Robustly stable multi-step-ahead model predictive control (MPC) based on parallel support vector machines (SVMs) with linear kernel was proposed. First, an analytical solution of optimal control laws of parallel SVMs ...Robustly stable multi-step-ahead model predictive control (MPC) based on parallel support vector machines (SVMs) with linear kernel was proposed. First, an analytical solution of optimal control laws of parallel SVMs based MPC was derived, and then the necessary and sufficient stability condition for MPC closed loop was given according to SVM model, and finally a method of judging the discrepancy between SVM model and the actual plant was presented, and consequently the constraint sets, which can guarantee that the stability condition is still robust for model/plant mismatch within some given bounds, were obtained by applying small-gain theorem. Simulation experiments show the proposed stability condition and robust constraint sets can provide a convenient way of adjusting controller parameters to ensure a closed-loop with larger stable margin.展开更多
This article investigates the robust current tracking control problem of three-phase grid-connected inverters with LCL filter under external disturbance by a dynamic state feedback control method.First,this paper cons...This article investigates the robust current tracking control problem of three-phase grid-connected inverters with LCL filter under external disturbance by a dynamic state feedback control method.First,this paper constructs an internal model to learn the information of the states and input of the grid-connected inverter under steady state.Second,by utilizing the internal model principle,the paper turns the tracking control problem into the robust stabilization control problem based on some appropriate coordinate transformations.Then,The paper designs a dynamics state feedback control law to deal with this robust stabilization problem,and thus the solution of the robust current tracking control problem of three-phase grid-connected inverters can be obtained.This control method can ensure the asymptotic stability of the closedloop system.Finally,the paper illustrates the effectiveness of the proposed control approach through several groups of simulations,and compares it with the feedforward control method to verify the robustness of the proposed control method to uncertain parameters.展开更多
This paper studies cooperative robust parallel operation of multiple actuators over an undirected communication graph.The plant is modeled as an uncertain linear system,and the actuators are linear and identical.Based...This paper studies cooperative robust parallel operation of multiple actuators over an undirected communication graph.The plant is modeled as an uncertain linear system,and the actuators are linear and identical.Based on the internal model principle,a distributed dynamic output feedback control law is proposed to achieve both robust output regulation of the closed-loop system and plant input sharing among the actuators.A practical example of five motors cooperatively driving an uncertain shaft under an external load torque is presented to show the effectiveness of the proposed control law.展开更多
Reinforcement learning(RL),as an important branch of machine learning,has recently achieved extensive attention and success in many applications.Its main idea is to enable agents to continuously learn to make optimal ...Reinforcement learning(RL),as an important branch of machine learning,has recently achieved extensive attention and success in many applications.Its main idea is to enable agents to continuously learn to make optimal decisions by trying to maximize a reward function for their actions and interactions with the environment.However,making highquality decisions in complex and uncertain real-world scenarios is a challenging task.The interference and attacks in such scenarios tend to destroy the existing strategies.Maintaining RL's optimal performance in various cases and adapting to changing environments remains an important challenge.This article presents a comprehensive review of recent advancements in robust reinforcement learning(RRL),and analyzes them from the perspectives of challenges,methodologies,and applications.It systematically evaluates current progress in RRL and summarizes the commonly used benchmark platforms.Finally,several open challenges are discussed to stimulate further research and guide future developments in this area.展开更多
The performance of deep recommendation models degrades significantly under data poisoning attacks.While adversarial training methods such as Vulnerability-Aware Training(VAT)enhance robustness by injecting perturbatio...The performance of deep recommendation models degrades significantly under data poisoning attacks.While adversarial training methods such as Vulnerability-Aware Training(VAT)enhance robustness by injecting perturbations into embeddings,they remain limited by coarse-grained noise and a static defense strategy,leaving models susceptible to adaptive attacks.This study proposes a novel framework,Self-Purification Data Sanitization(SPD),which integrates vulnerability-aware adversarial training with dynamic label correction.Specifically,SPD first identifies high-risk users through a fragility scoring mechanism,then applies self-purification by replacing suspicious interactions with model-predicted high-confidence labels during training.This closed-loop process continuously sanitizes the training data and breaks the protection ceiling of conventional adversarial training.Experiments demonstrate that SPD significantly improves the robustness of both Matrix Factorization(MF)and LightGCN models against various poisoning attacks.We show that SPD effectively suppresses malicious gradient propagation and maintains recommendation accuracy.Evaluations on Gowalla and Yelp2018 confirmthat SPD-trainedmodels withstandmultiple attack strategies—including Random,Bandwagon,DP,and Rev attacks—while preserving performance.展开更多
Urbanization is a significant driver of the loss of biodiversity and the disruption of ecosystems.Amphibians are especially vulnerable to the negative impact of urbanization as their life cycles and habitat requiremen...Urbanization is a significant driver of the loss of biodiversity and the disruption of ecosystems.Amphibians are especially vulnerable to the negative impact of urbanization as their life cycles and habitat requirements are complex.The present study investigated the effects of urbanization on amphibian predation networks in suburban Kunming in Yunnan,China and aimed to understand how predation network structure and stability vary with urbanization level.We constructed predation networks by analyzing the stomach contents of amphibians from 12d istinct urbanization gradients.We used the bipartite package in R to evaluate network robustness metrics such as modularity,nestedness,connectivity,and average shortest path length(ASPL).We found that urbanization level is negatively correlated with predation network connectivity(R=−0.67,Ρ=0.02),but there were no significant correlations between urbanization level and nestedness,modularity,or ASPL.Removal of the keystone species destabilized the predation networks at certain locations.The present work highlighted that maintaining prey quantity and diversity preserves predation network connectivity and stabilizes the overall network in urbanizing landscapes.It also underscored the critical role that keystone species play in sustaining network robustness.The results of this research provided insights into the ecological consequences of urbanization.They also suggested that conservation measures should protect the key species and habitats of amphibian predation networks and mitigate the negative impact of urban development on them.展开更多
