Article
Computer Science, Artificial Intelligence
Ziyu Lin, Jingliang Duan, Shengbo Eben Li, Haitong Ma, Jie Li, Jianyu Chen, Bo Cheng, Jun Ma
Summary: The research addresses the challenge of solving the finite-horizon HJB equation, proposes a new algorithm, and validates its effectiveness through simulations.
IEEE TRANSACTIONS ON NEURAL NETWORKS AND LEARNING SYSTEMS
(2023)
Article
Automation & Control Systems
Mayank Shekhar Jha, Didier Theilliol, Philippe Weber
Summary: This paper develops a novel ADP-based approach for solving the optimal tracking problem for completely unknown discrete time systems. Through suitable system transformation and an iterative ADP-based algorithm, the problem is transformed to a regulation problem and solved in an approximative sense. Simulation studies illustrate the effectiveness of the proposed algorithm.
OPTIMAL CONTROL APPLICATIONS & METHODS
(2023)
Article
Computer Science, Artificial Intelligence
Linghuan Kong, Shuang Zhang, Xinbo Yu
Summary: An approximate optimal scheme is proposed for an uncertain n-link robot subject to saturation nonlinearity. The proposed method takes into account model uncertainty in robotic dynamics and designs an optimal control under the frame of adaptive dynamic programming. The method is proved to be effective in stabilizing the unknown system and reducing control cost.
Article
Automation & Control Systems
Shangwei Zhao, Jingcheng Wang, Haotian Xu, Hongyuan Wang
Summary: In this paper, an approximate dynamic programming approach is proposed for handling the robust optimal tracking control problem in switched systems with uncertainties. A neural network based identifier is used to estimate the unknown system dynamics, and actor-critic neural networks are constructed to approximate the optimal control input and performance index. The convergence of the proposed approach is proved, and numerical simulations are conducted to validate its effectiveness.
INTERNATIONAL JOURNAL OF CONTROL AUTOMATION AND SYSTEMS
(2022)
Article
Automation & Control Systems
Cong Li, Yongchao Wang, Fangzhou Liu, Qingchen Liu, Martin Buss
Summary: This article presents a new formulation for model-free robust optimal regulation of continuous-time nonlinear systems using an incremental adaptive dynamic programming (IADP) approach. It leverages time delay estimation (TDE) technique and measured input-state data to achieve incremental stabilization under uncertainties, disturbances, and saturation. Numerical simulations validate its effectiveness and superiority in reducing energy expenditure and enhancing robustness.
INTERNATIONAL JOURNAL OF ROBUST AND NONLINEAR CONTROL
(2022)
Article
Engineering, Mechanical
Jianfeng Wang, Ping Zhang, Yan Wang, Zhicheng Ji
Summary: This paper investigates the problem of adaptive optimal tracking control for full-state constrained strict-feedback nonlinear systems with input delay. A novel control approach is developed by combining the backstepping design technique and adaptive dynamic programming (ADP) theory. The approach utilizes Pade approximation to handle input delay and barrier Lyapunov functions for state constraints. Neural networks are employed to approximate unknown functions. An adaptive backstepping feedforward controller is developed to convert the tracking task into an equivalent regulation problem. A critic network is constructed within the ADP framework to obtain the optimal control. The resulting controller consists of feedforward and feedback parts, while ensuring that all signals are uniformly ultimately bounded in the closed-loop system.
NONLINEAR DYNAMICS
(2023)
Article
Computer Science, Artificial Intelligence
Lijie Wang, Jiahong Xu, Yang Liu, C. L. Philip Chen
Summary: This article investigates the optimal consensus control problem for multiagent systems with input constraints. It proposes a single critic neural network with time-varying activation function for approximate optimal control and an improved learning law for weight update. It also designs an effective dynamic event-triggering mechanism to improve the utilization rate of communication resource. A simulation example is provided to support the effectiveness of the proposed method and the superiority of the designed mechanism.
IEEE TRANSACTIONS ON NEURAL NETWORKS AND LEARNING SYSTEMS
(2023)
Article
Automation & Control Systems
Antonio Sala, Leopoldo Armesto
Summary: This study introduces a new criterion for adaptive meshing in polyhedral partitions to interpolate value functions, employing an initial condition probability density function, uncertainty propagation, and temporal-difference error to determine the addition of new points. A collection of lemmas justifies the algorithmic proposal, with comparative analysis highlighting the advantages of this proposal over other options in literature. The developed methods are applied in simulation examples and an experimental robotic setup.
