Z. Yan received his bachelor degree in automation and computer-aided engineering in 2010 and PhD degree in computational intelligence in 2014, both from The Chinese University of Hong Kong, Hong Kong. He is a Chancellor’s research fellow at the Centre for Artificial Intelligence, University of Technology Sydney. Prior to this position, he was a researcher at Huawei Technologies and a software engineer at Alibaba Group. While working in industry, he oversaw the design and implementation of large-scale deep learning systems which supported a variety of commercial products in cloud computing, computer vision and logistics. His current research interests include neural networks and deep learning, optimization, and control. Two of his publications on memristive neural networks were listed as highly cited papers.
Can supervise: YES
Neural networks, optimization, deep learning, control
Sun, B, Cao, Y, Guo, Z, Yan, Z, Wen, S, Huang, T & Chen, Y 2020, 'Sliding Mode Stabilization of Memristive Neural Networks With Leakage Delays and Control Disturbance', IEEE Transactions on Neural Networks and Learning Systems, pp. 1-10.View/Download from: Publisher's site
Wang, B, Li, T, Yan, Z, Zhang, G & Lu, J 2020, 'DeepPIPE: A distribution-free uncertainty quantification approach for time series forecasting', NEUROCOMPUTING, vol. 397, pp. 11-19.View/Download from: Publisher's site
Wang, L, Yu, Z, Xiong, F, Yang, D, Pan, S & Yan, Z 2020, 'Influence Spread in Geo-Social Networks: A Multiobjective Optimization Perspective.', IEEE Transactions on Cybernetics.View/Download from: Publisher's site
As an emerging social dynamic system, geo-social network can be used to facilitate viral marketing through the wide spread of targeted advertising. However, unlike traditional influence spread problem, the heterogeneous spatial distribution has to incorporated into geo-social network environment. Moreover, from the perspective of business managers, it is indispensable to balance the tradeoff between the objective of influence spread maximization and objective of promotion cost minimization. Therefore, these two goals need to be seamlessly combined and optimized jointly. In this paper, considering the requirements of real-world applications, we develop a multiobjective optimization-based influence spread framework for geo-social networks, revealing the full view of Pareto-optimal solutions for decision makers. Based on the reverse influence sampling (RIS) model, we propose a similarity matching-based RIS sampling method to accommodate diverse users, and then transform our original problem into a weighted coverage problem. Subsequently, to solve this problem, we propose a greedy-based incrementally approximation approach and heuristic-based particle swarm optimization approach. Extensive experiments on two real-world geo-social networks clearly validate the effectiveness and efficiency of our proposed approaches.
Wang, M, Yan, Z, Wang, T, Cai, P, Gao, S, Zeng, Y, Wan, C, Wang, H, Pan, L, Yu, J, Pan, S, He, K, Lu, J & Chen, X 2020, 'Gesture recognition using a bioinspired learning architecture that integrates visual data with somatosensory data from stretchable sensors', NATURE ELECTRONICS.View/Download from: Publisher's site
Wang, S, Cao, Y, Guo, Z, Yan, Z, Wen, S & Huang, T 2020, 'Periodic Event-Triggered Synchronization of Multiple Memristive Neural Networks With Switching Topologies and Parameter Mismatch', IEEE Transactions on Cybernetics, pp. 1-11.View/Download from: Publisher's site
Wen, S, Liu, W, Yang, Y, Zhou, P, Guo, Z, Yan, Z, Chen, Y & Huang, T 2020, 'Multilabel Image Classification via Feature/Label Co-Projection', IEEE Transactions on Systems, Man, and Cybernetics: Systems, pp. 1-10.View/Download from: Publisher's site
Xiong, F, Shen, W, Chen, H, Pan, S, Wang, X & Yan, Z 2020, 'Exploiting Implicit Influence From Information Propagation for Social Recommendation.', IEEE transactions on cybernetics.View/Download from: Publisher's site
Social recommender systems have attracted a lot of attention from academia and industry. On social media, users' ratings and reviews can be observed by all users, and have implicit influence on their future ratings. When these users make subsequent decisions about an item, they may be affected by existing ratings on the item. Thus, implicit influence propagates among the users who rated the same items, and it has significant impact on users' ratings. However, implicit influence propagation and its effect on recommendation rarely have been studied. In this article, we propose an information propagation-based social recommendation method (SoInp) and model the implicit user influence from the perspective of information propagation. The implicit influence is inferred from ratings on the same items. We investigate the concrete effect of implicit user influence in the propagation process and introduce it into recommender systems. Furthermore, we incorporate the implicit user influence and explicit trust information in the matrix factorization framework. To demonstrate the performance, we conduct comprehensive experiments on real-world datasets to compare the proposed method with the state-of-the-art models. The results indicate that SoInp makes notable improvements in rating prediction.