With the growing global energy demand and the pressing need for a clean energy transition,supercapacitors(SCs)have demonstrated significant application potential in electric vehicles,wearable electronics,and renewable...With the growing global energy demand and the pressing need for a clean energy transition,supercapacitors(SCs)have demonstrated significant application potential in electric vehicles,wearable electronics,and renewable energy storage systems owing to their rapid charge-discharge capability,exceptional power density,and prolonged cycle life.The improvement of their overall performance fundamentally depends on the synergistic design of electrode materials and electrolyte systems,as well as the precise regulation of the electrode-electrolyte interface.This review focuses on the key components of supercapacitors,systematically reviewing the design strategies of high-performance electrode materials,outlining recent advances in novel electrolyte systems,and comprehensively discussing the critical roles of interfacial reinforcement and optimization in enhancing device energy density,power performance,and cycling stability.Furthermore,interfacial engineering strategies and innovations in device architecture are proposed to address interfacial degradation in flexible SCs under mechanical stress.Finally,key future research directions are highlighted,including the development of high-voltage and wide-temperature-range electrolyte systems and the integrated advancement of multiscale in situ characterization techniques and theoretical modeling.This review aims to provide theoretical guidance and innovative strategies for material design,contributing toward the realization of next-generation supercapacitors with enhanced energy density and reliability.展开更多
Development of lightweight and strong structural material using fast-growing poplar wood is promising for green and sustainable engineering.Herein,the overall performances of fast-growing natural poplar wood(NPW)are s...Development of lightweight and strong structural material using fast-growing poplar wood is promising for green and sustainable engineering.Herein,the overall performances of fast-growing natural poplar wood(NPW)are significantly enhanced via delignification,in situ growth of SiO_(2)followed by densification.The SiO_(2)/compresseddelignified-wood(SiO_(2)/CDW)nanocomposite obtained exhibits outstanding mechanical properties including a bending strength of 395.6 MPa,a tensile strength of 253.4 MPa,and a toughness of 7.1 MJ/m^(3),which is improved by 1548%,240%and 590%,respectively compared with NPW.In addition,the ignition time and burning time of SiO_(2)/CDW nanocomposite are prolonged by 700%and 112%compared to those of NPW.Moreover,the specific wear rate of SiO_(2)/CDW is 18×10^(-6)mm^(3)/Nm,which is 72.6%lower than that of NPW.Moreover,the spring-back ratios of SiO_(2)/CDW in 95%and in water are 45.2%and 66.7%,which are lower than those of CDW(64.6%and 92.4%).The SiO_(2)/CDW nanocomposite with enhanced mechanical,flame/water retardant and wear performances are promising to meet the needs of modern engineering as green and sustainable materials.展开更多
This paper proposes a robust control-oriented identification method for errors-in-variables(EIV)systems in output feedbacks using frequency-response(FR)experimental data.An important relation between such a closed-loo...This paper proposes a robust control-oriented identification method for errors-in-variables(EIV)systems in output feedbacks using frequency-response(FR)experimental data.An important relation between such a closed-loop EIV system and its coprime factor(CF)uncertainty description is first derived,based on which the FR measurements suitable for plant CF identification are able to be generated.Different factorizations of a given controller in the closed-loop system can be made best use to adjust right coprime factors(RCFs)of the plant so as to realize an improvement on the signal-to-noise ratio of identification experimental data.Subsequently,a nominal RCF model is estimated by linear matrix inequalities from the applicable FR measurements and its associated worst-case errors are quantified from a priori and a posteriori information on the underlying system.A resulting RCF perturbation model set can then be described by the nominal RCF model and its worst-case error bounds.Such a model set capable of being stabilized by the given controller is ready for its robust stabilizing controller redesign and robust performance analysis.Finally,a numerical simulation is given to show the efficacy of the proposed identification method.展开更多
Evaluating the adversarial robustness of classification algorithms in machine learning is a crucial domain.However,current methods lack measurable and interpretable metrics.To address this issue,this paper introduces ...Evaluating the adversarial robustness of classification algorithms in machine learning is a crucial domain.However,current methods lack measurable and interpretable metrics.To address this issue,this paper introduces a visual evaluation index named confidence centroid skewing quadrilateral,which is based on a classification confidence-based confusion matrix,offering a quantitative and visual comparison of the adversarial robustness among different classification algorithms,and enhances intuitiveness and interpretability of attack impacts.We first conduct a validity test and sensitive analysis of the method.Then,prove its effectiveness through the experiments of five classification algorithms including artificial neural network(ANN),logistic regression(LR),support vector machine(SVM),convolutional neural network(CNN)and transformer against three adversarial attacks such as fast gradient sign method(FGSM),DeepFool,and projected gradient descent(PGD)attack.展开更多
In this paper,we consider a robust semi-infinite interval-valued optimization problem with inequality constraints having an uncertain parameter.The parametric representation of the aforesaid problem is also considered...In this paper,we consider a robust semi-infinite interval-valued optimization problem with inequality constraints having an uncertain parameter.The parametric representation of the aforesaid problem is also considered in order to derive the necessary and sufficient optimality conditions.Furthermore,we formulate a mixed-type dual problem and derive duality results which associate the robust weak efficient solution of the primal and its dual problems.Several examples are given to illustrate the results in the manuscript.展开更多