ENGINEERING APPLICATIONS OF ARTIFICIAL INTELLIGENCE
(2022)
Article
Automation & Control Systems
Qinglai Wei, Zehua Liao, Ruizhuo Song, Pinjia Zhang, Zhuo Wang, Jun Xiao
Summary: This article solves the optimal control scheme for ice-storage air conditioning (IAC) system using a data-based adaptive dynamic programming (ADP) method for the first time. The developed ADP method improves system efficiency and reduces operation costs according to numerical results. The superiority of the developed algorithm is verified in comparison results.
IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS
(2021)
Article
Mathematics, Applied
Heng Zhang
Summary: This paper proposes a novel adaptive dynamic programming (ADP)-based model-free policy iteration (PI) algorithm to solve an infinite-horizon continuous-time linear quadratic stochastic (LQS) optimal control problem, which includes both control and state variables in the diffusion term of system dynamics. By using Ito's lemma and expectations, a relationship among the state trajectory, control input, and matrices to be solved is described. The ADP-based model-free algorithm is then developed to approximate the optimal control from collected data without requiring information about all system coefficient matrices. Convergence analysis is provided under mild conditions, and numerical examples demonstrate the effectiveness of the proposed algorithm.
JOURNAL OF APPLIED MATHEMATICS AND COMPUTING
(2023)
Article
Automation & Control Systems
Songsong Cheng, Mingjian Zhu, Yuhui Fu, Xiaohan Fang, Yuan Fan
Summary: This paper proposes a novel dynamic event-triggered approximate optimal control strategy for nonlinear continuous-time systems, achieving system stability, cost minimization and resource savings simultaneously. The optimal controller is obtained through HJB equation and neural network weight calculation. The dynamic event-triggered condition results in higher efficiency and the Lyapunov method proves the boundedness of system states.
INTERNATIONAL JOURNAL OF CONTROL AUTOMATION AND SYSTEMS
(2022)
Article
Automation & Control Systems
Qinglai Wei, Liao Zhu, Tao Li, Derong Liu
Summary: This article develops a new time-varying adaptive dynamic programming (ADP) algorithm to solve finite-horizon optimal control problems for a class of discrete-time affine nonlinear systems. Inspired by the pseudolinear method, the nonlinear system can be approximated by a series of time-varying linear systems. The paper proves the convergence of the states of the time-varying linear systems to the states of the discrete-time affine nonlinear systems and the convergence of the iterative value functions and control laws to the optimal ones. Numerical results are provided to verify the effectiveness of the proposed method.
IEEE TRANSACTIONS ON AUTOMATIC CONTROL
(2022)
Article
Automation & Control Systems
Xiong Yang, Haibo He, Xiangnan Zhong
Summary: The paper studies the constrained optimization problem of a class of uncertain nonlinear interconnected systems, proposing an SPI algorithm to solve the Hamilton-Jacobi-Bellman equations of constrained auxiliary subsystems, implementing the algorithm through an actor-critic structure, and validating the method through simulation.
IEEE TRANSACTIONS ON CYBERNETICS
(2021)
Article
Computer Science, Artificial Intelligence
Danyu Lin, Shan Xue, Derong Liu, Mingming Liang, Yonghua Wang
Summary: In this paper, a problem of multiplayer hierarchical decision-making for non-affine systems is solved using adaptive dynamic programming. The control dynamics are obtained and combined with the original system dynamics, transforming the non-affine multiplayer system into a general affine form. The hierarchical decision problem is modeled as a Stackelberg game, and a neural network is used to reconstruct the augmented system and approximate the value function. The feasibility and effectiveness of the algorithm are confirmed through simulation.