© 2018 Elsevier B.V. This paper proposes a sparse fully convolutional network (FCN) for face labeling. FCN has demonstrated strong capabilities in learning representations for semantic segmentation. However, it often suffers from heavy redundancy in parameters and connections. To ease this problem, group Lasso regularization and intra-group Lasso regularization are utilized to sparsify the convolutional layers of the FCN. Based on this framework, parameters that correspond to the same output channel are grouped into one group, and these parameters are simultaneously zeroed out during training. For the parameters in groups that are not zeroed out, intra-group Lasso provides further regularization. The essence of the regularization framework lies in its ability to offer better feature selection and higher sparsity. Moreover, a fully connected conditional random fields (CRF) model is used to refine the output of the sparse FCN. The proposed approach is evaluated on the LFW face dataset with the state-of-the-art performance. Compared with a non-regularized FCN, the sparse FCN reduces the number of parameters by 91.55% while increasing the segmentation performance by 11% relative error reduction.
In this paper, an improved you only look once (YOLOv3) algorithm is proposed to make the detection effect better and improve the performance of a tennis ball detection robot. The depth-separable convolution network is combined with the original YOLOv3 and the residual block is added to extract the features of the object. The feature map output by the residual block is merged with the target detection layer through the shortcut layer to improve the network structure of YOLOv3. Both the original model and the improved model are trained by the same tennis ball data set. The results show that the recall is improved from 67.70% to 75.41% and the precision is 88.33%, which outperforms the original 77.18%. The recognition speed of the model is increased by half and the weight is reduced by half after training. All these features provide a great convenience for the application of the deep neural network in embedded devices. Our goal is that the robot is capable of picking up more tennis balls as soon as possible. Inspired by the maximum clique problem (MCP), the pointer network (Ptr-Net) and backtracking algorithm (BA) are utilized to make the robot find the place with the highest concentration of tennis balls. According to the training results, when the number of tennis balls is less than 45, the accuracy of determining the concentration of tennis balls can be as high as 80%.
Han, M, Bao, Y, Sun, Z, Wen, S, Xia, L, Zhao, J, Du, J & Yan, Z 2019, 'Automatic Segmentation of Human Placenta Images with U-Net', IEEE Access, vol. 7, pp. 180083-180092.View/Download from: Publisher's site
© 2013 IEEE. Placenta is closely related to the health of the fetus. Abnormal placental function will affect the normal development of the fetus, and in severe cases, even endanger the life of the fetus. Therefore, accurate and quantitative evaluation of placenta has important clinical significance. It is a common method to segment human placenta with semantic segmentation. However, manual segmentation relies too much on the professional knowledge and clinical experience of the staff, and it will also consume a lot of time. Therefore, based on u-net, we propose an automatic segmentation method of human placenta, which reduces manual intervention and greatly speeds up the segmentation, making large-scale segmentation possible. The human placenta data set we used was labeled by experts, which was obtained from prenatal examinations of 11 pregnant women, about 1,110 images. It was a comprehensive and clinically significant data set. By training the network with such data set, the robustness of the model will be better. After testing on the data set, the segmentation effect is basically consistent with the manual segmentation effect.
Li, G & Yan, Z 2019, 'Reconstruction of sparse signals via neurodynamic optimization', International Journal of Machine Learning and Cybernetics, vol. 10, pp. 15-26.View/Download from: Publisher's site
Lu, J, Yan, Z, Han, J & Zhang, G 2019, 'Data-Driven Decision-Making (D3M): Framework, Methodology, and Directions', IEEE Transactions on Emerging Topics in Computational Intelligence, vol. 3, no. 4, pp. 286-296.View/Download from: Publisher's site
© 2017 IEEE. A decision problem, according to traditional principles, is approached by finding an optimal solution to an analytical programming decision model, which is known as model-driven decision-making. The fidelity of the model determines the quality and reliability of the decision-making; however, the intrinsic complexity of many real-world decision problems leads to significant model mismatch or infeasibility in deriving a model using the first principle. To overcome the challenges that are present in the big data era, both researchers and practitioners emphasize the importance of making decisions that are backed up by data related to decision tasks, a process called data-driven decision-making (D3M). By building on data science, not only can decision models be predicted in the presence of uncertainty or unknown dynamics, but also inherent rules or knowledge can be extracted from data and directly utilized to generate decision solutions. This position paper systematically discusses the basic concepts and prevailing techniques in data-driven decision-making and clusters-related developments in technique into two main categories: programmable data-driven decision-making (P-D3M) and nonprogrammable data-driven decision-making (NP-D3M). This paper establishes a D3M technical framework, main methodologies, and approaches for both categories of D3M, as well as identifies potential methods and procedures for using data to support decision-making. It also provides examples of how D3M is implemented in practice and identifies five further research directions in the D3M area. We believe that this paper will directly support researchers and professionals in their understanding of the fundamentals of D3M and of the developments in technical methods.