Cognitive unmanned aerial vehicle(UAV)is promising to tackle the spectrum scarcity problem faced by UAV communications.However,the secure information transmission is challenging due to the open nature of the spectrum ...Cognitive unmanned aerial vehicle(UAV)is promising to tackle the spectrum scarcity problem faced by UAV communications.However,the secure information transmission is challenging due to the open nature of the spectrum sharing.In order to tackle this issue,a cognitive UAV network with cooperative jamming is studied in this paper.A robust resource allocation and trajectory joint optimization problem is formulated by considering the practical case that the channel state information(CSI)cannot be accurately obtained.An iterative algorithm is proposed to address this challenging non-convex problem.Simulation results demonstrate that the worst case robust resource allocation design can realize the secure communications even under the imperfect CSI.Moreover,compared with other benchmark schemes,the proposed scheme can achieve secure performance improvement.展开更多
Vehicle re-identification(ReID)is a challenging task in intelligent transportation,and urban surveillance systems due to its complications in camera viewpoints,vehicle scales,and environmental conditions.Recent transf...Vehicle re-identification(ReID)is a challenging task in intelligent transportation,and urban surveillance systems due to its complications in camera viewpoints,vehicle scales,and environmental conditions.Recent transformer-based approaches have shown impressive performance by utilizing global dependencies,these models struggle with aspect ratio distortions and may overlook fine-grained local attributes crucial for distinguishing visually similar vehicles.We introduce a framework based on Swin Transformers that addresses these challenges by implementing three components.First,to improve feature robustness and maintain vehicle proportions,our Aspect Ratio-Aware Swin Transformer(AR-Swin)preserve the native ratio via letterbox,uses a non-square(16×8)patch-embedding stem,and keeps fixed 7×7 token windows.Second,we introduce a Dynamic Feature Fusion Network(DFFNet)that adaptively integrates global Swin features with local attribute embeddings;such as color and vehicle type enablingmore discriminative representations.Third,our Regional Attention Blocks incorporate regionalmasks into the transformer’s windowed attentionmechanism,effectively highlighting critical details like manufacturer logos or lights.On VeRi-776,we obtain 82.55 mAP,97.26 Rank-1 and 99.23 Rank-5,and on VehicleID we obtain 91.8 Rank-1 and 97.75 Rank-5.The design is drop-in for Swin backbones and emphasizes robustness without increasing architectural complexity.Code:https://github.com/sft110/Swinvreid.展开更多
As a cornerstone for applications such as autonomous driving,3D urban perception is a burgeoning field of study.Enhancing the performance and robustness of these perception systems is crucial for ensuring the safety o...As a cornerstone for applications such as autonomous driving,3D urban perception is a burgeoning field of study.Enhancing the performance and robustness of these perception systems is crucial for ensuring the safety of next-generation autonomous vehicles.In this work,we introduce a novel neural scene representation called Street Detection Gaussians(SDGs),which redefines urban 3D perception through an integrated architecture unifying reconstruction and detection.At its core lies the dynamic Gaussian representation,where time-conditioned parameterization enables simultaneous modeling of static environments and dynamic objects through physically constrained Gaussian evolution.The framework’s radar-enhanced perception module learns cross-modal correlations between sparse radardata anddense visual features,resulting ina22%reduction inocclusionerrors compared tovisiononly systems.A breakthrough differentiable rendering pipeline back-propagates semantic detection losses throughout the entire 3D reconstruction process,enabling the optimization of both geometric and semantic fidelity.Evaluated on the Waymo Open Dataset and the KITTI Dataset,the system achieves real-time performance(135 Frames Per Second(FPS)),photorealistic quality(Peak Signal-to-Noise Ratio(PSNR)34.9 dB),and state-of-the-art detection accuracy(78.1%Mean Average Precision(mAP)),demonstrating a 3.8×end-to-end improvement over existing hybrid approaches while enabling seamless integration with autonomous driving stacks.展开更多
The Nelder-Mead simplex method is a well-known algorithm enabling the minimization of functions that are not available in closed-form and that need not be differentiable or convex.Furthermore,it is particularly parsim...The Nelder-Mead simplex method is a well-known algorithm enabling the minimization of functions that are not available in closed-form and that need not be differentiable or convex.Furthermore,it is particularly parsimonious on the number of function evaluations,thus making it preferable to convex optimization paradigms in the case,common when dealing with control design problems,that the objective function of the optimization problem is non-differentiable,non-convex,and its closed-form is not available or difficult to be computed analytically.The main goal of this paper is to show how the joint use of the Nelder-Mead simplex method and the Morrison algorithm can be successfully used to solve relevant and challenging control problems that cannot be easily solved using analytic methods.In particular,it is shown how the problems of strong stabilization,static output feedback stabilization,and design of robust controllers having fixed structure can be framed as optimization problems,which,in turn,can be efficiently solved by coupling the two above mentioned algorithms.The performance of this procedure is compared with state-of-the-art techniques on dozens of static output feedback benchmark case studies,and its effectiveness is demonstrated by several examples.展开更多
Mesh models are among the primary representations for storing 3-D objects,encapsulating detailed geometric information.3-D mesh watermarking,in particular,plays a central role in the protection of 3-D content.However,...Mesh models are among the primary representations for storing 3-D objects,encapsulating detailed geometric information.3-D mesh watermarking,in particular,plays a central role in the protection of 3-D content.However,frequency-domain methods rely on complex parameterization and spectral decomposition,which are sensitive to mesh topology and resolution and often introduce perceptible artifacts.Spatial-domain techniques,on the other hand,typically embed watermarks in global or randomly selected regions,leading to visible distortions and reduced robustness.To address the above limitations and protect model copyright without compromising the original aesthetic quality,we propose a deterministice PCA-synchronized 3Dmeshwatermarkingmethodwith fullerene-guided carrier selection.First,a deterministic principal component analysis(PCA)-based mesh synchronization algorithm is employed to align the models to a canonical pose.Next,a fullerene-inspired carrier selection strategy is employed to determine the watermark carriers,leveraging the structural characteristics of fullerene molecules to achieve a more rational and effective carrier selection.Finally,to balance the embedding strength and enhance visual quality,the watermark information is embedded using an APQIM(Adaptive Parity-Check Quantization Index Modulation)scheme.The experimental results show that our method can achieve high visual quality with scalable capacity and strong robustness compared with existing methods.The watermarking scheme can resist various attacks,including simplification,smoothing,Gaussian noise,translation,and rotation.展开更多
