). Moreover, BB-WOA achieves computational improvements, reducing execution time by approximately 38.30%, 64.13%, and 78.53% over B-WOA, Random Search (RS), and GA, respectively. Environmentally, the proposed method reduces energy consumption by 38.48% relative to B-WOA, 64.18% compared to RS, and 52.55% relative to GA, while simultaneously selecting significantly fewer features (a reduction of up to 70.77%). These results underscore the effectiveness and sustainability of BB-WOA, positioning it as a highly competitive and environmentally friendly solution.

This article addresses the challenges posed by multimode denial of service (DoS) and deception attacks in observer-based sliding mode control (SMC) for fuzzy nonlinear systems. A novel multimode DoS attack model is introduced, incorporating time-varying sojourn probabilities to provide a more accurate and computationally efficient representation of the stochastic nature of these attacks. This model overcomes the limitations of traditional Markov-based models by capturing dynamic attack behaviors. Deception attacks are modeled as unbounded nonlinear functions, and adaptive neural networks (NNs) are employed to approximate their complex behaviors, significantly reducing their detrimental impact on system stability. A fuzzy sliding surface is designed based on the switching rule for sojourn probabilities, and an observer-based SMC law is proposed to ensure the mean square estimation upper bound of the fuzzy nonlinear systems, guaranteeing stability despite the presence of cyberattacks. Finally, the validity and superiority of the proposed control strategy are demonstrated through a tunnel diode circuit model.

This study introduces a novel integrated control and communication framework specifically designed for two-dimensional systems. First, a super-twisting-based second-order sliding mode control strategy is proposed to against uncertainties and effectively eliminate chattering. Second, a self-triggered communication protocol is developed, markedly reducing communication overhead by proactively scheduling transmissions without continuous error monitoring. Third, a dynamic binary encoding strategy is introduced, dynamically adjusting quantization intervals according to real-time historical data, thus improving signal accuracy and reducing quantization errors. Explicit sufficient conditions are derived, guaranteeing both practical reachability of the sliding manifold and the ultimate boundedness of the closed-loop system. Additionally, an optimization framework employing the Grey Wolf Optimizer algorithm is formulated to optimize sliding mode parameters, further reducing the convergence bounds. Comprehensive simulation results illustrate the superior performance and effectiveness of the proposed framework compared to existing methods.

Text-driven 3D scene editing has attracted considerable interest due to its convenience and user-friendliness. However, methods that rely on implicit 3D representations, such as Neural Radiance Fields (NeRF), while effective in rendering complex scenes, are hindered by slow processing speeds and limited control over specific regions of the scene. Moreover, existing approaches, including Instruct-NeRF2NeRF and GaussianEditor, which utilize multi-view editing strategies, frequently produce inconsistent results across different views when executing text instructions. This inconsistency can adversely affect the overall performance of the model, complicating the task of balancing the consistency of editing results with editing efficiency. To address these challenges, we propose a novel method termed Single-View to 3D Object Editing via Gaussian Splatting (SVGS), which is a single-view text-driven editing technique based on 3D Gaussian Splatting (3DGS). Specifically, in response to text instructions, we introduce a single-view editing strategy grounded in multi-view diffusion models, which reconstructs 3D scenes by leveraging only those views that yield consistent editing results. Additionally, we employ sparse 3D Gaussian Splatting as the 3D representation, which significantly enhances editing efficiency. We conducted a comparative analysis of SVGS against existing baseline methods across various scene settings, and the results indicate that SVGS outperforms its counterparts in both editing capability and processing speed, representing a significant advancement in 3D editing technology. For further details, please visit our project page at: https://amateurc.github.io/svgs.github.io/ .

This work delves into an optimal hybrid transmission strategy for remote state estimation (RSE), where some smart sensors observe various systems and transmit their local state estimates to a remote estimator. Inspired by the high bandwidth of a high-frequency (HF) link alongside the robust reliability and low energy consumption of a low-frequency (LF) link, a novel hybrid transmission strategy is proposed. This strategy allows smart sensors to utilize either the HF link or the LF link dynamically, each with distinct channel characteristics and energy consumption. To achieve highly reliable and energy-efficient transmission, it is imperative to devise an optimal transmission scheduling strategy that dictates the selection of the link and channel allocation. Formulating this challenge as a Markov decision process (MDP), a sufficient condition ensuring the existence of an optimal deterministic and stationary (ODS) policy is established. The structural characteristics of the optimal policy are derived. Furthermore, a deep reinforcement learning (DRL) approach, specifically the dueling double deepQ-network (D3QN) algorithm, is employed to approximate the optimal policy. Finally, the structural results and the effectiveness of the DRL algorithm are verified by simulation examples.

This article addresses discrete-time Takagi–Sugeno fuzzy switching systems utilizing a dual-channel transmission framework combined with an innovative switching rule. A novel duration-based switching mechanism is proposed, capturing realistic multiple-distribution characteristics by employing a joint distribution function involving both the current mode and its duration. Unlike traditional semi-Markov-based transition probabilities, this method emphasizes sojourn probabilities, significantly reducing the parameter estimation complexity and computational burden. The developed mechanism supports generalized duration distributions, surpassing traditional single-distribution limitations and thus lowering conservatism in the analysis. To improve the modeling accuracy and practical applicability, mode-dependent fuzzy membership functions are adopted. Observer-based feedback control strategies are implemented to effectively estimate state variables, addressing practical challenges in state measurement. In addition, the dual-channel strategy is introduced to mitigate packet loss effects and enhance data transmission reliability. The effectiveness and superiority of the proposed method are demonstrated through simulations on a single-link robotic arm.

This article investigates distributed state estimation of complex networks (CNs) with limited communication capacity. A random transmission strategy is used to overcome the communication capacity constraint between two nodes. A distributed state estimator that makes use of the partially available measurements is designed. To handle the cross-term, Young’s inequality is employed, and an upper bound (UB) for the state prediction error covariance (PEC) is derived. An optimal estimation gain is then devised based on the derived state PEC. The stability of the UB is analyzed using a vectorization approach, and then a sufficient condition for stability is obtained. Finally, a numerical simulation is carried out to validate the effectiveness of the proposed distributed estimator and confirm the accuracy of the derived UB.

The automatic design of a 3D tooth model plays a crucial role in dental digitization. However, current approaches face challenges in compositional 3D tooth generation because both the layouts and shapes of missing teeth need to be optimized. In addition, collision conflicts are often omitted in 3D Gaussian-based compositional 3D generation, where objects may intersect with each other due to the absence of explicit geometric information on the object surfaces. Motivated by graph generation through diffusion models and collision detection using 3D Gaussians, we propose an approach named DM-CFO for compositional tooth generation, where the layout of missing teeth is progressively restored during the denoising phase under both text and graph constraints. Then, the Gaussian parameters of each layout-guided tooth and the entire jaw are alternately updated using score distillation sampling (SDS). Furthermore, a regularization term based on the distances between the 3D Gaussians of neighboring teeth and the anchor tooth is introduced to penalize tooth intersections. Experimental results on three tooth-design datasets demonstrate that our approach significantly improves the multiview consistency and realism of the generated teeth compared with existing methods.