Article
Mathematics, Applied
Tomasz Sluzalec, Rafal Grzeszczuk, Sergio Rojas, Witold Dzwinel, Maciej Paszynski
Summary: This paper presents how to construct a deep neural network (DNN) expert for predicting quasi-optimal hp-refinements in finite element problems with singularities. The main approach is to train the DNN expert during the execution of the self-adaptive hp-finite element method (hp-FEM) algorithm and utilize it to predict further hp refinements. A two-grid paradigm self-adaptive hp-FEM algorithm is used for training, where a fine mesh is employed to provide optimal hp refinements for coarse mesh elements. The DNN expert is trained to predict the location of singularities and continue selecting quasi-optimal hp-refinements, maintaining exponential convergence of the method.
COMPUTERS & MATHEMATICS WITH APPLICATIONS
(2023)
Article
Computer Science, Artificial Intelligence
Mingming Ha, Ding Wang, Derong Liu
Summary: In this article, a novel value iteration scheme is proposed, which introduces a relaxation factor and combines with other methods to accelerate and guarantee the convergence. The theoretical results and numerical examples demonstrate its fast convergence speed and stability.
IEEE TRANSACTIONS ON NEURAL NETWORKS AND LEARNING SYSTEMS
(2023)
Article
Automation & Control Systems
Shunchao Zhang, Bo Zhao, Derong Liu, Cesare Alippi, Yongwei Zhang
Summary: In this article, an event-triggered robust control (ETRC) method is investigated for multi-player nonzero-sum games of continuous-time input constrained nonlinear systems with mismatched uncertainties. The method transforms the robust control problem into an optimal regulation problem by constructing an auxiliary system and designing an appropriate value function. A critic neural network (NN) is used to approximate the value function of each player and obtain control laws. The method reduces computational burden and communication bandwidth by updating the control laws when events occur. The effectiveness of the developed ETRC method is demonstrated through theoretical analysis and examples.
INTERNATIONAL JOURNAL OF ROBUST AND NONLINEAR CONTROL
(2023)
Article
Computer Science, Artificial Intelligence
Qiuye Wu, Bo Zhao, Derong Liu, Marios M. Polycarpou
Summary: This paper proposes an event-triggered adaptive dynamic programming method to solve the decentralized tracking control problem for input constrained unknown nonlinear interconnected systems. A neural-network-based local observer is established to reconstruct the system dynamics using local input-output data and desired trajectories. The DTC problem is transformed into an optimal control problem using a nonquadratic value function. The DTC policy is obtained by solving the local Hamilton-Jacobi-Bellman equation through the observer-critic architecture, with weights tuned by the experience replay technique. Simulation examples demonstrate the effectiveness of the proposed scheme.
Article
Computer Science, Artificial Intelligence
Mingduo Lin, Bo Zhao, Derong Liu
Summary: A novel policy gradient (PG) adaptive dynamic programming method is proposed for nonlinear discrete-time zero-sum games with unknown dynamics. A policy iteration algorithm is used to approximate the Q-function and the control and disturbance policies using neural network approximators. The control and disturbance policies are then updated using the PG method based on the iterative Q-function. The experience replay technique is applied to improve training stability and data usage efficiency. Simulation results show the effectiveness of the proposed method.
Article
Computer Science, Artificial Intelligence
Mingming Liang, Derong Liu
Summary: This article presents a novel neural-network-based optimal event-triggered impulsive control method. The proposed method utilizes a general-event-based impulsive transition matrix (GITM) to represent the evolving characteristics of all system states across impulsive actions. Through the developed event-triggered impulsive adaptive dynamic programming (ETIADP) algorithm and its high-efficiency version (HEIADP), the optimization problems for stochastic systems with event-triggered impulsive controls are addressed. The results show that the proposed methods can reduce computational and communication burdens and fulfill the desired goals.
IEEE TRANSACTIONS ON NEURAL NETWORKS AND LEARNING SYSTEMS
(2023)
Article
Automation & Control Systems
Runqi Chai, Derong Liu, Tianhao Liu, Antonios Tsourdos, Yuanqing Xia, Senchun Chai
Summary: This paper presents an integrated real-time trajectory planning and tracking control framework for autonomous ground vehicles (AGV) parking maneuver problems, utilizing deep neural networks and recurrent network structures. Two transfer learning strategies are applied to adapt the motion planner for different AGV types. Experimental studies demonstrate enhanced performance in fulfilling parking missions.