Wen, S, Xiao, S, Yang, Y, Yan, Z, Zeng, Z & Huang, T 2019, 'Adjusting Learning Rate of Memristor-based Multilayer Neural Networks via Fuzzy Method', IEEE Transactions on Computer - Aided Design of Integrated Circuits and Systems, vol. 38, no. 6, pp. 1084-1094.View/Download from: Publisher's site
IEEE Back Propagation (BP) based on stochastic gradient descent (SGD) is the prevailing method to train multilayer neural networks (MNNs) with hidden layers. However, the existence of the physical separation between memory arrays and arithmetic module makes it inefficient and ineffective to implement BP in conventional digital hardware. Although CMOS may alleviate some problems of the hardware implementation of MNNs, synapses based on complementary metal-oxide-semiconductor (CMOS) cost too much power and areas in very large scale integrated (VLSI) circuits. As a novel device, memristor shows promises to overcome this shortcoming due to its ability to closely integrate processing and memory. This paper proposes a novel circuit for implementing a synapse based on a memristor and two MOSFET tansistors (p-type and n-type). Compared with a CMOS-only circuit, the proposed one reduced the area consumption by 92%-98%. In addition, we develop a fuzzy method for the adjustment of the learning rates of MNNs, which increases the learning accuracy by 2%-3% compared with a constant learning rate. Meanwhile, the fuzzy adjustment method is robust and insensitive to parameter changes due to the approximate reasoning. Furthermore, the proposed methods can be extended to memristor-based multi-layer convolutional neural network for complex tasks. The novel architecture behaves in a human-liking thinking process.
Le, X, Chen, S, Yan, Z & Xi, J 2018, 'A Neurodynamic Approach to Distributed Optimization With Globally Coupled Constraints', IEEE Transactions on Cybernetics, vol. 48, no. 11.View/Download from: Publisher's site
IEEE In this paper, a distributed neurodynamic approach is proposed for constrained convex optimization. The objective function is a sum of local convex subproblems, whereas the constraints of these subproblems are coupled. Each local objective function is minimized individually with the proposed neurodynamic optimization approach. Through information exchange between connected neighbors only, all nodes can reach consensus on the Lagrange multipliers of all global equality and inequality constraints, and the decision variables converge to the global optimum in a distributed manner. Simulation results of two power system cases are discussed to substantiate the effectiveness and characteristics of the proposed approach.
Wen, S, Xie, X, Yan, Z, Huang, T & Zeng, Z 2018, 'General memristor with applications in multilayer neural networks.', Neural networks : the official journal of the International Neural Network Society, vol. 103, pp. 142-149.View/Download from: Publisher's site
Memristor describes the relationship between charge and flux. Although several window functions for memristors based on the HP linear and nonlinear dopant drift models have been studied, most of them are inadequate to capture the full characteristics of memristors. To address this issue, this paper proposes a unified window function to describe a general memristor with restrictions of its parameters given. Compared with other window functions, the proposed function demonstrates high validity and accuracy. In order to make the simulation results have high consistency with the results of actual circuit, we apply the new window function to the simulation of a memristor-based multilayer neural network (MNN) circuit. The overall accuracy will vary with the change of control parameters in the window function. It implies that the proposed model can guide the design of actual memristor-based circuits.
Le, X, Yan, Z & Xi, J 2017, 'A Collective Neurodynamic System for Distributed Optimization with Applications in Model Predictive Control', IEEE Transactions on Emerging Topics in Computational Intelligence, vol. 1, no. 4, pp. 305-314.View/Download from: Publisher's site
© 2017 IEEE. A collective neurodynamic system is presented to distributed convex optimization subject to linear equality and box constraints in framework of an autonomous multiagent network. The overall objective to be minimized takes an additive form of multiple local objective functions. Agents in the system, each of which is modeled by a recurrent neural network, cooperatively, and autonomously develop their dynamic behaviors based on real-time interactions with their neighbors. In specific, each recurrent neural network has knowledge on a local objective only with no access to the overall objective function. It exchanges neuronal state with neighboring agents via a predefined communication topology. Guided by their individual neurodynamics and the collective efforts, all agents reach consensus at the global optimal solution to the distributed convex optimization. Illustrative examples are provided to substantiate the theoretical properties. A case study on model predictive control is reported to further validate the collective neurodynamic system.
