In this paper, a finite-time observer-based fault-tolerant control (F-FTC) scheme is proposed to address path following problem of an underactuated surface vehicle suffering from input saturation and actuator gain degradation. First, a finite-time observer is designed to accurately detect faults and complex system uncertainties within the faulty system. Second, a smooth dead-zone operator-based model is applied to avoid the potential non-differentiability in controller. Finally, the F-FTC is established by virtue of a first-order filtering system. Lyapunov theory is used to verify the closed-loop system stability while simulations experiments validate the merits of the proposed F-FTC scheme.
Alzheimer's disease is a progressive neurological disorder. The disease is not reversible, but mild cognitive impairment is a transitional state between Alzheimer's disease and cognitively normal, it is a stage in which the brain is so minimally diseased that it can be treated or slowed or prevented from developing further lesions. At present, there are many methods for the aided diagnosis of mild cognitive impairment, among which the method based on fMRI medical images has emerged in recent years. However, the current aided diagnosis methods in this direction have some problems such as low accuracy and cumbersome feature extraction. A novel feature extraction method is put forward, namely the mean time series difference method in this paper. The feature extraction method effectively improves the accuracy in the auxiliary diagnosis task of mild cognitive impairment. An enhanced broad learning system with mean time series difference also was proposed. The results show that the enhanced broad learning system with mean time series can not only effectively optimize the feature extraction process, but also the accuracy of mild cognitive impairment classification tasks., which has significance for the clinical auxiliary diagnosis of mild cognitive impairment.
Ageism has become a significant concern in today's society, impacting the well-being of the elderly. Internalized ageism, the internal acceptance of negative stereotypes related to aging, leads to self-ageism and negative self-perceptions among older individuals. This study summarizes 11 key factors influencing self-ageism in the elderly and employs the Decision-making Trial and Evaluation Laboratory (DEMATEL) method to analyze data from 26 questionnaires, exploring the direction and degree of interaction between each pair of factors. Based on a comprehensive analysis, it is found that Educational level, Income and expenditure, and Elderly diseases are the most influential factors contributing to self-ageism among the elderly in Vietnam. Therefore, focusing on these three factors as key starting points and considering the current implementation status of elderly policies in Vietnam, this study proposes improvement recommendations. Through an exploration of internalized ageism among Vietnamese elderly individuals, this research aims to offer insights for targeted interventions to enhance positive aging experiences and improve the well-being of the elderly.
The need for flexible and effective 5G communication networks has prompted studies into novel approaches for network slicing, a fundamental paradigm that permits network service personalization for a range of applications. In this research, we provide a unique technique that combines Graph Convolution Networks (GCN) and Grey Wolf Optimization (GWO) strategies to improve the adaptive configuration of 5G network slices. GCN, a software capable of recording detailed network interactions, and GWO, well-known for its metaheuristic optimization, combine to offer a strong foundation for tackling dynamic network slicing challenges. We optimize GWO settings via rigorous testing and demonstrate the effectiveness of the integrated strategy by achieving gains in important performance measures including latency, throughput, and resource use. The investigation not only advances the continuing development of 5G networks but also highlights how cutting-edge machine learning and optimization techniques can completely transform network slicing approaches.
Rivers and agricultural canals form a network that serves as an important habitat for aquatic ecosystems. Urbanization, field development, and water pollution have led to the deterioration of aquatic environments, resulting in the loss of bio-diversity therein, making ecosystem conservation and restoration a pressing issue. Freshwater fish that inhabit agricultural channel networks are significantly affected by changes in flow regimes due to irrigation, and it is essential to consider seasonal changes in both natural and anthropogenic flow dynamics for effective ecosystem conservation. Species distribution models are valuable tools for quantitatively assessing species-habitat relationships. This study examines the applicability of multiobjective fuzzy genetics-based machine learning (MoFGBML) to fish habitat assessment for major species recorded in an ecohydraulic survey dataset of an irrigation system located in a suburb of Tokyo. MoFGBML generates fuzzy if-then rule based species distribution models with different tradeoffs between accuracy and interpretability. Our focus is the model interpretability and knowledge extraction from the species distribution model, specifically feature importance and rules in relation to the model accuracy. We also show the importance of the temporal information in fish habitat assessment.