IEEE TRANSACTIONS ON AUTOMATION SCIENCE AND ENGINEERING
(2023)
Article
Computer Science, Artificial Intelligence
Danyu Lin, Shan Xue, Derong Liu, Mingming Liang, Yonghua Wang
Summary: In this paper, a problem of multiplayer hierarchical decision-making for non-affine systems is solved using adaptive dynamic programming. The control dynamics are obtained and combined with the original system dynamics, transforming the non-affine multiplayer system into a general affine form. The hierarchical decision problem is modeled as a Stackelberg game, and a neural network is used to reconstruct the augmented system and approximate the value function. The feasibility and effectiveness of the algorithm are confirmed through simulation.
Article
Computer Science, Artificial Intelligence
Chujian Zeng, Bo Zhao, Derong Liu
Summary: This paper proposes a neuro-dynamic programming-based fault tolerant control scheme for a class of nonlinear systems, considering the occurrence of both actuator and sensor faults simultaneously. The scheme combines a descriptor observer with an adaptive observer to estimate system states and multiple faults. By employing a critic neural network, the approximate optimal control policy is obtained for the fault-free system. An FTC law is developed to suppress the influence of actuator faults by combining the estimations of actuator faults with the approximate optimal control policy. The stability of the closed-loop nonlinear system is analyzed using the Lyapunov stability theorem.
Article
Automation & Control Systems
Yongwei Zhang, Bo Zhao, Derong Liu, Shunchao Zhang
Summary: In this article, the event-triggered robust control problem of unknown multiplayer nonlinear systems with constrained inputs and uncertainties is investigated using adaptive dynamic programming. A neural network-based identifier is constructed to relax the requirement of system dynamics. By designing a nonquadratic value function, the stabilization problem is converted into a constrained optimal control problem. The approximate solution of the event-triggered Hamilton-Jacobi equation is obtained using a critic network with a novel weight updating law, and the Lyapunov stability theorem ensures that the multiplayer system is uniformly ultimately bounded.
IEEE TRANSACTIONS ON CYBERNETICS
(2023)
Article
Automation & Control Systems
Mingming Liang, Yonghua Wang, Derong Liu
Summary: In this study, a novel general impulsive transition matrix is defined to reveal the transition dynamics and probability distribution evolution patterns between impulsive events. Based on this matrix, policy iteration-based impulsive adaptive dynamic programming algorithms are developed to solve optimal impulsive control problems. The algorithms demonstrate convergence to the optimal impulsive performance index function and allow for optimization on computing devices with low memory spaces.
IEEE TRANSACTIONS ON CYBERNETICS
(2023)
Proceedings Paper
Automation & Control Systems
Jinquan Lin, Bo Zhao, Derong Liu
Summary: In this paper, an integral reinforcement learning (IRL)-based approximate optimal control (AOC) method is developed for unknown nonaffine systems using dynamic feedback. The optimal control policy for nonaffine systems cannot be explicitly expressed due to the unknown input gain matrix. Thus, a dynamic feedback signal is introduced to transform the nonaffine system into an augmented affine system. The AOC for unknown nonaffine systems is formulated by designing an appropriate value function for the augmented affine system, and the IRL method is adopted to derive the approximate solution of the Hamilton-Jacobi-Bellman equation.
2023 IEEE 12TH DATA DRIVEN CONTROL AND LEARNING SYSTEMS CONFERENCE, DDCLS
(2023)
Article
Automation & Control Systems
Bo Zhao, Guang Shi, Derong Liu
Summary: This article investigates local control problems for nonlinear interconnected systems by using adaptive dynamic programming (ADP) with particle swarm optimization (PSO). It constructs a proper local value function and employs a local critic neural network to solve the local Hamilton-Jacobi-Bellman equation. The event-triggering mechanism is introduced to determine the sampling time instants and ensure asymptotic stability through Lyapunov stability analysis.
IEEE TRANSACTIONS ON SYSTEMS MAN CYBERNETICS-SYSTEMS
(2023)
Article
Automation & Control Systems
Mingming Liang, Derong Liu
Summary: This article focuses on designing the optimal impulsive controller (IMC) of continuous-time nonlinear systems and proposes a new adaptive dynamic programming algorithm with high generality and feasibility. The introduced policy-improving mechanism makes the algorithm more flexible for memory-limited computing devices.