Yan, Z, Fan, J & Wang, J 2017, 'A Collective Neurodynamic Approach to Constrained Global Optimization', IEEE Transactions on Neural Networks and Learning Systems, vol. 28, no. 5, pp. 1206-1215.View/Download from: Publisher's site
Global optimization is a long-lasting research topic in the field of optimization, posting many challenging theoretic and computational issues. This paper presents a novel collective neurodynamic method for solving constrained global optimization problems. At first, a one-layer recurrent neural network (RNN) is presented for searching the Karush-Kuhn-Tucker points of the optimization problem under study. Next, a collective neuroydnamic optimization approach is developed by emulating the paradigm of brainstorming. Multiple RNNs are exploited cooperatively to search for the global optimal solutions in a framework of particle swarm optimization. Each RNN carries out a precise local search and converges to a candidate solution according to its own neurodynamics. The neuronal state of each neural network is repetitively reset by exchanging historical information of each individual network and the entire group. Wavelet mutation is performed to avoid prematurity, add diversity, and promote global convergence. It is proved in the framework of stochastic optimization that the proposed collective neurodynamic approach is capable of computing the global optimal solutions with probability one provided that a sufficiently large number of neural networks are utilized. The essence of the collective neurodynamic optimization approach lies in its potential to solve constrained global optimization problems in real time. The effectiveness and characteristics of the proposed approach are illustrated by using benchmark optimization problems.
Yan, Z, Le, X & Wang, J 2016, 'Tube-Based Robust Model Predictive Control of Nonlinear Systems via Collective Neurodynamic Optimization', IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS, vol. 63, no. 7, pp. 4377-4386.View/Download from: Publisher's site
Guo, Z, Wang, J & Yan, Z 2015, 'Global Exponential Synchronization of Two Memristor-Based Recurrent Neural Networks with Time Delays via Static or Dynamic Coupling', IEEE TRANSACTIONS ON SYSTEMS MAN CYBERNETICS-SYSTEMS, vol. 45, no. 2, pp. 235-249.View/Download from: Publisher's site
In this paper, a one-layer recurrent neural network is proposed for solving nonconvex optimization problems subject to general inequality constraints, designed based on an exact penalty function method. It is proved herein that any neuron state of the proposed neural network is convergent to the feasible region in finite time and stays there thereafter, provided that the penalty parameter is sufficiently large. The lower bounds of the penalty parameter and convergence time are also estimated. In addition, any neural state of the proposed neural network is convergent to its equilibrium point set which satisfies the Karush-Kuhn-Tucker conditions of the optimization problem. Moreover, the equilibrium point set is equivalent to the optimal solution to the nonconvex optimization problem if the objective function and constraints satisfy given conditions. Four numerical examples are provided to illustrate the performances of the proposed neural network.