This paper develops an advanced framework for the operational optimization of integrated multi-energy systems that encompass electricity,gas,and heating networks.Introducing a cutting-edge stochastic gradient-enhanced...This paper develops an advanced framework for the operational optimization of integrated multi-energy systems that encompass electricity,gas,and heating networks.Introducing a cutting-edge stochastic gradient-enhanced distributionally robust optimization approach,this study integrates deep learning models,especially generative adversarial networks,to adeptly handle the inherent variability and uncertainties of renewable energy and fluctuating consumer demands.The effectiveness of this framework is rigorously tested through detailed simulations mirroring real-world urban energy consumption,renewable energy production,and market price fluctuations over an annual period.The results reveal substantial improvements in the resilience and efficiency of the grid,achieving a reduction in power distribution losses by 15%and enhancing voltage stability by 20%,markedly outperforming conventional systems.Additionally,the framework facilitates up to 25%in cost reductions during peak demand periods,significantly lowering operational costs.The adoption of stochastic gradients further refines the framework’s ability to continually adjust to real-time changes in environmental and market conditions,ensuring stable grid operations and fostering active consumer engagement in demand-side management.This strategy not only aligns with contem-porary sustainable energy practices but also provides scalable and robust solutions to pressing challenges in modern power network management.展开更多
The outstanding growth in the applications of large language models(LLMs)demonstrates the significance of adaptive and efficient prompt engineering tactics.The existing methods may not be variable,vigorous and streaml...The outstanding growth in the applications of large language models(LLMs)demonstrates the significance of adaptive and efficient prompt engineering tactics.The existing methods may not be variable,vigorous and streamlined in different domains.The offered study introduces an immediate optimization outline,named PROMPTx-PE,that is going to yield a greater level of precision and strength when it comes to the assignments that are premised on LLM.The proposed systemfeatures a timely selection schemewhich is informed by reinforcement learning,a contextual layer and a dynamic weighting module which is regulated by Lyapunov-based stability guidelines.The PROMPTx-PE dynamically varies the exploration and exploitation of the prompt space,depending on real-time feedback and multi-objective reward development.Extensive testing on both benchmark(GLUE,SuperGLUE)and domain-specific data(Healthcare-QA and Industrial-NER)demonstrates a large best performance to be 89.4%and a strong robustness disconnect with under 3%computation expense.The results confirm the effectiveness,consistency,and scalability of PROMPTx-PE as a platform of adaptive prompt engineering based on recent uses of LLMs.展开更多
This paper suggests a way to improve teamwork and reduce uncertainties in operations by using a game theory approach involving multiple virtual power plants(VPP).A generalized credibility-based fuzzy chance constraint...This paper suggests a way to improve teamwork and reduce uncertainties in operations by using a game theory approach involving multiple virtual power plants(VPP).A generalized credibility-based fuzzy chance constraint programming approach is adopted to address uncertainties stemming from renewable generation and load demand within individual VPPs,while robust optimization techniques manage electricity and thermal price volatilities.Building upon this foundation,a hierarchical Nash-Stackelberg game model is established across multiple VPPs.Within each VPP,a Stackelberg game resolves the strategic interaction between the operator and photovoltaic prosumers(PVP).Among VPPs,a cooperative Nash bargaining model coordinates alliance formation.The problem is decomposed into two subproblems:maximizing coalitional benefits,and allocating cooperative surpluses via payment bargaining,solved distributively using the alternating direction method of multipliers(ADMM).Case studies demonstrate that the proposed strategy significantly enhances the economic efficiency and uncertainty resilience of multi-VPP alliances.展开更多
In multi-modal emotion recognition,excessive reliance on historical context often impedes the detection of emotional shifts,while modality heterogeneity and unimodal noise limit recognition performance.Existing method...In multi-modal emotion recognition,excessive reliance on historical context often impedes the detection of emotional shifts,while modality heterogeneity and unimodal noise limit recognition performance.Existing methods struggle to dynamically adjust cross-modal complementary strength to optimize fusion quality and lack effective mechanisms to model the dynamic evolution of emotions.To address these issues,we propose a multi-level dynamic gating and emotion transfer framework for multi-modal emotion recognition.A dynamic gating mechanism is applied across unimodal encoding,cross-modal alignment,and emotion transfer modeling,substantially improving noise robustness and feature alignment.First,we construct a unimodal encoder based on gated recurrent units and feature-selection gating to suppress intra-modal noise and enhance contextual representation.Second,we design a gated-attention crossmodal encoder that dynamically calibrates the complementary contributions of visual and audio modalities to the dominant textual features and eliminates redundant information.Finally,we introduce a gated enhanced emotion transfer module that explicitly models the temporal dependence of emotional evolution in dialogues via transfer gating and optimizes continuity modeling with a comparative learning loss.Experimental results demonstrate that the proposed method outperforms state-of-the-art models on the public MELD and IEMOCAP datasets.展开更多
基金Project(2002CB312200) supported by the National Key Fundamental Research and Development Program of China project(60574019) supported by the National Natural Science Foundation of China
文摘Robustly stable multi-step-ahead model predictive control (MPC) based on parallel support vector machines (SVMs) with linear kernel was proposed. First, an analytical solution of optimal control laws of parallel SVMs based MPC was derived, and then the necessary and sufficient stability condition for MPC closed loop was given according to SVM model, and finally a method of judging the discrepancy between SVM model and the actual plant was presented, and consequently the constraint sets, which can guarantee that the stability condition is still robust for model/plant mismatch within some given bounds, were obtained by applying small-gain theorem. Simulation experiments show the proposed stability condition and robust constraint sets can provide a convenient way of adjusting controller parameters to ensure a closed-loop with larger stable margin.