This paper investigates the optimal control problem for interconnected systems with unknown system dynamics through a two-stage reinforcement learning method. First, to address the impact of interconnection term, the optimal control problem for interconnected systems is transformed into obtaining the solution of game algebraic Riccati equations within the framework of$H_{\infty }$control method. Furthermore, existing optimal control approaches for interconnected systems necessitate precise knowledge of system dynamics, which is difficult to obtain accurately or involves high costs. Thus, we introduce a two-stage reinforcement learning method. The admissible control policies are obtained using the homotopy-based iteration method in the first stage. Then, the optimal control policies are obtained through the policy iteration method in the second stage. The two-stage method not only eliminates the requirement for system dynamics and initial admissible control policies but also ensures convergence speed and accuracy, significantly enhancing its practicality. Finally, a two-machine power system example is provided to validate the feasibility of the two-stage method. Note to Practitioners—Interconnected systems, a class of systems composed of multiple local subsystems, find wide applications in various fields such as power systems, transportation networks, and spatially interconnected systems. Particularly, the optimal control problem of interconnected systems has gradually become a focal point of current research. However, the current research on the optimal control problem of interconnected systems is still constrained by the system dynamics and the initial stability of the system. To relax these limitations, this paper introduces a two-stage method. A homotopy-based iteration approach is employed to obtain control policy and interconnection policy that make the system closed-loop stable, thus achieving the optimal solution. Furthermore, the data-driven approach overcomes the limitations imposed by system dynamics. The feasibility of the two-stage method is illustrated by a two-machine power system model.

This study investigates the security control issue for fuzzy singularly perturbed systems (FSPSs) subjected to Denial-of-Service (DoS) attacks. To accurately characterize the dynamics of DoS attacks, an innovative duration-dependent switching rule is proposed, guided by a joint probability distribution function that incorporates both the current attack mode and its dwell time. Unlike traditional Markov/semi-Markov chains, this approach utilizes duration-conditioned sojourn probabilities to eliminate the necessity for explicit transition probabilities estimation, significantly simplifying implementation and reducing computational complexity in multi-mode attack scenarios. Additionally, an attack compensation mechanism explicitly designed to maintain system stability is developed. Furthermore, a non-monotonic Lyapunov function aligned with the proposed switching logic is introduced, resulting in less conservative stability criteria that guarantee the mean-square stability of closed-loop FSPSs. The effectiveness and resilience of the proposed methodology are substantiated through a numerical simulation and a practical case study, highlighting its superior performance in mitigating the adverse effects of DoS attacks compared to existing methodologies. Note to Practitioners—With the continuous advancement in network-based control technologies, singularly perturbed systems have gained significant attention due to their inherent capability to handle dynamics that evolve across multiple time scales. Practical engineering systems frequently encounter uncertainties, including external disturbances, equipment failures, and cyber threats such as Denial-of-Service (DoS) attacks, which can profoundly impact system states and operational stability. Particularly, DoS attacks introduce substantial risks depending on their duration and operational mode. Recognizing these practical concerns, this study proposes an advanced joint probability distribution model that explicitly incorporates both the current attack mode and its duration. By adopting a duration-dependent sojourn probability based on general distributions, the model closely aligns with real-world operational conditions. Additionally, industrial control systems experiencing DoS attacks often face critical information loss, potentially leading to severe instabilities, such as oscillations or divergence. To address these risks, this paper presents a sophisticated gain compensation mechanism designed explicitly for maintaining robust system performance under attack-induced disruptions. This approach ensures that reliable control gains remain available even when essential data is compromised, significantly enhancing system resilience and stability under challenging attack conditions.

This paper investigates the fuzzy censoring control problem for two-dimensional (2-D) discrete fuzzy systems characterized by the Roesser model. A novel event-triggered random protocol based on nonhomogeneous sojourn probability is proposed to effectively alleviate communication loads in 2-D fuzzy systems. To address signal fading effects, a censoring strategy is introduced to quantify the severity of signal degradation in transmitted data. The censored control input is subsequently modeled through a Tobit framework, thereby facilitating the design of an integrated state feedback control scheme for 2-D fuzzy systems. By leveraging Lyapunov stability theory, sufficient conditions are derived to guarantee the asymptotic mean-square stability of the closed-loop system while satisfying prescribed performance specifications. Furthermore, parameters are optimized via genetic algorithms to mitigate redundant conservatism associated with linear matrix inequalities. Finally, a gas absorption, water stream heating and air drying industrial case are carried out to verify the theoretical rigor and practical effectiveness of the proposed methodology.

This study develops the space-time sampled-data control problem for memristor-based reaction-diffusion neural networks (MRDNNs) using a memory event-triggering scheme. Unlike traditional Markov switching models with fixed transition probabilities, the nonhomogeneous sojourn probabilities are capable of describing the switching behavior more accurately. By effectively leveraging historical transmitted packets, an innovative mode-dependent memory event-triggering scheme that incorporates nonhomogeneous sojourn probability information is introduced, optimizing communication resource utilization. A space-time sampled-data control law is designed by sampling both spatial and temporal domains, significantly reducing network communication resource consumption while achieving the desired system performance. The validity and superiority of the proposed space-time sampled-data control strategy are demonstrated through a simulation example.

This paper proposes a novel prescribed performance control scheme for stochastic micro-electro-mechanical system (MEMS) gyroscopes, addressing three critical issues overlooked by existing methods: control torque oscillation during rapid convergence, deviation in steady-state tracking errors in a global asymmetric design, and violation of monotonic constraints due to insufficient input capability. To tackle these challenges, the paper proposes a quadratic prescribed performance function design, a local asymmetric constraint design, and a customized non-monotonic design. These innovations effectively resolve the technical difficulties and establish comprehensive performance specifications for stochastic MEMS gyroscopes. The proposed scheme ensures boundedness in probability for all closed-loop signals and convergence of the tracking error to an arbitrarily small residual within a prescribed time. Simulation results confirm the effectiveness and superiority of the scheme.

This article proposes a novel framework of nonhomogeneous sojourn probabilities in the 2-D domain and investigates the observer-based security control issue for 2-D fuzzy switched systems (SS) under an adaptive-threshold-based saturation function. In comparison with the existing homogeneous/nonhomogeneous Markov processes, the proposed nonhomogeneous sojourn probability framework enhances measurability and reduces computational complexity, making it a more reasonable approach. The security control challenge for 2-D fuzzy SS is addressed in the context of randomly occurring attacks with uncertain occurrence rates, where the uncertainty is norm-bounded and asymmetrical around the nominal rate. To mitigate the effects of these abnormal attacks, a novel adaptive-threshold-based saturation function is employed, ensuring that transmitted information remains within an acceptable range. Leveraging Lyapunov theories, sufficient conditions are attained to guarantee the asymptotically mean square stable with noise attenuation performance. The validity of the proposed theoretical results is demonstrated through a simulation example.

The principal innovative contribution of this study resides in the introduction of a category of fractional delayed large-scale neural networks characterized by intricate topological structures. Additionally, this article provides a comprehensive exploration of novel outcomes linked to fractional order-induced bifurcations in large-scale networks. In the initial step, the correlation of the artificial neural network and the graphical neural network is established through the Mason's diagram method. Subsequently, the system's characteristic equations are derived by employing the Coates' flow graph decomposition method. Moving on, through the concept of the global element, an exhaustive investigation delves into the distribution of eigenroots. The sum of synaptic transmission delays among neurons is considered as a bifurcation parameter, with an analysis focused on the stability of the trivial equilibrium and the existence of the Hopf bifurcation. Following this, the optimal fractional order-dependent stability interval is determined using the implicit function array curve method, presenting a novel approach for critical value determination. Finally, the drawn conclusions are substantiated through multiple sets of computer simulations. It is indicated that an increase in delay precipitates the onset of Hopf bifurcation. Moreover, a reduction in the fractional order significantly improves the steady-state performance of the system. However, once the fractional order value descends below the left stability boundary, the system's stability is compromised, leading to the emergence of periodic oscillations. The prediction algorithm proposed in this article offers valuable insights into selecting the appropriate fractional order for large-scale complex networks.