Yan, Z & Wang, J 2015, 'Nonlinear Model Predictive Control Based on Collective Neurodynamic Optimization', IEEE TRANSACTIONS ON NEURAL NETWORKS AND LEARNING SYSTEMS, vol. 26, no. 4, pp. 840-850.View/Download from: Publisher's site
Guo, Z, Wang, J & Yan, Z 2014, 'A systematic method for analyzing robust stability of interval neural networks with time-delays based on stability criteria', NEURAL NETWORKS, vol. 54, pp. 112-122.View/Download from: Publisher's site
Guo, Z, Wang, J & Yan, Z 2014, 'Attractivity Analysis of Memristor-Based Cellular Neural Networks With Time-Varying Delays', IEEE TRANSACTIONS ON NEURAL NETWORKS AND LEARNING SYSTEMS, vol. 25, no. 4, pp. 704-717.View/Download from: Publisher's site
Guo, Z, Wang, J & Yan, Z 2014, 'Passivity and Passification of Memristor-Based Recurrent Neural Networks With Time-Varying Delays', IEEE TRANSACTIONS ON NEURAL NETWORKS AND LEARNING SYSTEMS, vol. 25, no. 11, pp. 2099-2109.View/Download from: Publisher's site
Yan, Z & Wang, J 2014, 'Robust Model Predictive Control of Nonlinear Systems With Unmodeled Dynamics and Bounded Uncertainties Based on Neural Networks', IEEE TRANSACTIONS ON NEURAL NETWORKS AND LEARNING SYSTEMS, vol. 25, no. 3, pp. 457-469.View/Download from: Publisher's site
Guo, Z, Wang, J & Yan, Z 2013, 'Global exponential dissipativity and stabilization of memristor-based recurrent neural networks with time-varying delays', NEURAL NETWORKS, vol. 48, pp. 158-172.View/Download from: Publisher's site
Yan, Z & Wang, J 2012, 'Model Predictive Control for Tracking of Underactuated Vessels Based on Recurrent Neural Networks', IEEE JOURNAL OF OCEANIC ENGINEERING, vol. 37, no. 4, pp. 717-726.View/Download from: Publisher's site
Yan, Z & Wang, J 2014, 'Nonlinear and robust model predictive control of systems with unmodeled dynamics based on supervised learning and neurodynamic optimization' in Liu, D, Alippi, C, Zhao, D & Zhang, H (eds), Frontiers of Intelligent Control and Information Processing, World Scientific Publishing, Singapore, pp. 193-220.View/Download from: Publisher's site
© 2015 by World Scientific Publishing Co. Pte. Ltd. Model predictive control is an optimization-based control strategy which has achieved enormous successes in numerous real-world applications. MPC generates control signals by means of real-time optimization of a performance index over a finite moving horizon of predicted future, subject to system constraints. A major challenge of the MPC research and development lies in the realization of nonlinear and robust MPC approaches, especially to cases where unmodeled dynamics exist. This chapter presents novel MPC approaches to nonlinear systems with unmodeled dynamics based on neural networks. At first, a nonlinear system with unmodeled dynamics is decomposed by means of Jacobian linearization to an affine part and a higher-order unknown term. The linearization residues, together with the unmodeled dynamics, are then modeled by using a feedforward neural network called the Extreme Learning Machine via supervised learning. When the controlled system is affected by bounded additive disturbances, the minimax methodology is exploited to achieve robustness. The nonlinear and robust MPC problems are formulated as constrained convex optimization problems and iteratively solved by applying neurodynamic optimization approaches. The applied neurodynamic optimization approaches can compute the optimal control signals in real-time, which shed a light for real-time implementability of MPC technology. Simulation results are provided to substantiate the effectiveness and characteristics of the proposed approaches.
Wang, B, Luo, H, Lu, J, Li, T, Zhang, G, Yan, Z & Zheng, Y 2019, 'Deep uncertainty quantification: A machine learning approach for weather forecasting', Proceedings of the ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, ACM SIGKDD International Conference on Knowledge Discovery & Data Mining, ACM, Anchorage, USA, pp. 2087-2095.View/Download from: Publisher's site
© 2019 Association for Computing Machinery. Weather forecasting is usually solved through numerical weather prediction (NWP), which can sometimes lead to unsatisfactory performance due to inappropriate setting of the initial states. In this paper, we design a data-driven method augmented by an effective information fusion mechanism to learn from historical data that incorporates prior knowledge from NWP. We cast the weather forecasting problem as an end-to-end deep learning problem and solve it by proposing a novel negative log-likelihood error (NLE) loss function. A notable advantage of our proposed method is that it simultaneously implements single-value forecasting and uncertainty quantification, which we refer to as deep uncertainty quantification (DUQ). Efficient deep ensemble strategies are also explored to further improve performance. This new approach was evaluated on a public dataset collected from weather stations in Beijing, China. Experimental results demonstrate that the proposed NLE loss significantly improves generalization compared to mean squared error (MSE) loss and mean absolute error (MAE) loss. Compared with NWP, this approach significantly improves accuracy by 47.76%, which is a state-of-the-art result on this benchmark dataset. The preliminary version of the proposed method won 2nd place in an online competition for daily weather forecasting1
Wang, B, Yan, Z, Lu, J & Zhang, G 2018, 'Explore Uncertainty in Residual Networks for Crowds Flow Prediction', International Joint Conference on Neural Networks, IEEE, Rio de Janeiro, Brazil, pp. 1-7.View/Download from: Publisher's site
The residual network has witnessed a great success in computer vision particularly on classification tasks, however, it has not been well studied in regression. In this work, we show its competence in a regression task - crowds flow prediction, which has strong implication to city safety and management. The problem of crowds flow prediction is challenging due to its fast dynamics. To address this issue, we explore residual learning with Gaussian regularization and propose a novel convolutional neural network called Gaussian noise residual networks (Noise-ResNet). Compared with the benchmark ST-ResNet on crowds flow prediction, the proposed architecture has three advantages: 1) Superior performance. Especially, it attains the state-of-the-art results on benchmark dataset BikeNYC. 2) Light architecture. Noise-ResNet only utilises one residual unit rather than STResNet with multiple ones, which greatly reduces the training time. 3) Interpretable input sequences. Noise-ResNet takes an input sequence that only considers the most important periodic data and closeness data, which makes the learning process more interpretable. Furthermore, experimental results substantiate that the Noise-ResNet can outperform ResNet with dropout on the same regression task.