IEEE TRANSACTIONS ON SYSTEMS MAN CYBERNETICS-SYSTEMS
(2023)
Article
Automation & Control Systems
Ke Wang, Chaoxu Mu, Zhen Ni, Derong Liu
Summary: This paper presents a novel composite obstacle avoidance control method that generates safe motion trajectories for autonomous systems in an adaptive manner. The method combines model-based policy iteration and state-following-based approximation in a safe reinforcement learning framework. The proposed learning-based controller achieves stable reaching of target points while maintaining a safe distance from obstacles. The effectiveness of the method is demonstrated through simulations and comparisons with other avoidance control methods.
IEEE TRANSACTIONS ON AUTOMATION SCIENCE AND ENGINEERING
(2023)
Article
Computer Science, Artificial Intelligence
Bo Zhao, Yongwei Zhang, Derong Liu
Summary: This article presents a cooperative motion/force control scheme for modular reconfigurable manipulators (MRMs) based on adaptive dynamic programming (ADP). The dynamic model of the entire MRM system is treated as a set of joint modules interconnected by coupling torque, and the Jacobian matrix is mapped into each joint. A neural network is used as a robust decentralized observer, and an improved local value function is constructed for each joint module. The control scheme is achieved by using force feedback compensation and is proven to be uniformly ultimately bounded through Lyapunov stability analysis.
IEEE TRANSACTIONS ON NEURAL NETWORKS AND LEARNING SYSTEMS
(2023)
Article
Computer Science, Artificial Intelligence
Hamdan Abdellatef, Lina J. Karam
Summary: This paper proposes performing the learning and inference processes in the compressed domain to reduce computational complexity and improve speed of neural networks. Experimental results show that modified ResNet-50 in the compressed domain is 70% faster than traditional spatial-based ResNet-50 while maintaining similar accuracy. Additionally, a preprocessing step with partial encoding is suggested to improve resilience to distortions caused by low-quality encoded images. Training a network with highly compressed data can achieve good classification accuracy with significantly reduced storage requirements.
Article
Computer Science, Artificial Intelligence
Victor R. Barradas, Yasuharu Koike, Nicolas Schweighofer
Summary: Inverse models are essential for human motor learning as they map desired actions to motor commands. The shape of the error surface and the distribution of targets in a task play a crucial role in determining the speed of learning.
Article
Computer Science, Artificial Intelligence
Ting Zhou, Hanshu Yan, Jingfeng Zhang, Lei Liu, Bo Han
Summary: We propose a defense strategy that reduces the success rate of data poisoning attacks in downstream tasks by pre-training a robust foundation model.
Article
Computer Science, Artificial Intelligence
Hao Sun, Li Shen, Qihuang Zhong, Liang Ding, Shixiang Chen, Jingwei Sun, Jing Li, Guangzhong Sun, Dacheng Tao
Summary: In this paper, the convergence rate of AdaSAM in the stochastic non-convex setting is analyzed. Theoretical proof shows that AdaSAM has a linear speedup property and decouples the stochastic gradient steps with the adaptive learning rate and perturbed gradient. Experimental results demonstrate that AdaSAM outperforms other optimizers in terms of performance.
Article
Computer Science, Artificial Intelligence
Juntong Yun, Du Jiang, Li Huang, Bo Tao, Shangchun Liao, Ying Liu, Xin Liu, Gongfa Li, Disi Chen, Baojia Chen
Summary: In this study, a dual manipulator grasping detection model based on the Markov decision process is proposed. By parameterizing the grasping detection model of dual manipulators using a cross entropy convolutional neural network and a full convolutional neural network, stable grasping of complex multiple objects is achieved. Robot grasping experiments were conducted to verify the feasibility and superiority of this method.
Article
Computer Science, Artificial Intelligence
Miaohui Zhang, Kaifang Li, Jianxin Ma, Xile Wang
Summary: This paper proposes an unsupervised person re-identification (Re-ID) method that uses two asymmetric networks to generate pseudo-labels for each other by clustering and updates and optimizes the pseudo-labels through alternate training. It also designs similarity compensation and similarity suppression based on the camera ID of pedestrian images to optimize the similarity measure. Extensive experiments show that the proposed method achieves superior performance compared to state-of-the-art unsupervised person re-identification methods.