The bipartite consensus (BC) issue for nonlinear multiagent systems (NMASs) with unknown system dynamics information is investigated in this article. Initially, the dynamics of NMASs are represented using the Takagi-Sugeno (T-S) fuzzy model. Subsequently, to achieve distributed control, a minmax game policy is introduced, where each agent aims to minimize its performance index while its neighbors attempt to maximize it. Consequently, the BC problem for NMASs is reformulated as a zero-sum game, transforming the controller design into solving a set of game algebraic Riccati equations (GAREs). To solve such equations, a novel scaling off-policy iteration (PI) algorithm is proposed. The key features of the proposed learning algorithm can be outlined in three main aspects: 1) during the learning process, the reliance on system dynamics is relaxed; 2) compared with the PI method, the requirement for initial admissible control policies is eliminated; and 3) a more rapid convergence speed is achieved than traditional value iteration. Finally, the effectiveness and advantages of the proposed method are validated through a simulation example and a series of comparative experiments.

This article presents an adaptive neural network control scheme with prescribed performance for stochastic nonlinear systems. Unlike existing adaptive stochastic control schemes that primarily utilize deterministic neural networks for approximations in complex stochastic environments, we employ stochastic neural networks to approximate the stochastic nonlinear terms, effectively resolving the "memory overflow" issue. Moreover, we propose a novel prescribed performance design method, which distinguishes itself from the previous prescribed performance control schemes by integrating a quadratic characteristic capable of suppressing transient input vibrations, along with a local asymmetric characteristic that optimize both transient output overshoot and steady-state error bias. Furthermore, the proposed control scheme is implemented within a fixed-time framework to ensure that all closed-loop systems are fixed-time bounded in probability, with the tracking error consistently within the predefined performance bounds. Simulation results validate the effectiveness of the proposed control scheme.

This study explores the protocol-based sampled-data control for T-S fuzzy reaction–diffusion neural networks (RDNNs) with nonhomogeneous sojourn probabilities (NSPs). By incorporating a deterministic switching signal, a new framework of NSPs is developed to characterize the random behaviors of fuzzy RDNNs. Unlike previous studies, this work introduces multiasynchronous switching among fuzzy RDNNs, triggering conditions, and controllers using dynamic asynchrony models, effectively capturing mode transitions through NSPs and detection probabilities. An improved adaptive event-triggered protocol is created by integrating fuzzy rules and detection probability information. Moving beyond traditional time-domain sampled-data control strategies, a space–time sampled-data control approach is proposed to significantly reduce communication load. Benefiting from Lyapunov theory, criteria are attained for ensuring the mean-square exponential stability of the systems under consideration. Ultimately, the proposed space–time sampled-data control strategy is validated through a simulation example, highlighting its effectiveness and superiority.

Many machine learning problems can be formulated as non-convex multi-player games. Due to non-convexity, it is challenging to obtain the existence condition of the global Nash equilibrium (NE) and design theoretically guaranteed algorithms. This paper studies a class of non-convex multi-player games, where players' payoff functions consist of canonical functions and quadratic operators. We leverage conjugate properties to transform the complementary problem into a variational inequality (VI) problem using a continuous pseudo-gradient mapping. We prove the existence condition of the global NE as the solution to the VI problem satisfies a duality relation. We then design an ordinary differential equation to approach the global NE with an exponential convergence rate. For practical implementation, we derive a discretized algorithm and apply it to two scenarios: multi-player games with generalized monotonicity and multi-player potential games. In the two settings, step sizes are required to be O(1/k) and O(1/√k) to yield the convergence rates of O(1/ k) and O(1/√k), respectively. Extensive experiments on robust neural network training and sensor network localization validate our theory. Our code is available at https://github.com/GuanpuChen/Global-NE.

Turing patterns in diffusion neural networks are strongly associated with the performance of artificial intelligence model. However, two practical considerations are rarely mentioned in the mechanism research on the problem of diffusion-coupled cellular neural networks (CNNs): the effects of memristor and three-dimensional (3-D) network structure. Furthermore, facilitating the formation of the determined Turing pattern is expected to make the neural network exhibit the desired intelligence. This paper first realizes the 3-D diffusive networking for a primary cell circuit with memristor characteristics and utilizes a proportional-derivative (PD) control strategy to drive the pattern formation. Next, the characteristic equation is derived using the spatial eigenfunction-based decoupling method. Then, by tracing the root distribution of the characteristic equation, some analytical conditions for the stability or forming Turing patterns are deduced. In addition, the central manifold reduction and linear analysis approaches are utilized to derive the amplitude equations of 3-D Turing patterns and investigate their stability. At last, some numerical simulations are provided to illustrate the results. It is also demonstrated that PD control and memristor occupy dominant positions in various CNNs’ Turing patterns. After determining the pattern stability, the implementation of PD control can be considered an effective means to facilitate stable networks to experience Turing instability and generate switchable 3-D pattern.

This paper aims to consider the finite-time synchronization of a class of discontinuous fuzzy competitive networks with non-differential leakage delays. Different from the previous competitive networks and leakage-delayed networks, the activation functions are discontinuous and the leakage delays are non-differential. Without using the Lyapunov function methods, the synchronization in a finite-time is investigated based on the comparison principle and inequality techniques, which are used for the first time in the study of the competitive networks and leakage-delayed networks. Moreover, the feedback controller and bounded feedback controller are designed to help achieve the finite-time synchronization. The bounded feedback controller is formulated by the functionarctanand a new inequality lemma is given, which plays a vital role in achieving the synchronization via bounded feedback controller. Although both controllers can help achieve the finite-time synchronization, it can be found that there is a big difference that the value of feedback controller is greatly influenced by system's state variables, but bounded controller is not. Such fact leads to different settling-times. The conclusion is that bounded controller has more advantages in shortening convergence time through numerical simulations.

This study examines observer-based sliding mode control for stochastic sampling Takagi–Sugeno fuzzy systems under a stochastic communication protocol (SCP). Aimed at reducing network bandwidth use and lessening the negative effects of signal transmission disruptions, the sampling signal sequence exhibits dual stochastic characteristics, influenced by both stochastic sampling and network attacks. The stochastic intervals between samples are governed by a Markov chain, and the samples are vulnerable to denial-of-service attacks during network transmission. Moreover, a novel SCP is developed utilizing sojourn probability and residence time information, enhancing communication efficiency. Subsequently, a suitable sliding mode control law is crafted based on the observed data, providing a sufficient condition for the mean square exponentially ultimate boundedness of the closed-loop system. The efficacy of the constructed methodology is demonstrated using a tunnel diode circuit model.

This work proposes a fully distributed moving horizon estimation method with an event-triggered communication strategy over wireless sensor networks. The proposed method calculates a local state estimation of each sensor by minimizing a quadratic objective function, which involves a fused arrival cost that is computed in a distributed manner. This approach adjusts data transmission rate by selectively transmitting local information and allows the incorporation of constraints on the noise and state variables. In addition, the estimation error is proved to be uniformly bounded in the mean square sense under the condition that the network topology is strongly connected and the system is collectively observable. Finally, a target tracking example is presented to demonstrate the validity of the proposed approach.

Data deduplication is of vital importance for mobile cloud computing to cope with the explosive growth of outsourced mobile data. In order to ensure the privacy of sensitive mobile data against an untrusted cloud, Message-Locked Encryption (MLE) has been proposed to enable deduplication over ciphertext. However, MLE prohibits data access control since it uses deterministic content-derived encryption keys. Recently, a lightweight rekeying-aware encrypted deduplication system (REED) has been proposed to achieve dynamic access control for secure data deduplication. However, REED is vulnerable to key-retaining attack and stub-retaining attack, which leads to insecure access revocation, and thus cannot support secure dynamic access control. In response, we present AC-Dedup, an encrypted deduplication storage system that supportssecure dynamic access controlfor mobile cloud storage. At the core of AC-Dedup are two novel encryption techniques namedmixed message locked encryptionandrandom stub re-encryptionto resist the two types of attacks, respectively. To the best of our knowledge, AC-Dedup is the first practical system that achieves secure data deduplication and secure dynamic access control simultaneously. We conduct security analysis and experimental evaluation on mobile device and cloud platform with real-world IoT datasets. The results show that AC-Dedup enables secure and efficient dynamic access control while preserving deduplication effectiveness.