The adaptive secure controller is developed for state-constrained uncertain cyber-physical systems against sensor and actuator attacks. Firstly, fuzzy logic systems are utilized to estimate the unknown nonlinear function. Secondly, the non-linear mapping function is introduced to transform the state-constrained system into an unconstrained one and directly deal with the state-constrained issue. By ensuring the boundedness of the transformed system state, the state-constrained condition is not violated in the absence of the feasibility condition of the barrier Lyapunov function. Then, the deception attacks suffered by the sensor channel and actuator channel are modeled as a multiplicative form to decouple the control of network attacks from state constraints. With the help of the command filter, the derivatives of virtual control signals can be directly obtained and the fuzzy adaptive controller is designed to ensure that all signals of the closed-loop system are bounded. Finally, the effectiveness of the developed method is further verified by a numerical simulation.
In this paper, reinforcement learning is employed to address the cooperative hunting task for a swarm of underactuated unmanned surface vehicles (USVs) in maritime warfare scenarios. By virtue of the multi-agent deep deterministic policy gradient algorithm, the cooperative swarm hunting strategy (CSHS) is developed by defining new reward functions that can guide the USV swarm in a continuous manner, thereby avoiding sparse rewards. Moreover, Unity3D-based virtual environments are built for cost-efficient learning of the proposed CSHS. The results demonstrate that the CSHS is able to not only avoid obstacles but also exhibits strong generalization among different scenarios, within the entire USV swarm hunting process.
In this paper, we address the fast finite-time stability problem for a class of uncertain stochastic nonlinear systems. The dead-zone phenomenon is a general and complex challenge in the field of nonlinear systems, which is frequently encountered in real control systems, especially in critical components such as actuators and sensors. In conjunction with the proposed backstepping approach, an adaptive fuzzy control design strategy is established and a corresponding dead-zone controller is constructed. Second, the approximation capability of fuzzy logic systems is utilized for the simultaneous handling of stochastic nonlinear systems with completely unknown nonlinearities as well as disturbance terms. Finally, a numerical simulation is provided to verify the correctness of the theoretical analysis.
With the explosive growth of data in the financial industry, efficient collection, analysis, and utilization of this data have become urgent challenges. This study aims to develop a financial big data collection and intelligent decision-making system based on multimodal large language models (MM-LLM). The system utilizes Google's AI mega-model, Gemini, to achieve functionalities such as classification, retrieval, analysis, intelligent evaluation, and recommendation of financial big data. Through parameter configuration on the management side, the system can flexibly adapt to the data collection and analysis needs in different financial scenarios, significantly improving the efficiency of data processing and the accuracy of decision-making. Innovative solutions are provided to address the specific requirements of investors in analyzing vast announcement data and evaluating individual stocks, assisting them in making wiser investment decisions. The system adopts a front-end and back-end separation architecture, based on distributed microservices, and integrates technologies such as Vue3, Springboot, Nacos, Dubbo, Redis cluster, MySQL, Elasticsearch, Gemini, Ansj, and Kafka. By developing and applying this system, the study demonstrates the vast potential of multimodal large language models in the field of financial big data, laying a solid foundation for future intelligent financial decision-making,
In this paper proposed an effective fuzzy expectation-maximization phoneme prediction method in diffusion model-based dysarthria voice conversion (FEMPPDM-DVC) which is accessible to (i) training without parallel data (ii) converting a longer duration of the audio data (iii) preserves speaker identity. By integrating Fuzzy Expectation-Maximization (FEM) clustering and Diffusion model-based dysarthria voice conversion approach, the proposed method is able gradually generate normal utterances. Feature extraction is performed using Mel-frequency cepstral coefficients (MFCCs), a robust method in acoustic feature extraction in Text-to-Speech systems. Ensures the effectiveness and accuracy, through repeated parameter adjustment using the FEM clustering algorithm. The conversion network is a diffusion model-based structure, which can convert normal utterances to dysarthria utterances in the forward diffusion process. After forward diffusion process, the dysarthria utterance will go through a reverse diffusion process to convert dysarthria voice to normal utterance. Once converted, a GAN-based vocoder is applied to convert mel-frequency spectrogram to waveform. Objective evaluation is conducted on the Saarbrücken Voice Database (SVD) dataset show that FEMDDPM-DVC is able to improve the intelligibility and naturalness of dysarthria utterances in 15 kinds of dysarthria voice. The proposed method is compared with five other dysarthria voice conversion methods, show that the proposed method has the most performance among other method.