基金Supported by the Fundamental Research Funds for the Central Universities(2024ZYGXZR047)the National Natural Science Foundation of China(62373156)the Guangdong Basic and Applied Basic Research Foundation(2024A1515011736)。
文摘This article investigates the robust current tracking control problem of three-phase grid-connected inverters with LCL filter under external disturbance by a dynamic state feedback control method.First,this paper constructs an internal model to learn the information of the states and input of the grid-connected inverter under steady state.Second,by utilizing the internal model principle,the paper turns the tracking control problem into the robust stabilization control problem based on some appropriate coordinate transformations.Then,The paper designs a dynamics state feedback control law to deal with this robust stabilization problem,and thus the solution of the robust current tracking control problem of three-phase grid-connected inverters can be obtained.This control method can ensure the asymptotic stability of the closedloop system.Finally,the paper illustrates the effectiveness of the proposed control approach through several groups of simulations,and compares it with the feedforward control method to verify the robustness of the proposed control method to uncertain parameters.
基金Supported by the Shenzhen Key Laboratory of Control Theory and Intelligent Systems (ZDSYS20220330161800001)the National Natural Science Foundation of China (62303207)the Guangdong Basic and Applied Basic Research Foundation (2024A1515010725)。
文摘This paper studies cooperative robust parallel operation of multiple actuators over an undirected communication graph.The plant is modeled as an uncertain linear system,and the actuators are linear and identical.Based on the internal model principle,a distributed dynamic output feedback control law is proposed to achieve both robust output regulation of the closed-loop system and plant input sharing among the actuators.A practical example of five motors cooperatively driving an uncertain shaft under an external load torque is presented to show the effectiveness of the proposed control law.
文摘Reinforcement learning(RL),as an important branch of machine learning,has recently achieved extensive attention and success in many applications.Its main idea is to enable agents to continuously learn to make optimal decisions by trying to maximize a reward function for their actions and interactions with the environment.However,making highquality decisions in complex and uncertain real-world scenarios is a challenging task.The interference and attacks in such scenarios tend to destroy the existing strategies.Maintaining RL's optimal performance in various cases and adapting to changing environments remains an important challenge.This article presents a comprehensive review of recent advancements in robust reinforcement learning(RRL),and analyzes them from the perspectives of challenges,methodologies,and applications.It systematically evaluates current progress in RRL and summarizes the commonly used benchmark platforms.Finally,several open challenges are discussed to stimulate further research and guide future developments in this area.
文摘The performance of deep recommendation models degrades significantly under data poisoning attacks.While adversarial training methods such as Vulnerability-Aware Training(VAT)enhance robustness by injecting perturbations into embeddings,they remain limited by coarse-grained noise and a static defense strategy,leaving models susceptible to adaptive attacks.This study proposes a novel framework,Self-Purification Data Sanitization(SPD),which integrates vulnerability-aware adversarial training with dynamic label correction.Specifically,SPD first identifies high-risk users through a fragility scoring mechanism,then applies self-purification by replacing suspicious interactions with model-predicted high-confidence labels during training.This closed-loop process continuously sanitizes the training data and breaks the protection ceiling of conventional adversarial training.Experiments demonstrate that SPD significantly improves the robustness of both Matrix Factorization(MF)and LightGCN models against various poisoning attacks.We show that SPD effectively suppresses malicious gradient propagation and maintains recommendation accuracy.Evaluations on Gowalla and Yelp2018 confirmthat SPD-trainedmodels withstandmultiple attack strategies—including Random,Bandwagon,DP,and Rev attacks—while preserving performance.
基金supported by Yunnan Fundamental Research Projects(202501BD070001-081).
文摘Urbanization is a significant driver of the loss of biodiversity and the disruption of ecosystems.Amphibians are especially vulnerable to the negative impact of urbanization as their life cycles and habitat requirements are complex.The present study investigated the effects of urbanization on amphibian predation networks in suburban Kunming in Yunnan,China and aimed to understand how predation network structure and stability vary with urbanization level.We constructed predation networks by analyzing the stomach contents of amphibians from 12d istinct urbanization gradients.We used the bipartite package in R to evaluate network robustness metrics such as modularity,nestedness,connectivity,and average shortest path length(ASPL).We found that urbanization level is negatively correlated with predation network connectivity(R=−0.67,Ρ=0.02),but there were no significant correlations between urbanization level and nestedness,modularity,or ASPL.Removal of the keystone species destabilized the predation networks at certain locations.The present work highlighted that maintaining prey quantity and diversity preserves predation network connectivity and stabilizes the overall network in urbanizing landscapes.It also underscored the critical role that keystone species play in sustaining network robustness.The results of this research provided insights into the ecological consequences of urbanization.They also suggested that conservation measures should protect the key species and habitats of amphibian predation networks and mitigate the negative impact of urban development on them.