In this brief, we consider the stability of inertial memristor-based neural networks with time-varying delays. First, delayed inertial memristor-based neural networks are modeled as continuous systems in the flux-current-voltage-time domain via the mathematical model of Hewlett-Packard (HP) memristor. Then, they are reduced to delayed inertial neural networks with interval parameters uncertainties. Quasi-equilibrium points and quasi-stability are proposed. Quasi-stability criteria of delayed inertial memristor-based neural networks are obtained by matrix measure method, the Halanay inequality, and uncertainty technologies. In the end, a numerical example is provided to show the validity of our results.

Ant Colony Optimization (ACO) is an acclaimed method for solving combinatorial problems proposed by Marco Dorigo in 1992 and has since been enhanced and hybridized many times. This paper proposes a novel modification of the algorithm, based on the introduction of a two-dimensional pheromone into a single-criteria ACO. The complex structure of the pheromone is supposed to increase ants’ awareness when choosing the next edge of the graph, helping them achieve better results than in the original algorithm. The proposed modification is general and thus can be applied to any ACO-type algorithm. We show the results based on a representative instance of TSPLIB and discuss them in order to support our claims regarding the efficiency and efficacy of the proposed approach.

Computing generates intelligence. With this statement we do not mean computing’s capabilities of manipulating numbers, shapes, symbols, and even logics. What we mean is the ingenious design of computing structures which serve as the basis of intelligence generation during program running. In this panel discussion, we consider how to obtain such capabilities through some computing paradigms as examples, including principal computing, logic computing, discriminative computing, and generative computing. The panelists express their thoughts about the inherent advantages and disadvantages of each of these paradigms, in terms of their adaptivity, interpretability, generality and specificity, and dives into detailed discussions about Large Language Models (LLMs), a mainstream generative paradigm which leverages the strengths of large pre-trained models and downstream prompt tuning to deliver combined intelligence, superior to most existing frameworks in natural language processing. The panel outlines potential challenges of the generative paradigm, with a strong focus on LLMs, and emphasizes that future directions of such models will need to address (1) tackling bias, discrimination, and transparency challenges; (2) delivering logical answers with high specificity; (3) enabling personalized, lightweight, and rapid updating mechanisms; (4) assessing accreditation, tracing, and misusages; and (5) ensuring sustainable LLMs.

Gas composition and light mode of industrial greenhouses are some of the most determining factors in the process of growing vegetable crops in greenhouse conditions. The intelligentisation of information technologies for monitoring and control based on artificial intelligence methods can increase the efficiency of agrotechnical procedures for greenhouse cultivation. One of the efficient approaches in today's world practice is the development and implementation of trustworthy hybrid decision-support systems based on the techniques of Fuzzy logic and Human-in-the-Loop. The research object is non-stationary processes of complex intelligent transformation of measurement data on the concentration of carbon dioxide and effective energy illumination in the growing zone of industrial greenhouses. The scientific novelty and practical value of the obtained research results consist in creating a computer model for aggregation and intelligent processing of agricultural monitoring data for greenhouses. The developed computer model is fully transformed into peripheral-level software of Internet of Things systems for agricultural purposes. This makes it possible to implement the for-computing architecture of monitoring systems in greenhouses. The obtained research results make it possible to optimise the resources used in growing crops in greenhouse conditions through the implementation of hardware and software intelligent monitoring tools that are adaptive to the types and periods of crop vegetation. The scientific and applied effect of the research is creating the novel approach of development and practical use of intelligent technologies for agrotechnical monitoring by substantiating the methodological provisions for synthesis of structural and algorithmic organisation of the corresponding software and hardware solutions.

In this article, a novel hybrid reinforcement$Q$-learning control method is proposed to solve the adaptive fuzzy$H_{\infty }$control problem of discrete-time nonlinear Markov jump systems based on the Takagi–Sugeno fuzzy model. First, the core problem of adaptive fuzzy$H_{\infty }$control is converted to solving fuzzy game coupled algebraic Riccati equation, which can hardly be solved by mathematical methods directly. To solve this problem, an offline parallel hybrid learning algorithm is first designed, where system dynamics should be known as a prior. Furthermore, an online parallel$Q$-learning hybrid learning algorithm is developed. The main characteristics of the proposed online hybrid learning algorithms are threefold: 1) system dynamics are avoided during the learning process; 2) compared with the policy iteration method, the restriction of the initial stable control policy is removed; and 3) compared with the value iteration method, a faster convergence rate can be obtained. Finally, we provide a tunnel diode circuit system model to validate the effectiveness of the present learning algorithm.

This paper focuses on the fuzzy control problem for a class of semi-Markov jump singularly perturbed nonlinear systems with actuator saturation. Nonlinearities in the underlying systems are tackled with the Takagi-Sugeno fuzzy model. As distinct from the previous achievements, the case that the semi- Markov kernel with partially known information is taken into account such that the underlying systems can be extended to a wider scope. Given that actuator saturation is ubiquitous in engineering systems, the convex hull method is embraced to address this circumstance. To reduce the conservativeness of the obtained results, on the basis of Lyapunov stability theory and LaSalle's invariance principle, some criteria related to the sojourn time are presented. Next, the mean-square exponential stability of the underlying system is obtained based on the derived results. Finally, further experiment to verify the feasibility of the proposed methodology is given by a tunnel diode circuit model.

Text-to-image synthesis is a challenging problem, in which a complex scene contains diverse objects of various sizes and sub-images of objects belonging to the same class have diverse forms from different perspectives. Thus, synthesis models have difficulty in capturing varied objects in the complex scene. To alleviate these problems, we devise an independent object-level decomposing and enhancing generative adversarial networks, denoted as InDecGAN, to synthesize complex images and capture varied objects in a complex scene. Specifically, InDecGAN fully utilizes the independent object-level information, bounding boxes and high-resolution images of objects in training, by employing independent object-level pathways to synthesize varied objects. The independent object-level pathway integrates an independent object-level adversarial loss and the bounding box information to learn the visual features of objects independently, then, the main pathway exploits the features provided by the object-level pathway to compose the full scene and synthesize images. In addition, we analyze the generalization properties of the proposed InDecGAN and demonstrate the improvement from the perspective of the model architecture. Moreover, extensive experiments conducted on a widely used dataset are presented to demonstrate that the proposed model with an independent object-level pathway produces synthesized images of significantly improved quality.

This article studies the asynchronous fault detection filter problem for discrete-time memristive neural networks with a stochastic communication protocol (SCP) and denial-of-service attacks. Aiming at alleviating the occurrence of network-induced phenomena, a dwell-time-based SCP is scheduled to coordinate the packet transmission between sensors and filter, whose deterministic switching signal arranges the proper feedback switching information among the homogeneous Markov processes (HMPs) for different scenarios. A variable obeying the Bernoulli distribution is proposed to characterize the randomly occurring denial-of-service attacks, in which the attack rate is uncertain. More specifically, both dwell-time-based SCP and denial-of-service attacks are modeled by means of compensation strategy. In light of the mode mismatches between data transmission and filter, a hidden Markov model (HMM) is adopted to describe the asynchronous fault detection filter. Consequently, sufficient conditions of stochastic stability of memristive neural networks are devised with the assistance of Lyapunov theory. In the end, a numerical example is applied to show the effectiveness of the theoretical method.

This article presents a decentralized learning control method for a class of partially unknown nonlinear systems with asymmetric control input constraints and mismatched interconnections via a novel dynamic event-triggering condition. By employing an integral reinforcement learning strategy, the system drift dynamics can be avoided in the learning process. Meanwhile, a critic neural network is designed to obtain the approximated value function and tuned by using the gradient descent approach. Furthermore, a novel dynamic event-triggering condition is designed to determine the occurrence of an event by introducing a dynamic variable. By using the Lyapunov theory, all signals in the closed-loop system are proved to be uniformly ultimately bounded. Finally, we present a nonlinear interconnected system and an interconnected power system to verify the effectiveness of the proposed method.