The ARIMA model is a popular way to forecast financial time series. However, many time series such as financial or biological time series often have nonlinear properties which cause in model errors during the evolutionary of fitting ARIMA models, and this results in significant prediction errors. To overcome this dilemma, a nonlinear autoregressive model is proposed which combines the method of LASSO, ARIMA, and BP neural network in the time-series modellings. Firstly, the LASSO model is used to reduce the dimensions of the input exogenous variables where the exchange rate is seen as endogenous variables, and the selected variables are those playing important roles, at the same time, the multicollinearity can also be eliminated. Secondly, based on the observing values of endogenous and exogenous variables, the BP neural networks with these selected variables are trained to obtain the nonlinear prediction models which are used to make predictions. Furthermore, in order to check whether there exist the correlations in the residual sequences, the white noise tests are proposed. If it fails, we use the ARIMA model to make further predictions to obtain more accurate predictions. Thirdly, the reciprocal variance methods are used to combine the results from the BP neural network and the ARIMA model, and the ideal predictive values are obtained. Finally, the proposed LASSO-ARIMA-BP neural network model are applied to predict the exchange rate. Comparing with other models such as random forest, ARIMA, and BP neural network using different evaluation metrics, it illustrates that the proposed combined model is more accurate in the exchange rate predictions.
This paper proposes a fuzzy optimal adaptive prescribed performance control strategy for nonlinear multi-agent systems. Firstly, to solve the prescribed performance problem, an error transformation function is used to transform the constraint problem into a bounded problem. Subsequently, a distributed fuzzy control protocol is proposed, utilizing a reinforcement learning algorithm. The control scheme ensures that all closed-loop signals are bounded, and the consensus tracking error remains within a predefined range. Finally, the effectiveness of the proposed control scheme is verified through simulation.
Person re-identification (Re-ID) is an important research field in computer vision and pedestrian detection. However, there are many complex and highly uncertain factors in the real world, such as occlusion, appearance similarity, and motion blur. These factors pose serious challenges to obtaining accurate and robust feature representation. In order to deploy person Re-Idalgorithms on mobile devices and other terminal systems, it is necessary to consider both model complexity and recognition accuracy. To address these uncertainties and enhance the real-time detection accuracy of person Re-ID, we propose a lightweight attention network based on fuzzy logic (FLA-Net). In the backbone network, a pair of complementary attention mechanisms are embedded to capture the discriminative features of pedestrians. Additionally, fuzzy logic is introduced into the attention module to re-weight the feature maps, optimizing the accuracy and robustness of feature representation by adjusting the fuzzy membership degree of pixel values in local regions. Finally, we employ the local horizontal pooling operation to extract fine-grained information from the network, facilitating the capture of discriminative pedestrian features. Experimental analysis of the public datasets Market1501 and DukeMTMC-reID demonstrates that FLA-Net is superior to the state-of-the-art lightweight person Re-Idmethods.
Robust navigable segmentation and obstacle detection are crucial for an autonomous surface vehicle. In this paper, to effectively tackle the severe challenge of distinguishing obstacles from surface disturbances by virtue of inter-frame correlation, a bilateral spatio-temporal contextual obstacle seg-mentation network (STOS-Net) is innovatively devised. In the context branch, to capture spatio-temporal contextual information, by utilizing the shared encoder to extract features involving previous and current frames, a spatio-temporal attention module is proposed, such that potential relationships among features of consecutive frames can be effectively established, and thereby distinguishing surface disturbance and actual obstacle. Subse-quently, the Sobel edge detection operator is deployed to generate boundary maps, which guide the detail branch to produce more precise edge feature maps, thereby significantly enhancing the accuracy of waterline segmentation. Moreover, to address the significant multi-scale variations caused by the size of obstacles, the multi-scale information from the two branches is integrated by a multi-layer perceptron module. The findings demonstrate that the innovative STOS-Net attains an F1 score of 93.5 on the MODS dataset, along with a localization error of 12.1 pixels for waterline segmentation.
This paper investigates the secure consensus control problem for nonlinear multi-agent systems (MASs) under de-ception attacks. The main characteristic of the proposed secure consensus control method is its ability to mitigate undesirable system behaviors caused by deception attacks at the communication links among different agents. In particular, the relationship between the compromised signals and the output signals of MASs is analyzed. Furthermore, a novel Nussbaum function and an adaptive fuzzy law are respectively proposed to tackle the unknown control direction and to estimate the unknown terms caused by deception attacks. Simulation results are presented to demonstrate the validity of the proposed adaptive consensus secure control method.