Wang, B, Yan, Z, Lu, J, Zhang, G & Li, T 2018, 'Deep multi-task learning for air quality prediction', Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), International Conference on Neural Information Processing, Springer Link, Siem Reap, Cambodia, pp. 93-103.View/Download from: Publisher's site
© 2018, Springer Nature Switzerland AG. Predicting the concentration of air pollution particles has been an important task of urban computing. Accurately measuring and estimating makes the citizen and governments can behave with suitable decisions. In order to predict the concentration of several air pollutants at multiple monitoring stations throughout the city region, we proposed a novel deep multi-task learning framework based on residual Gated Recurrent Unit (GRU). The experimental results on the real world data from London region substantiate that the proposed deep model has manifest superiority than shallow models and outperforms 9 baselines.
Wang, B, Yan, Z, Lu, J, Zhang, G & Li, T 2018, 'Road traffic flow prediction using deep transfer learning', Data Science and Knowledge Engineering for Sensing Decision Support, Conference on Data Science and Knowledge Engineering for Sensing Decision Support (FLINS 2018), World Scientific, Belfast, Northern Ireland, pp. 331-338.View/Download from: Publisher's site
Traffic flow prediction is a long-standing problem. Over the recent years, deep learning has gradually achieved a satisfying success on this task, but it depends on abundant historical traffic data. A realistic problem is that some new-established transportation networks only have few data which is not enough to train a robust deep learning model. To address this problem, we first explore and apply the transfer learning and fine-tuning to the field of transportation and propose a novel transferable traffic deep learning model, called TT-DL which can predict real-time traffic flow in data-strapped roads by transferring knowledge from data-rich roads. Our experimental results show that transfer learning is better than any other initialization methods. This indicates that traffic network has its special structure and there exists transferable knowledge between different traffic areas.
Yan, Z, Le, X, Wen, S & Lu, J 2018, 'A continuous-time recurrent neural network for sparse signal reconstruction via ℓ1 minimization', 8th International Conference on Information Science and Technology, ICIST 2018, International Conference on Information Science and Technology, IEEE, Cordoba, Spain, pp. 43-49.View/Download from: Publisher's site
© 2018 IEEE. This paper presents a neurodynamic model for solving e1 minimization problems for sparse signal reconstruction. The essence of the proposed approach lies in its capability to operate in continuous time, which enables it to outperform most existing iterative e1-solvers in dynamic environments. The model is described by a goal-seeking recurrent neural network and it evolves according to its deterministic neurodynamics. It is proved that the model globally converges to the optimal solution to the e1-minimization problem under study. The connection weights of the neural network model are determined by using subgradient projection methods and the activation function is designed based on subdifferential. Due to its simple structure, the hardware implementation of this neurodynamic model is viable and cost-effective, which sheds light on real-time sparse signal recovery via large scale e1 minimization formulations.
Yan, Z, Lu, J & Zhang, G 2018, 'Distributed model predictive control of linear systems with coupled constraints based on collective neurodynamic optimization', AI 2018: AI 2018: Advances in Artificial Intelligence (LNAI), Australasian Joint Conference on Artificial Intelligence, Spinger, Wellington, New Zealand, pp. 318-328.View/Download from: Publisher's site
© Springer Nature Switzerland AG 2018. Distributed model predictive control explores an array of local predictive controllers that synthesize the control of subsystems independently yet they communicate to efficiently cooperate in achieving the closed-loop control performance. Distributed model predictive control problems naturally result in sequential distributed optimization problems that require real-time solution. This paper presents a collective neurodynamic approach to design and implement the distributed model predictive control of linear systems in the presence of globally coupled constraints. For each subsystem, a neurodynamic model minimizes its cost function using local information only. According to the communication topology of the network, neurodynamic models share information to their neighbours to reach consensus on the optimal control actions to be carried out. The collective neurodynamic models are proven to guarantee the global optimality of the model predictive control system.