The Sun's activity has an enormous impact on the climate and technology on the Earth. The NASA Solar Dynamic Observatory (SDO) spacecraft produces an enormous number of images a day. We propose a technique for fast retrieving and indexing these images. We use image preprocessing, novel edge detection based on a nonparametric Parzen kernel estimator for discovering active regions and a convolutional autoencoder to generate a concise solar image description. Such real-valued hashes can be used to retrieve similar Sun images. Moreover, they can be a basis to classify or predict Solar activity. The experiments showed excellent accuracy and relatively good speed of the proposed method.

This paper deals with global synchronization problem of multiple discrete-time Markovian jump memristor-based neural networks (DTMJMNNs) with mixed mode-dependent delays via a novel event-triggered impulsive coupling control (ETICC). The parameters of the multiple DTMJMNNs and the mixed time delays (both discrete and distributed delays) switch randomly according to a Markov chain. In the ETICC strategy, the controller does not work all the time, but only works at impulse instants determined by specific events. In particular, the coupling matrix can be non-Laplacian. By using the Lyapunov stability theory, linear matrix inequalities (LMIs), and the Kronecker product, some sufficient conditions for global synchronization of multiple DTMJMNNs under the event-triggered strategy are derived. Two examples are presented to test the validity of the theoretical analysis results.

In this article, we study the finite-time stabilization and the asymptotic stabilization with probability one of Markovian jump Boolean control networks (MJBCNs) by sampled-data state feedback controls (SDSFCs). Based on the semi-tensor product (STP), we introduce an augmented variable multiplied by the vector form of the switching signal and the state of MJBCN. We find that under SDSFC, the sequence of the states of the augmented variable at sampling instants satisfies the Markov property. Based on the convergences of the switching signal and the augmented variable, we obtain the sufficient and necessary criteria for the finite-time stabilization and the asymptotic stabilization of MJBCNs by SDSFCs, respectively. Moreover, for the two kinds of stabilization, the feedback matrices of SDSFCs are constructed, respectively. Finally, the obtained results are applied to an apoptosis network and a model of the lactose operon in the Escherichia Coli.

A new type of asymptotic stability for nonlinear hybrid neutral stochastic systems with constant delays was investigated recently, where the criteria depended on the delays’ sizes. Unfortunately, developed theory so far is not sufficient to deal with challenging problems of the decay rate, time-varying delays, and nonautonomous issues. These problems have not been tackled in the existing literature. Consequently, under the weak constraints, this article focuses on the general decay, including the exponential stability and the polynomial stability, for nonlinear nonautonomous hybrid neutral stochastic systems with time-varying delays by the approach of the multiple degenerate functionals. Moreover, this article derives the interesting assertions related to the general$H_{\infty }$stability and the polynomial growth at most.

In this article, we study the discrete-time decentralized optimization problems of multiagent systems by an event-triggering interaction scheme, in which each agent privately knows its local convex cost function, and collectively minimizes the total cost functions. The underlying interaction and the corresponding weight matrix are required to be undirected connected and doubly stochastic, respectively. To resolve this optimization problem collaboratively, we propose a decentralized event-triggering algorithm (DETA) that is based on the consensus theory and inexact gradient tracking technique. DETA involves each agent interacting with its neighboring agents only at some independent event-triggering sampling time instants. Under the assumptions that the global convex cost function is coercive and has Lipschitz continuous gradient, we prove that DETA steers all agents' states to an optimal solution even with nonuniform constant step sizes. Moreover, our analysis also shows that DETA converges at a rate of O(1/√t) if the step sizes are uniform and do not exceed some upper bounds. We illustrate the effectiveness of DETA on a canonical simple decentralized parameter estimation problem.

This article focuses on the composite$\mathcal {H}_{\infty }$synchronization problem for jumping reaction–diffusion neural networks (NNs) with multiple kinds of disturbances. Due to the existence of disturbance effects, the performance of the aforementioned system would be degraded; therefore, improving the control performance of closed-loop NNs is the main goal of this article. Notably, for these disturbances, one of them can be described as a norm-bounded, and the other is generated by an exogenous model. In order to reject the above one kind of disturbance, a disturbance observer is developed. Furthermore, combining the disturbance observer approach and conventional state-feedback control scheme, a composite disturbance rejection controller is specifically designed to compensate for the influences of the disturbances. Then, some criteria are established based on the general Lyapunov stability theory, which can ensure that the synchronization error system is stochastically stable and satisfies a fixed$\mathcal {H}_{\infty } $performance level. A simulation example is finally presented to verify the availability of our developed method.

In this article, a periodic self-triggered impulsive (PSTI) control scheme is proposed to achieve synchronization of neural networks (NNs). Two kinds of impulsive gains with constant and random values are considered, and the corresponding synchronization criteria are obtained based on tools from impulsive control, event-driven control theory, and stability analysis. The designed triggering protocol is simpler, easier to implement, and more flexible compared with some previously reported algorithms as the protocol combines the advantages of the periodic sampling and event-driven control. In addition, the chaotic synchronization of NNs via the presented PSTI sampling is further applied to encrypt images. Several examples are also utilized to illustrate the validity of the presented synchronization algorithm of NNs based on PSTI control and its potential applications in image processing.

The finite-time synchronization problem is investigated for the master-slave complex-valued memristive neural networks in this article. A novel Lyapunov-function based finite-time stability criterion with impulsive effects is proposed and utilized to design the decentralized finite-time synchronization controller. Not only the settling time but also the attractive domain with respect to the impulsive gain and average impulsive interval, as well as initial values is derived according to the sufficient synchronization condition. Two examples are outlined to illustrate the validity of our hybrid control strategy.

This paper investigates a distributed static and dynamic self-triggered impulsive control for nonlinear multiagent systems (MASs) where the impulsive gains follow a normal distribution, respectively. By integrating the distributed self-triggered control scheme with the impulsive control approach, a novel distributed impulsive controller is developed. The goal of the consensus of MASs can be realized using the proposed methods and several consensus criteria are obtained. Our schemes have some distinct superiorities, including the impulsive gains obeying a normal distribution, avoiding the continuous communication, and reducing the sampling frequency. Hence, compared with the existing literature, the conservativeness coming from the limitation of impulse gain and the sampling frequency is degraded, and it effectively extends the generality of the method in the practical application. Finally, the effectiveness of the theoretical results is demonstrated by two simulations.

This article studies the constrained optimization problems in the quaternion regime via a distributed fashion. We begin with presenting some differences for the generalized gradient between the real and quaternion domains. Then, an algorithm for the considered optimization problem is given, by which the desired optimization problem is transformed into an unconstrained setup. Using the tools from the Lyapunov-based technique and nonsmooth analysis, the convergence property associated with the devised algorithm is further guaranteed. In addition, the designed algorithm has the potential for solving distributed neurodynamic optimization problems as a recurrent neural network. Finally, a numerical example involving machine learning is given to illustrate the efficiency of the obtained results.

This article considers global exponential synchronization almost surely (GES a.s.) for a class of switched discrete-time neural networks (DTNNs). The considered system switches from one mode to another according to transition probability (TP) and evolves with mode-dependent average dwell time (MDADT), i.e., TP-based MDADT switching, which is more practical than classical average dwell time (ADT) switching. The logarithmic quantization technique is utilized to design mode-dependent quantized output controllers (QOCs). Noticing that external perturbations are unavoidable, actuator fault (AF) is also considered. New Lyapunov-Krasovskii functionals and analytical techniques are developed to obtain sufficient conditions to guarantee the GES a.s. It is discovered that the TP matrix plays an important role in achieving the GES a.s., the upper bound of the dwell time (DT) of unsynchronized subsystems can be very large, and the lower bound of the DT of synchronized subsystems can be very small. An algorithm is given to design the control gains, and an optimal algorithm is provided for reducing conservatism of the given results. Numerical examples demonstrate the effectiveness and the merits of the theoretical analysis.