基金supported by the National Natural Science Foundation of China(Nos.52072208 and 52261160384)supported by the Postdoctoral Fellowship Program(Grade B)of China Postdoctoral Science Foundation under Grant Number GZB20250057China Postdoctoral Science Foundation(2025M770223).
文摘With the growing global energy demand and the pressing need for a clean energy transition,supercapacitors(SCs)have demonstrated significant application potential in electric vehicles,wearable electronics,and renewable energy storage systems owing to their rapid charge-discharge capability,exceptional power density,and prolonged cycle life.The improvement of their overall performance fundamentally depends on the synergistic design of electrode materials and electrolyte systems,as well as the precise regulation of the electrode-electrolyte interface.This review focuses on the key components of supercapacitors,systematically reviewing the design strategies of high-performance electrode materials,outlining recent advances in novel electrolyte systems,and comprehensively discussing the critical roles of interfacial reinforcement and optimization in enhancing device energy density,power performance,and cycling stability.Furthermore,interfacial engineering strategies and innovations in device architecture are proposed to address interfacial degradation in flexible SCs under mechanical stress.Finally,key future research directions are highlighted,including the development of high-voltage and wide-temperature-range electrolyte systems and the integrated advancement of multiscale in situ characterization techniques and theoretical modeling.This review aims to provide theoretical guidance and innovative strategies for material design,contributing toward the realization of next-generation supercapacitors with enhanced energy density and reliability.
基金the financial support from the National Natural Science Foundation of China(No.52303082)Natural Science Foundation of Hubei Province(No.2023AFB375)Fundamental Research Funds for Central Universities of China(No.2022CDJQY-004)。
文摘Development of lightweight and strong structural material using fast-growing poplar wood is promising for green and sustainable engineering.Herein,the overall performances of fast-growing natural poplar wood(NPW)are significantly enhanced via delignification,in situ growth of SiO_(2)followed by densification.The SiO_(2)/compresseddelignified-wood(SiO_(2)/CDW)nanocomposite obtained exhibits outstanding mechanical properties including a bending strength of 395.6 MPa,a tensile strength of 253.4 MPa,and a toughness of 7.1 MJ/m^(3),which is improved by 1548%,240%and 590%,respectively compared with NPW.In addition,the ignition time and burning time of SiO_(2)/CDW nanocomposite are prolonged by 700%and 112%compared to those of NPW.Moreover,the specific wear rate of SiO_(2)/CDW is 18×10^(-6)mm^(3)/Nm,which is 72.6%lower than that of NPW.Moreover,the spring-back ratios of SiO_(2)/CDW in 95%and in water are 45.2%and 66.7%,which are lower than those of CDW(64.6%and 92.4%).The SiO_(2)/CDW nanocomposite with enhanced mechanical,flame/water retardant and wear performances are promising to meet the needs of modern engineering as green and sustainable materials.
文摘This paper proposes a robust control-oriented identification method for errors-in-variables(EIV)systems in output feedbacks using frequency-response(FR)experimental data.An important relation between such a closed-loop EIV system and its coprime factor(CF)uncertainty description is first derived,based on which the FR measurements suitable for plant CF identification are able to be generated.Different factorizations of a given controller in the closed-loop system can be made best use to adjust right coprime factors(RCFs)of the plant so as to realize an improvement on the signal-to-noise ratio of identification experimental data.Subsequently,a nominal RCF model is estimated by linear matrix inequalities from the applicable FR measurements and its associated worst-case errors are quantified from a priori and a posteriori information on the underlying system.A resulting RCF perturbation model set can then be described by the nominal RCF model and its worst-case error bounds.Such a model set capable of being stabilized by the given controller is ready for its robust stabilizing controller redesign and robust performance analysis.Finally,a numerical simulation is given to show the efficacy of the proposed identification method.
文摘Evaluating the adversarial robustness of classification algorithms in machine learning is a crucial domain.However,current methods lack measurable and interpretable metrics.To address this issue,this paper introduces a visual evaluation index named confidence centroid skewing quadrilateral,which is based on a classification confidence-based confusion matrix,offering a quantitative and visual comparison of the adversarial robustness among different classification algorithms,and enhances intuitiveness and interpretability of attack impacts.We first conduct a validity test and sensitive analysis of the method.Then,prove its effectiveness through the experiments of five classification algorithms including artificial neural network(ANN),logistic regression(LR),support vector machine(SVM),convolutional neural network(CNN)and transformer against three adversarial attacks such as fast gradient sign method(FGSM),DeepFool,and projected gradient descent(PGD)attack.
基金supported by the MATRICES,SERB-DST,New Delhi,India(No.MTR/2021/000002).
文摘In this paper,we consider a robust semi-infinite interval-valued optimization problem with inequality constraints having an uncertain parameter.The parametric representation of the aforesaid problem is also considered in order to derive the necessary and sufficient optimality conditions.Furthermore,we formulate a mixed-type dual problem and derive duality results which associate the robust weak efficient solution of the primal and its dual problems.Several examples are given to illustrate the results in the manuscript.