Ren, G, Wen, S, Yan, Z, Hu, R, Zeng, Z & Cao, Y 2016, 'Power load forecasting based on support vector machine and particle swarm optimization', Proceedings of the World Congress on Intelligent Control and Automation (WCICA), pp. 2003-2008.View/Download from: Publisher's site
© 2016 IEEE. Accurate electric load forecasting is significant for the operation of the power systems and electricity markets. This paper proposes a particle swarm optimization with support vector machine (PSOSVM) to forecast annual power load. Based on radial basis function, support vector machine (SVM) is utilized to determine the structure and initial values of the parameters. Then, particle swarm optimization (PSO) is employed to optimize the parameters of the SVM model. In order to utilize the proposed method, practical data are divided into two parts, one is for training, the other is for testing. The combined method, PSOSVM, can effectively predict annual power load.
Bi, S, Zhang, G, Xue, X & Yan, Z 2015, 'Real-time robust model predictive control of mobile robots based on recurrent neural networks', Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), International Conference on Neural Information Processing, Springer, Istanbul, Turkey, pp. 289-296.View/Download from: Publisher's site
© Springer International Publishing Switzerland 2015. This paper presents a novel model predictive control (MPC) approach to tracking control of mobile robots based on recurrent neural networks (RNNs). The tracking control problem is firstly formulated as a sequential dynamic optimization problem in framework of MPC. Then a novel neurodynamic approach is developed for computing the optimal control signals in real time, where multiple RNNs are applied in a collective fashion. The proposed approach enables MPC of mobile robots to be synthesized in real time. Simulation results are provided to substantiate the effectiveness of the proposed approach.
Yan, Z, Bi, S & Xue, X 2015, 'Global exponential anti-synchronization of coupled memristive chaotic neural networks with time-varying delays', Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), International Symposium on Neural Networks, Springer, Jeju, South Korea, pp. 192-201.View/Download from: Publisher's site
© Springer International Publishing Switzerland 2015. This paper investigates the problem of global exponential anti-synchronization of a class of memristive chaotic neural networks with time-varying delays. First, a memrsitive neural network is modeled. Then, considering the state-dependent properties of the memristor, a new fuzzy model employing parallel distributed compensation (PDC) provides a new way to analyze the complicated memristive neural networks with only two subsystems. And the controller is dependent on the output of the system in the case of packed circuits. An illustrative example is also presented to show the effectiveness of the results.
Zhang, P, Yan, Z & Wang, J 2014, 'Obstacle and singularity avoidance for kinematically redundant manipulators based on neurodynamic optimization', 5th International Conference on Intelligent Control and Information Processing, ICICIP 2014 - Proceedings, International Conference on Intelligent Control and Information Processing, IEEE, Dalian, China, pp. 460-465.View/Download from: Publisher's site
© 2014 IEEE. With wide applications of kinematically redundant manipulators in robotics, obstacle and singularity avoidance emerge as critical issues to be addressed. Correspondingly, three problems have to be considered, including the determination of critical points on a given manipulator, the computation of joint velocities using inverse kinematics, and the analysis of singularity caused by configurations of manipulators. In this paper, these tasks are formulated as a convex quadratic programming (QP) subject to equality and inequality constraints with time-varying parameters where physical constraints such as joint physical limits are also incorporated directly into the formulation. To solve the QP problem in real time, a recurrent neural network called the improved dual neural network is applied, which has lower structural complexity compared with existing neural networks for solving this particular problem. The effectiveness of the proposed approaches is demonstrated by simulation results based on the Mitsubishi PA10-7C manipulator.
Le, X, Wang, J & Yan, Z 2014, 'Neurodynamics-based robust pole assignment for synthesizing second-order control systems via output feedback based on a convex feasibility problem reformulation', INISTA 2014 - IEEE International Symposium on Innovations in Intelligent Systems and Applications, Proceedings, IEEE International Symposium on Innovations in Intelligent Systems and Applications, pp. 1-6.View/Download from: Publisher's site
A neurodynamic optimization approach is proposed for robust pole assignment problem of second-order control systems via output feedback. With a suitable robustness measure serving as the objective function, the robust pole assignment problem is formulated as a quasi-convex optimization problem with linear constraints. Next, the problem further is reformulated as a convex feasibility problem. Two coupled recurrent neural networks are applied for solving the optimization problem with guaranteed optimality and exact pole assignment. Simulation results are included to substantiate the effectiveness of the proposed approach. © 2014 IEEE.