This article investigates the output regulation problem of probabilistic$k$-valued logical systems with delays by an intermittent control scheme. Two types of event-triggered control are designed via semi-tensor product (STP) of matrices. According to the algebraic state-space representation of probabilistic$k$-valued systems with delays, the problem is transformed into the existence of solutions of algebraic equations. Then we obtain the sufficient and necessary condition for the output regulation problem of probabilistic$k$-valued logical systems with delays. Two types of approaches are given to design the event-triggered control laws. Eventually, two exemplifications are provided to explain the effectiveness and utility of the acquired results.

This article addresses the investigation of sliding-mode control (SMC) for slow-sampling singularly perturbed systems (SPSs) with Markov jump parameters. As a new attempt, the SMC strategy is considered in the study of discrete-time Markov jump SPSs. Subsequently, in order to design a sliding-mode controller to ensure the stability of the proposed system, a novel integral sliding surface is constructed, and an SMC law is synthesized to ensure the reachability of the sliding surface. Through the utilization of Lyapunov stability and SMC theory, sufficient conditions are derived to ensure the state trajectories of the system are driven to a predefined sliding surface and the closed-loop sliding mode dynamics are stochastically stable. Finally, the applicability of the proposed SMC strategy is verified by a numerical example and a practical electric circuit model.

In this paper, the optimization of the finite-time consensus of a multiagent system is studied when the leader is disturbed by white noise. Different from the consensus problem itself, this paper is devoted to optimizing the performance of the consensus and the constructive method and martingale method are used to guarantee the effectiveness of the optimization. Another contribution is that the problem of the optimal control with the piecewise constant controller gains under a stochastic environment is solved, and the reformed dynamic principle and the reformed Hamilton–Jacobi–Bellman (HJB) partial differential equations (PDEs) are proposed by using the splitting method and Feynman–Kac formula. Furthermore, since the multiagent system often runs on a high-dimensional space, a heuristic method is proposed based on the derivation of the reformed HJB PDEs and the scalar of the computational complexity of the PDEs is equivalent to that of a system of ordinary differential equations with the same dimensions when the multiagent system can be described as a piecewise linear system. Finally, the heuristic method is compared with usual filtering methods for solving engineering problems by a numerical example and it is found that the heuristic method achieves higher precision in finite time.

Image compression is one of the essential methods of image processing. Its most prominent advantage is the significant reduction of image size allowing for more efficient storage and transfer. However, lossy compression is associated with the loss of some image details in favor of reducing its size. In compressed images, the deficiencies are manifested by noticeable defects in the form of artifacts; the most common are block artifacts, ringing effect, or blur. In this article, we propose three models of fully convolutional networks with different configurations and examine their abilities in reducing compression artifacts. In the experiments, we research the extent to which the results are improved for models that will process the image in a similar way to the compression algorithm, and whether the initialization with predefined filters would allow for better image reconstruction than developed solely during learning.

In this paper, we propose a recursive variant of the Parzen kernel density estimator (KDE) to track changes of dynamic density over data streams in a nonstationary environment. In stationary environments, well-established traditional KDE techniques have nice asymptotic properties. Their existing extensions to deal with stream data are mostly based on various heuristic concepts (losing convergence properties). In this paper, we study recursive KDEs, called recursive concept drift tracking KDEs, and prove their weak (in probability) and strong (with probability one) convergence, resulting in perfect tracking properties as the sample size approaches infinity. In three theorems and subsequent examples, we show how to choose the bandwidth and learning rate of a recursive KDE in order to ensure weak and strong convergence. The simulation results illustrate the effectiveness of our algorithm both for density estimation and classification over time-varying stream data.

This paper studies the fixed-time anti-synchronization (FTAS) of discontinuous reaction-diffusion neural networks (DRDNNs) with both time-varying coefficients and time delay. First, differential inclusion theory is used to deal with the influence caused by discontinuous activations. In addition, a new fixed-time convergence theorem is used to handle the time-varying coefficients. Second, a novel state-feedback control algorithm and integral state-feedback control algorithm are proposed to realize FTAS of DRDNNs. During the generalized (adaptive) pinning control strategy, a guideline is proposed to select neurons to pin the designed controller. Furthermore, we present several criteria on FTAS by using the generalized Lyapunov function method. Different from the traditional Lyapunov function with negative definite derivative, the derivative of the Lyapunov function can be positive in this paper. Finally, we give two numerical simulations to substantiate the merits of the obtained results.

In this brief, the robust control invariance problem of probabilistic Boolean control networks (PBCNs) is investigated by a class of event-triggered control (ETC), which is an intermittent control scheme in essential. By resorting to the semi-tensor product (STP) technique, a PBCN with ETC can be equivalently described in a form of an algebraic linear system. Based on which, a matrix testing condition is derived to judge whether the given set can be a robust ETC invariant set (RETCIS). Subsequently, a necessary and sufficient condition is developed for the existence of event-triggered controllers. Meanwhile, all feasible event-triggered controllers are designed for guaranteeing the given set to be an RETCIS. Finally, a biological example is employed to demonstrate the availability of theoretical results.

This paper proposes a quaternion-valued one-layer recurrent neural network approach to resolve constrained convex function optimization problems with quaternion variables. Leveraging the novel generalized Hamilton-real (GHR) calculus, the quaternion gradient-based optimization techniques are proposed to derive the optimization algorithms in the quaternion field directly rather than the methods of decomposing the optimization problems into the complex domain or the real domain. Via chain rules and Lyapunov theorem, the rigorous analysis shows that the deliberately designed quaternion-valued one-layer recurrent neural network stabilizes the system dynamics while the states reach the feasible region in finite time and converges to the optimal solution of the considered constrained convex optimization problems finally. Numerical simulations verify the theoretical results.

In the literature, the effects of switching with average dwell time (ADT), Markovian switching, and intermittent coupling on stability and synchronization of dynamic systems have been extensively investigated. However, all of them are considered separately because it seems that the three kinds of switching are different from each other. This article proposes a new concept to unify these switchings and considers global exponential synchronization almost surely (GES a.s.) in an array of neural networks (NNs) with mixed delays (including time-varying delay and unbounded distributed delay), switching topology, and stochastic perturbations. A general switching mechanism with transition probability (TP) and mode-dependent ADT (MDADT) (i.e., TP-based MDADT switching in this article) is introduced. By designing a multiple Lyapunov-Krasovskii functional and developing a set of new analytical techniques, sufficient conditions are obtained to ensure that the coupled NNs with the general switching topology achieve GES a.s., even in the case that there are both synchronizing and nonsynchronizing modes. Our results have removed the restrictive condition that the increment coefficients of the multiple Lyapunov-Krasovskii functional at switching instants are larger than one. As applications, the coupled NNs with Markovian switching topology and intermittent coupling are employed. Numerical examples are provided to demonstrate the effectiveness and the merits of the theoretical analysis.

In this paper, we present our novel approach to recommender systems based on a neuro-fuzzy approach. The neuro-fuzzy approach allows for deciding to recommend or not to recommend processed items for a user. By using it, we can understand the decision through analyzing rules of decision paths. Our method gives a possibility to learn and simulate users decisions based on their actions in our test environment. Finally, a rank list of top-rated items is delivered to the user based on simulated rank for each of them. We develop our AI framework to perform tests with the use of CUDA technology. Additionally, we develop a user interface in the form of a web application. It gives the possibility to perform simulations of real users. To compare our approach with a deep learning based method, we perform tests on the MovieLens 20M Dataset. It should be noted that the architecture of the data module of our system allowed for reasonably easy integration with MovieLens data.