基金National Key R&D Program of China under Grant 2020YFB1807602the National Natural Science Foundation of China under Grant 62222107,Grant 62071223,Grant 62031012Young Elite Scientist Sponsorship Program by CAST。
文摘Cognitive unmanned aerial vehicle(UAV)is promising to tackle the spectrum scarcity problem faced by UAV communications.However,the secure information transmission is challenging due to the open nature of the spectrum sharing.In order to tackle this issue,a cognitive UAV network with cooperative jamming is studied in this paper.A robust resource allocation and trajectory joint optimization problem is formulated by considering the practical case that the channel state information(CSI)cannot be accurately obtained.An iterative algorithm is proposed to address this challenging non-convex problem.Simulation results demonstrate that the worst case robust resource allocation design can realize the secure communications even under the imperfect CSI.Moreover,compared with other benchmark schemes,the proposed scheme can achieve secure performance improvement.
基金supported by SDAIA-KFUPM Joint Research Center of Artificial Intelligence,Deanship of Research,King Fahd University of Petroleum and Minerals,under Grant#CAI02562(JRC-AI-RFP-17).
文摘Vehicle re-identification(ReID)is a challenging task in intelligent transportation,and urban surveillance systems due to its complications in camera viewpoints,vehicle scales,and environmental conditions.Recent transformer-based approaches have shown impressive performance by utilizing global dependencies,these models struggle with aspect ratio distortions and may overlook fine-grained local attributes crucial for distinguishing visually similar vehicles.We introduce a framework based on Swin Transformers that addresses these challenges by implementing three components.First,to improve feature robustness and maintain vehicle proportions,our Aspect Ratio-Aware Swin Transformer(AR-Swin)preserve the native ratio via letterbox,uses a non-square(16×8)patch-embedding stem,and keeps fixed 7×7 token windows.Second,we introduce a Dynamic Feature Fusion Network(DFFNet)that adaptively integrates global Swin features with local attribute embeddings;such as color and vehicle type enablingmore discriminative representations.Third,our Regional Attention Blocks incorporate regionalmasks into the transformer’s windowed attentionmechanism,effectively highlighting critical details like manufacturer logos or lights.On VeRi-776,we obtain 82.55 mAP,97.26 Rank-1 and 99.23 Rank-5,and on VehicleID we obtain 91.8 Rank-1 and 97.75 Rank-5.The design is drop-in for Swin backbones and emphasizes robustness without increasing architectural complexity.Code:https://github.com/sft110/Swinvreid.
文摘As a cornerstone for applications such as autonomous driving,3D urban perception is a burgeoning field of study.Enhancing the performance and robustness of these perception systems is crucial for ensuring the safety of next-generation autonomous vehicles.In this work,we introduce a novel neural scene representation called Street Detection Gaussians(SDGs),which redefines urban 3D perception through an integrated architecture unifying reconstruction and detection.At its core lies the dynamic Gaussian representation,where time-conditioned parameterization enables simultaneous modeling of static environments and dynamic objects through physically constrained Gaussian evolution.The framework’s radar-enhanced perception module learns cross-modal correlations between sparse radardata anddense visual features,resulting ina22%reduction inocclusionerrors compared tovisiononly systems.A breakthrough differentiable rendering pipeline back-propagates semantic detection losses throughout the entire 3D reconstruction process,enabling the optimization of both geometric and semantic fidelity.Evaluated on the Waymo Open Dataset and the KITTI Dataset,the system achieves real-time performance(135 Frames Per Second(FPS)),photorealistic quality(Peak Signal-to-Noise Ratio(PSNR)34.9 dB),and state-of-the-art detection accuracy(78.1%Mean Average Precision(mAP)),demonstrating a 3.8×end-to-end improvement over existing hybrid approaches while enabling seamless integration with autonomous driving stacks.
基金partially supported by the Italian Ministry for Research in the framework of the 2020 Program for Research Projects of National Interest(2020RTWES4)。
文摘The Nelder-Mead simplex method is a well-known algorithm enabling the minimization of functions that are not available in closed-form and that need not be differentiable or convex.Furthermore,it is particularly parsimonious on the number of function evaluations,thus making it preferable to convex optimization paradigms in the case,common when dealing with control design problems,that the objective function of the optimization problem is non-differentiable,non-convex,and its closed-form is not available or difficult to be computed analytically.The main goal of this paper is to show how the joint use of the Nelder-Mead simplex method and the Morrison algorithm can be successfully used to solve relevant and challenging control problems that cannot be easily solved using analytic methods.In particular,it is shown how the problems of strong stabilization,static output feedback stabilization,and design of robust controllers having fixed structure can be framed as optimization problems,which,in turn,can be efficiently solved by coupling the two above mentioned algorithms.The performance of this procedure is compared with state-of-the-art techniques on dozens of static output feedback benchmark case studies,and its effectiveness is demonstrated by several examples.
基金supported by the National Natural Science Foundation of China(NSFC)under Grant 62272331the Key Laboratory of Data Protection and Intelligent Management,Ministry of Education,Sichuan University and the Fundamental Research Funds for the Central Universities under Grant SCU2023D008.