This paper implements a dynamic goalkeeper defense policy based on Deep Reinforcement Learning (DRL) for a soccer robot. This study proposed a multi-directional training approach to enable soccer goalkeepers to learn defense strategies for acting an appropriate manner. This paper using the Soft Actor-Critic (SAC) as DRL architecture and the Gazebo simulator as an environment simulation platform. Utilizing progressive learning with good reward function design to enable soccer robot goalkeepers to explore and learn in a simulated environment. It efficiently completed the better defense task. Finally, the correctness of the defense policy is evaluated by testing and validating it in a multi-directional area. The experimental results demonstrated the designed SAC containing the superior performance for successful defense rates in this paper.
It is well known that the Gaussian mixture model (GMM) is an exemplary model-based clustering approach that captures data distribution patterns to facilitate clustering. However, when it faces to multi-view clustering tasks, the classical GMM algorithm often encounters limitations in scalability. Multi-view clustering entails employing diverse views or feature sets to cluster data, thereby achieving comprehensive and accurate clustering outcomes. The classical GMM algorithm fails to fully exploit the correlated information across multiple views, resulting in a decrease in clustering accuracy. This paper proposes a new GMM-based trans-fer multi-view clustering algorithm named MT-GMM. This algorithm transfers valuable knowledge from other views to guide and enhance the clustering task of the current view, thereby improving the clustering performance specifically for the current view. Experimental results conducted on real-world datasets demonstrate that the proposed MT-GMM outperforms several other multi-view clustering methods, as it can effectively harness the complementary information between different views accurately capture the structures of data.
In bioinformatics, classifying protein sequences into anticancer peptides (ACPs) and non-ACPs is crucial yet challenging due to the inherent uncertainties of biological data. This study introduces a novel fuzzy neural network (FNN) model that integrates fuzzy logic within neural network architectures, enhancing the handling of ambiguity and improving classification accuracy. Our model, tested against several conventional machine learning models and recent studies, demonstrated superior specificity (83.28%) and overall accuracy (79.14%), marking a significant advancement in the identification of therapeutically relevant peptides. The integration of fuzzy logic not only optimized the performance but also increased the interpretability of the results, making it a valuable tool for complex bioinformatic analyses. These findings underscore the potential of fuzzy systems to refine predictive capabilities in computational biology, aligning perfectly with the themes of enhancing fuzzy theory applications in practical and impactful ways.
This paper presents a control design methodology for rotary inverted pendulum based on a fuzzy descriptor system. In this methodology, the nonlinear dynamics of the rotary inverted pendulum is exactly presented by a fuzzy descriptor system. After that, based on the fuzzy descriptor system and applying the Lyapunov stability theorem, a decay rate fuzzy control design is proposed for stabilizing the rotary inverted pendulum. Furthermore, for practical application, the constraint of input voltage is also considered in the control design. The control design is represented in terms of linear matrix inequality (LMI) conditions which can be efficiently solved by the LMI solvers of MATLAB Robust Control Toolbox. Finally, simulation and experiment results are provided to demonstrate the effectiveness of the proposed fuzzy-descriptor-system-based control design methodology.
This research focuses on detecting driver fatigue in real-time. Driver drowsiness is a serious issue that often leads to traffic accidents, making it a particularly hazardous concern. Therefore, preventing and reducing driver fatigue and distraction is an important area of research. The study aims to engineer a real-time, non-intrusive drowsiness detection system image processing and fuzzy logic techniques. To determine the level of fatigue, the system analyzes facial features, including the state of the eyes and mouth. However, the calculation of the states of the eyes and mouth involves various uncertainties, such as vagueness and imprecision. Therefore, a Belief Rule-Based Approach (BRBA) is implemented due to its proficiency in managing uncertainties. In a predefined rule base, each input to an antecedent attribute is mapped to a belief distribution. Subsequently, within this rule base, the Equilibrium Optimizer (EO) method is employed to execute inference processes. A novel Belief Rule-Based (BRB) system is developed by meticulously analyzing the associated risks of fatigue and drowsiness, with the objective of accurately determining the risk levels of various driver states.