In this paper, the method of concept drift detection in time-varying data stream mining is considered. The Restricted Boltzmann Machine (RBM) is proposed to be applied as a drift detector. The RBMs which are able to learn joint probability distributions of attribute values and their classes were taken into account. Properly learned they contain a compressed information about the underlying data distribution. The RBM learned on a part of the data stream can be used to determine possible changes in the data stream probability distribution. Two evaluation measures are applied as indicators of possible sudden or gradual changes: the reconstruction error and the free energy. In experiments conducted on synthetic datasets, both measures proved to be well suited for the task of concept drift detection.

One of the greatest challenges in data mining is related to processing and analysis of massive data streams. Contrary to traditional static data mining problems, data streams require that each element is processed only once, the amount of allocated memory is constant and the models incorporate changes of investigated streams. A vast majority of available methods have been developed for data stream classification and only a few of them attempted to solve regression problems, using various heuristic approaches. In this paper, we develop mathematically justified regression models working in a time-varying environment. More specifically, we study incremental versions of generalized regression neural networks, called IGRNNs, and we prove their tracking properties - weak (in probability) and strong (with probability one) convergence assuming various concept drift scenarios. First, we present the IGRNNs, based on the Parzen kernels, for modeling stationary systems under nonstationary noise. Next, we extend our approach to modeling time-varying systems under nonstationary noise. We present several types of concept drifts to be handled by our approach in such a way that weak and strong convergence holds under certain conditions. Finally, in the series of simulations, we compare our method with commonly used heuristic approaches, based on forgetting mechanism or sliding windows, to deal with concept drift. Finally, we apply our concept in a real life scenario solving the problem of currency exchange rates prediction.

The most popular tools for stream data mining are based on decision trees. In previous 15 years, all designed methods, headed by the very fast decision tree algorithm, relayed on Hoeffding's inequality and hundreds of researchers followed this scheme. Recently, we have demonstrated that although the Hoeffding decision trees are an effective tool for dealing with stream data, they are a purely heuristic procedure; for example, classical decision trees such as ID3 or CART cannot be adopted to data stream mining using Hoeffding's inequality. Therefore, there is an urgent need to develop new algorithms, which are both mathematically justified and characterized by good performance. In this paper, we address this problem by developing a family of new splitting criteria for classification in stationary data streams and investigating their probabilistic properties. The new criteria, derived using appropriate statistical tools, are based on the misclassification error and the Gini index impurity measures. The general division of splitting criteria into two types is proposed. Attributes chosen based on type- splitting criteria guarantee, with high probability, the highest expected value of split measure. Type- criteria ensure that the chosen attribute is the same, with high probability, as it would be chosen based on the whole infinite data stream. Moreover, in this paper, two hybrid splitting criteria are proposed, which are the combinations of single criteria based on the misclassification error and Gini index.

In recent years plenty of new algorithms for data stream classification were developed. The occurrence of different concept drift types in data streams turned out to be especially challenging. Much attention was paid to the ensemble methods because of their desired properties. However, the problem of deciding how many components should be stored in the ensemble is still an open issue. Therefore in this article we show a theoretically justified method of determining the proper ensemble size automatically. The performance of the proposed algorithm was experimentally tested and compared with other known methods.

In this paper, a new method for constructing decision trees for stream data is proposed. First a new splitting criterion based on the misclassification error is derived. A theorem is proven showing that the best attribute computed in considered node according to the available data sample is the same, with some high probability, as the attribute derived from the whole infinite data stream. Next this result is combined with the splitting criterion based on the Gini index. It is shown that such combination provides the highest accuracy among all studied algorithms.

Decision trees are the commonly applied tools in the task of data stream classification. The most critical point in decision tree construction algorithm is the choice of the splitting attribute. In majority of algorithms existing in literature the splitting criterion is based on statistical bounds derived for split measure functions. In this paper we propose a totally new kind of splitting criterion. We derive statistical bounds for arguments of split measure function instead of deriving it for split measure function itself. This approach allows us to properly use the Hoeffding's inequality to obtain the required bounds. Based on this theoretical results we propose the Decision Trees based on the Fractions Approximation algorithm (DTFA). The algorithm exhibits satisfactory results of classification accuracy in numerical experiments. It is also compared with other existing in literature methods, demonstrating noticeably better performance.

In this paper a method for nonparametric regression estimation in non-stationary environment is presented. The Parzen kernels are used to design the recursive general regression neural networks to track changes of non-stationary system under non-stationary noise. The probabilistic properties of the proposed method are investigated. Experimental results are presented and discussed.

Since the Hoeffding tree algorithm was proposed in the literature, decision trees became one of the most popular tools for mining data streams. The key point of constructing the decision tree is to determine the best attribute to split the considered node. Several methods to solve this problem were presented so far. However, they are either wrongly mathematically justified (e.g., in the Hoeffding tree algorithm) or time-consuming (e.g., in the McDiarmid tree algorithm). In this paper, we propose a new method which significantly outperforms the McDiarmid tree algorithm and has a solid mathematical basis. Our method ensures, with a high probability set by the user, that the best attribute chosen in the considered node using a finite data sample is the same as it would be in the case of the whole data stream.

In mining data streams the most popular tool is the Hoeffding tree algorithm. It uses the Hoeffding's bound to determine the smallest number of examples needed at a node to select a splitting attribute. In the literature the same Hoeffding's bound was used for any evaluation function (heuristic measure), e.g., information gain or Gini index. In this paper, it is shown that the Hoeffding's inequality is not appropriate to solve the underlying problem. We prove two theorems presenting the McDiarmid's bound for both the information gain, used in ID3 algorithm, and for Gini index, used in Classification and Regression Trees (CART) algorithm. The results of the paper guarantee that a decision tree learning system, applied to data streams and based on the McDiarmid's bound, has the property that its output is nearly identical to that of a conventional learner. The results of the paper have a great impact on the state of the art of mining data streams and various developed so far methods and algorithms should be reconsidered.

Most methods constituting the soft computing concept can not handle data with missing or unknown feature values. Neural networks are able to perfectly fit to data and fuzzy logic systems use interpretable knowledge. In the paper we incorporate rough set theory to neuro-fuzzy system of very specific type. This results in learning systems which can work when the set number of available feature values is changing. To achieve better accuracy learning systems can be combined into larger ensembles. In the paper the AdaBoost metalearning is used to create an ensemble of learning systems. The rough-neuro-fuzzy systems use knowledge comprised in the form of fuzzy rules to perform classification. Simulations on a well-known benchmark give legitimacy to use the method in real world applications.

In this paper, a novel approach towards self-generating fuzzy neural network (SGFNN) is proposed. The proposed approach is simple and effective and is able to generate a fuzzy neural network with high accuracy and compact structure. The structure learning algorithm of the proposed SGFNN combines criteria of rule generation with a pruning technology. The Kalman filter (KF) algorithm is used to adjust the consequent parameters of the SGFNN. The SGFNN is applied for function approximation, nonlinear system identification and time-series prediction problems. Simulation results and comparative studies with other algorithms demonstrate that a more compact architecture with high performance can be obtained by the proposed approach.

In this paper, the viability of using Fuzzy-Rule-Based Regression Modeling (FRM) algorithm for tool performance and degradation detection is investigated. The FRM is developed based on a multi-layered fuzzy-rule-based hybrid system with Multiple Regression Models (MRM) embedded into a fuzzy logic inference engine that employs Self Organizing Maps (SOM) for clustering. The FRM converts a complex nonlinear problem to a simplified linear format in order to further increase the accuracy in prediction and rate of convergence. The efficacy of the proposed FRM is tested through a case study - namely to predict the remaining useful life of a ball nose milling cutter during a dry machining process of hardened tool steel with a hardness of 52-54 HRc. A comparative study is further made between four predictive models using the same set of experimental data. It is shown that the FRM is superior as compared with conventional MRM, Back Propagation Neural Networks (BPNN) and Radial Basis Function Networks (RBFN) in terms of prediction accuracy and learning speed.