Article
Computer Science, Artificial Intelligence
Florian Bacho, Dominique Chu
Summary: This paper proposes a new approach called the Forward Direct Feedback Alignment algorithm for supervised learning in deep neural networks. By combining activity-perturbed forward gradients, direct feedback alignment, and momentum, this method achieves better performance and convergence speed compared to other local alternatives to backpropagation.
Article
Computer Science, Artificial Intelligence
Xiaojian Ding, Yi Li, Shilin Chen
Summary: This research paper addresses the limitations of recursive feature elimination (RFE) and its variants in high-dimensional feature selection tasks. The proposed algorithms, which introduce a novel feature ranking criterion and an optimal feature subset evaluation algorithm, outperform current state-of-the-art methods.
Article
Computer Science, Artificial Intelligence
Naoko Koide-Majima, Shinji Nishimoto, Kei Majima
Summary: Visual images observed by humans can be reconstructed from brain activity, and the visualization of arbitrary natural images from mental imagery has been achieved through an improved method. This study provides a unique tool for directly investigating the subjective contents of the brain.
Article
Computer Science, Artificial Intelligence
Huanjie Tao, Qianyue Duan
Summary: In this paper, a hierarchical attention network with progressive feature fusion is proposed for facial expression recognition (FER), addressing the challenges posed by pose variation, occlusions, and illumination variation. The model achieves enhanced performance by aggregating diverse features and progressively enhancing discriminative features.
Article
Computer Science, Artificial Intelligence
Zhenyi Wang, Pengfei Yang, Linwei Hu, Bowen Zhang, Chengmin Lin, Wenkai Lv, Quan Wang
Summary: In the face of the complex landscape of deep learning, we propose a novel subgraph-level performance prediction method called SLAPP, which combines graph and operator features through an innovative graph neural network called EAGAT, providing accurate performance predictions. In addition, we introduce a mixed loss design with dynamic weight adjustment to improve predictive accuracy.
Article
Computer Science, Artificial Intelligence
Yiyang Yin, Shuangling Luo, Jun Zhou, Liang Kang, Calvin Yu-Chian Chen
Summary: Medical image segmentation is crucial for modern healthcare systems, especially in reducing surgical risks and planning treatments. Transanal total mesorectal excision (TaTME) has become an important method for treating colon and rectum cancers. Real-time instance segmentation during TaTME surgeries can assist surgeons in minimizing risks. However, the dynamic variations in TaTME images pose challenges for accurate instance segmentation.
Article
Computer Science, Artificial Intelligence
Teng Cheng, Lei Sun, Junning Zhang, Jinling Wang, Zhanyang Wei
Summary: This study proposes a scheme that combines the start-stop point signal features for wideband multi-signal detection, called Fast Spectrum-Size Self-Training network (FSSNet). By utilizing start-stop points to build the signal model, this method successfully solves the difficulty of existing deep learning methods in detecting discontinuous signals and achieves satisfactory detection speed.
Article
Computer Science, Artificial Intelligence
Wenming Wu, Xiaoke Ma, Quan Wang, Maoguo Gong, Quanxue Gao
Summary: The layer-specific modules in multi-layer networks are critical for understanding the structure and function of the system. However, existing methods fail to accurately characterize and balance the connectivity and specificity of these modules. To address this issue, a joint learning graph clustering algorithm (DRDF) is proposed, which learns the deep representation and discriminative features of the multi-layer network, and balances the connectivity and specificity of the layer-specific modules through joint learning.
Article
Computer Science, Artificial Intelligence
Guanghui Yue, Guibin Zhuo, Weiqing Yan, Tianwei Zhou, Chang Tang, Peng Yang, Tianfu Wang
Summary: This paper proposes a novel boundary uncertainty aware network (BUNet) for precise and robust colorectal polyp segmentation. BUNet utilizes a pyramid vision transformer encoder to learn multi-scale features and incorporates a boundary exploration module (BEM) and a boundary uncertainty aware module (BUM) to handle boundary areas. Experimental results demonstrate that BUNet outperforms other methods in terms of performance and generalization ability.