文摘Mesh models are among the primary representations for storing 3-D objects,encapsulating detailed geometric information.3-D mesh watermarking,in particular,plays a central role in the protection of 3-D content.However,frequency-domain methods rely on complex parameterization and spectral decomposition,which are sensitive to mesh topology and resolution and often introduce perceptible artifacts.Spatial-domain techniques,on the other hand,typically embed watermarks in global or randomly selected regions,leading to visible distortions and reduced robustness.To address the above limitations and protect model copyright without compromising the original aesthetic quality,we propose a deterministice PCA-synchronized 3Dmeshwatermarkingmethodwith fullerene-guided carrier selection.First,a deterministic principal component analysis(PCA)-based mesh synchronization algorithm is employed to align the models to a canonical pose.Next,a fullerene-inspired carrier selection strategy is employed to determine the watermark carriers,leveraging the structural characteristics of fullerene molecules to achieve a more rational and effective carrier selection.Finally,to balance the embedding strength and enhance visual quality,the watermark information is embedded using an APQIM(Adaptive Parity-Check Quantization Index Modulation)scheme.The experimental results show that our method can achieve high visual quality with scalable capacity and strong robustness compared with existing methods.The watermarking scheme can resist various attacks,including simplification,smoothing,Gaussian noise,translation,and rotation.
基金supported by the National Key R&D Program of China(No.2021ZD0112700).
文摘This paper develops an advanced framework for the operational optimization of integrated multi-energy systems that encompass electricity,gas,and heating networks.Introducing a cutting-edge stochastic gradient-enhanced distributionally robust optimization approach,this study integrates deep learning models,especially generative adversarial networks,to adeptly handle the inherent variability and uncertainties of renewable energy and fluctuating consumer demands.The effectiveness of this framework is rigorously tested through detailed simulations mirroring real-world urban energy consumption,renewable energy production,and market price fluctuations over an annual period.The results reveal substantial improvements in the resilience and efficiency of the grid,achieving a reduction in power distribution losses by 15%and enhancing voltage stability by 20%,markedly outperforming conventional systems.Additionally,the framework facilitates up to 25%in cost reductions during peak demand periods,significantly lowering operational costs.The adoption of stochastic gradients further refines the framework’s ability to continually adjust to real-time changes in environmental and market conditions,ensuring stable grid operations and fostering active consumer engagement in demand-side management.This strategy not only aligns with contem-porary sustainable energy practices but also provides scalable and robust solutions to pressing challenges in modern power network management.
基金supported by the National Science and Technology Council(NSTC),Taiwan,under grant number 114-2221-E-182-041-MY3by Chang Gung University and Chang Gung Memorial Hospital under project number NERPD4Q0021.
文摘The outstanding growth in the applications of large language models(LLMs)demonstrates the significance of adaptive and efficient prompt engineering tactics.The existing methods may not be variable,vigorous and streamlined in different domains.The offered study introduces an immediate optimization outline,named PROMPTx-PE,that is going to yield a greater level of precision and strength when it comes to the assignments that are premised on LLM.The proposed systemfeatures a timely selection schemewhich is informed by reinforcement learning,a contextual layer and a dynamic weighting module which is regulated by Lyapunov-based stability guidelines.The PROMPTx-PE dynamically varies the exploration and exploitation of the prompt space,depending on real-time feedback and multi-objective reward development.Extensive testing on both benchmark(GLUE,SuperGLUE)and domain-specific data(Healthcare-QA and Industrial-NER)demonstrates a large best performance to be 89.4%and a strong robustness disconnect with under 3%computation expense.The results confirm the effectiveness,consistency,and scalability of PROMPTx-PE as a platform of adaptive prompt engineering based on recent uses of LLMs.
基金supported by Science and Technology Project of SGCC(Research on Distributed Cooperative Control of Virtual Power Plants Based on Hybrid Game)(5700-202418337A-2-1-ZX).
文摘This paper suggests a way to improve teamwork and reduce uncertainties in operations by using a game theory approach involving multiple virtual power plants(VPP).A generalized credibility-based fuzzy chance constraint programming approach is adopted to address uncertainties stemming from renewable generation and load demand within individual VPPs,while robust optimization techniques manage electricity and thermal price volatilities.Building upon this foundation,a hierarchical Nash-Stackelberg game model is established across multiple VPPs.Within each VPP,a Stackelberg game resolves the strategic interaction between the operator and photovoltaic prosumers(PVP).Among VPPs,a cooperative Nash bargaining model coordinates alliance formation.The problem is decomposed into two subproblems:maximizing coalitional benefits,and allocating cooperative surpluses via payment bargaining,solved distributively using the alternating direction method of multipliers(ADMM).Case studies demonstrate that the proposed strategy significantly enhances the economic efficiency and uncertainty resilience of multi-VPP alliances.
基金funded by“the Fanying Special Program of the National Natural Science Foundation of China,grant number 62341307”“the Scientific research project of Jiangxi Provincial Department of Education,grant number GJJ200839”“theDoctoral startup fund of JiangxiUniversity of Technology,grant number 205200100402”.
文摘In multi-modal emotion recognition,excessive reliance on historical context often impedes the detection of emotional shifts,while modality heterogeneity and unimodal noise limit recognition performance.Existing methods struggle to dynamically adjust cross-modal complementary strength to optimize fusion quality and lack effective mechanisms to model the dynamic evolution of emotions.To address these issues,we propose a multi-level dynamic gating and emotion transfer framework for multi-modal emotion recognition.A dynamic gating mechanism is applied across unimodal encoding,cross-modal alignment,and emotion transfer modeling,substantially improving noise robustness and feature alignment.First,we construct a unimodal encoder based on gated recurrent units and feature-selection gating to suppress intra-modal noise and enhance contextual representation.Second,we design a gated-attention crossmodal encoder that dynamically calibrates the complementary contributions of visual and audio modalities to the dominant textual features and eliminates redundant information.Finally,we introduce a gated enhanced emotion transfer module that explicitly models the temporal dependence of emotional evolution in dialogues via transfer gating and optimizes continuity modeling with a comparative learning loss.Experimental results demonstrate that the proposed method outperforms state-of-the-art models on the public MELD and IEMOCAP datasets.