In the paper the evolutionary strategy (μ, λ) is applied for learning flexible neuro-fuzzy systems [13]–[15]. In the process of evolution we determine: (i) fuzzy inference (Mamdani type or logical type - described by an S-implication), (ii) concrete fuzzy implication, if the logical type system is found in the process of evolution or concrete t-norm connecting antecedents and consequences, if the Mamdani type system is found in the process of evolution, (iii) concrete t-norm for aggregation of antecedents in each rule, (iv) concrete triangular norm describing aggregation operator, (v) shapes and parameters of fuzzy membership functions, (vi) weights describing importance of antecedents of rules and weights describing importance of rules, (vii) parameters of adjustable triangular norms. It should be noted that the crossover and mutation operators are chosen in a self-adaptive way. The method is tested using well known benchmarks.

Most methods constituting the soft computing concept can not handle data with missing or unknown features. Neural networks are able to perfectly fit to data and fuzzy logic systems use interpretable knowledge. To achieve better accuracy learning systems can be combined into larger ensembles. In this paper we combine logical neuro-fuzzy systems into the AdaBoost ensemble and extract fuzzy rules from the ensemble. The rules are used in rough-neuro-fuzzy classifier which can operate on data with missing values. The rough systems perform very well on these rules which was illustrated on a well known benchmark. The features were being removed to check the performance on incomplete data sets.

Takagi-Sugeno fuzzy systems are very common learning systems. The paper is about building classification ensembles from them and merging resulting rule bases. When merged, the rule base is more intelligible and easier to process. The merging is possible thanks to a modification of TS systems. Numerical simulations show that the modified systems perform very well

In the paper we develop a new method for designing and reduction of neuro-fuzzy systems. The method is based on the concept of the weighted triangular norms. In subsequent stages we reduce number of inputs, number of rules and number of antecedents. The method is tested using well known benchmarks.

Classification accuracy is nearly always improved after combining many systems. One of the most popular methods of multiple classification is boosting. In the paper we develop a method for merging fuzzy rule bases of neuro-fuzzy systems constituting an ensemble trained by the boosting algorithm.

Boosting is a method for learning combined classifiers. In a boosting ensemble of classifiers trained by the backpropagation algorithm, the learning rate takes much smaller value comparing with the backpropagation applied alone. We propose a method which overcomes the above drawback and test it on neuro-fuzzy systems constituting a classifier ensemble using some well known benchmarks.

In the paper, a new class of Takagi-Sugeno fuzzy systems is derived. Various parameters and weights are incorporated into construction of such systems. The approach presented in the paper introduces more flexibility to the structure and design of neuro-fuzzy systems.

We introduce a new class of operators called quasi-triangular norms. They are denoted by H and parameterized by a parameter /spl nu/:H(a/sub 1/,a/sub 2/,...,a/sub n/;/spl nu/). From the construction of function H, it follows that it becomes a t-norm for /spl nu/=0 and a dual t-conorm for /spl nu/=1. For /spl nu/ close to 0, function H resembles a t-norm and for /spl nu/ close to 1, it resembles a t-conorm. In the paper, we also propose adjustable quasi-implications and a new class of neuro-fuzzy systems. Most neuro-fuzzy systems proposed in the past decade employ "engineering implications" defined by a t-norm as the minimum or product. In our proposition, a quasi-implication I(a,b;/spl nu/) varies from an "engineering implication" T{a,b} to corresponding S-implication as /spl nu/ goes from 0 to 1. Consequently, the structure of neuro-fuzzy systems presented in This work is determined in the process of learning. Learning procedures are derived and simulation examples are presented.

Most neuro-fuzzy systems proposed in the past decade employ "engineering implications" defined by a t-norm, e.g. the minimum or the product. We apply a new class of operators called quasi-triangular norms for the construction of neuro-fuzzy systems. These operators depend on a certain parameter /spl nu/ and change their functional forms between a t-norm and a t-conorm. Consequently, the structure of neuro-fuzzy systems presented in the paper is determined in the process of learning. Learning procedures are derived and simulation examples are presented.

The current state of knowledge regarding nonstationary processes is significantly poorer then in the case of stationary signals. In many applications, signals are treated as stationary only because in this way it is easier to analyze them; in fact, they are nonstationary. Nonstationary processes are undoubtedly more difficult to analyze and their diversity makes application of universal tools impossible. In this paper we propose a new class of generalized regression neural networks working in nonstationary environment. The generalized regession neural networks (GRNN) studied in this paper are able to follow changes of the best model, i.e., time-varying regression functions. The novelty is summarized as follows: 1) We present adaptive GRNN tracking time-varying regression functions. 2) We prove convergence of the GRNN based on general learning theorems presented in Section IV. 3) We design in detail special GRNN based on the Parzen and orthogonal series kernels. In each case we precise conditions ensuring convergence of the GRNN to the best models described by regression function. 4) We investigate speed of convergence of the GRNN and compare performance of specific structures based on the Parzen kernel and orthogonal series kernel. 5) We study various nonstationarities (multiplicative, additive, "scale change," "movable argument") and design in each case the GRNN based on the Parzen kernel and orthogonal series kernel.

In this paper, we propose a new class of probabilistic neural networks (PNNs) working in nonstationary environment. The novelty is summarized as follows: 1) We formulate the problem of pattern classification in nonstationary environment as the prediction problem and design a probabilistic neural network to classify patterns having time-varying probability distributions. We note that the problem of pattern classification in the nonstationary case is closely connected with the problem of prediction because on the basis of a learning sequence of the length n, a pattern in the moment n + k, k > or = 1 should be classified. 2) We present, for the first time in literature, definitions of optimality of PNNs in time-varying environment. Moreover, we prove that our PNNs asymptotically approach the Bayes-optimal (time-varying) decision surface. 3) We investigate the speed of convergence of constructed PNNs. 4) We design in detail PNNs based on Parzen kernels and multivariate Hermite series.

In this paper, we present a neuro-fuzzy structure of the hierarchical prioritized structure (HPS) proposed by Yager. The HPS allows for easy hierarchization of a fuzzy rule-base. Our neuro-fuzzy system can be learned by the backpropagation algorithm and is relatively computationally efficient.

In this paper, we derive new neuro-fuzzy structures called flexible neuro-fuzzy inference systems or FLEXNFIS. Based on the input-output data, we learn not only the parameters of the membership functions but also the type of the systems (Mamdani or logical). Moreover, we introduce: 1) softness to fuzzy implication operators, to aggregation of rules and to connectives of antecedents; 2) certainty weights to aggregation of rules and to connectives of antecedents; and 3) parameterized families of T-norms and S-norms to fuzzy implication operators, to aggregation of rules and to connectives of antecedents. Our approach introduces more flexibility to the structure and design of neuro-fuzzy systems. Through computer simulations, we show that Mamdani-type systems are more suitable to approximation problems, whereas logical-type systems may be preferred for classification problems.

Presents a general approach to neuro fuzzy systems. It includes both Mamdani (constructive) and logical (destructive) fuzzy inference. Moreover, a new class of fuzzy inference (and corresponding neuro-fuzzy systems) is introduced.

A new algorithm for image compression, named predictive vector quantization (PVQ), is developed based on competitive neural networks and optimal linear predictors. The semi-closed-loop PVQ methodology is suggested. The experimental results are presented and the performance of the algorithm is discussed.

A nonparametric algorithm, based on multiple Fourier series, for identification of MISO (multidimensional-input-single-output) systems is investigated. It is shown that the algorithm converges for a wide class of noise.>

The methodology for nonparametric fitting of multivariate functions is proposed. The mean-square error convergence and the strong convergence are proved.

Van Ryzin and Greblicki showed that pattern recognition procedures derived from orthogonal series estimates of a probability density function are Bayes risk consistent. In this note it is proved that these procedures do not lose-under some additional conditions-their asymptotic properties even if the random environment is nonstationary.