Le, X, Yan, Z & Wang, J 2013, 'A neurodynamic optimization approach to robust pole assignment for synthesizing piecewise linear control systems', 2013 IEEE International Conference on Information and Automation, ICIA 2013, IEEE International Conference on Information and Automation, IEEE, Yinchuan, China, pp. 1334-1339.View/Download from: Publisher's site
This paper presents a neurodynamic optimization approach to robust pole assignment for synthesis of piecewise linear control systems via state feedback. The robust pole assignment is formulated as a pseudoconvex optimization problem with linear equality constraints where a robustness measure is considered as the objective function. The robustness is achieved by means of minimizing the spectral condition number of the closed-loop eigensystem. Two recurrent neural networks with guaranteed global convergence are applied for solving the optimization problem in real time. Simulation results are included to substantiate the effectiveness and demonstrate the characteristics of the proposed approach. © 2013 IEEE.
Le, X, Yan, Z & Wang, J 2014, 'Neurodynamics-based robust eigenstructure assignment for second-order descriptor systems', Proceedings of the International Joint Conference on Neural Networks, IEEE International Joint Conference on Neural Networks, IEEE, China, pp. 2770-2775.View/Download from: Publisher's site
© 2014 IEEE. In this paper, a neurodynamic optimization approach is proposed for robust eigenstructure assignment problem of second-order descriptor systems via state feedback control. With a novel robustness measure serving as the objective function, the robust eigenstructure assignment problem is formulated as a pseudoconvex optimization problem. Two coupled recurrent neural networks are applied for solving the optimization problem with guaranteed optimality and exact pole assignment. Simulation results are included to substantiate the effectiveness of the proposed approach.
Shan, Y, Yan, Z & Wang, J 2013, 'Model predictive control of underwater gliders based on a one-layer recurrent neural network', 2013 6th International Conference on Advanced Computational Intelligence, ICACI 2013 - Proceedings, International Conference on Advanced Computational Intelligence, IEEE, Hangzhou, China, pp. 328-333.View/Download from: Publisher's site
In this paper, a motion control problem for underwater gilders in longitudinal plane is considered. A recurrent neural network based model predictive control approach is developed. The model predictive control of underwater gliders is formulated as a time-varying constrained quadratic programming problem, which is solved by using a recurrent neural network called the simplified dual network in real-time. Simulation results are further presented to show the effectiveness and performance of the proposed model predictive control approach. © 2013 IEEE.
Wang, X, Yan, Z & Wang, J 2014, 'Model predictive control of multi-robot formation based on the simplified dual neural network', Proceedings of the International Joint Conference on Neural Networks, IEEE International Joint Conference on Neural Networks, IEEE, China, pp. 3161-3166.View/Download from: Publisher's site
© 2014 IEEE. This paper is concerned with formation control problems of multi-robot systems in framework of model predictive control. The formation control of robots herein is based on the leader-follower scheme. The followers are controlled by torques to track the desired trajectories to form and keep a formation. A model predictive control approach is proposed for solving the formation control problem, where the control problem is formulated as a dynamic quadratic optimization problem. A one-layer recurrent neural network called the simplified dual network is applied for computing the optimal control input in real time. Simulation results substantiate that the formation of robots can be well controlled by the proposed approach.
Yan, Z, Le, X & Wang, J 2014, 'Model predictive control of linear parameter varying systems based on a recurrent neural network', Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), International Conference on Theory and Practice of Natural Computing, Springer, Granada, Spain, pp. 255-266.
© Springer International Publishing Switzerland 2014. This paper presents a model predictive control approach to discrete-time linear parameter varying systems based on a recurrent neural network. The model predictive control problem is formulated as a sequential convex optimization, and it is solved by using a recurrent neural network in real time. The essence of the proposed approach lies in its real-time computational capability with extended applicability. Simulation results are provided to substantiate the effectiveness of the proposed model predictive control approach.
Yan, Z, Wang, J & Fan, J 2014, 'Machine-cell and part-family formation in cellular manufacturing using a two-phase clustering algorithm', IFAC Proceedings Volumes (IFAC-PapersOnline), International Federation of Automatic Control World Congress, IFAC, Cape Town, pp. 2605-2610.View/Download from: Publisher's site
© IFAC. This paper presents a two-phase clustering algorithm for machine-cell and part-family formation in the design of cellular manufacturing systems. The proposed algorithm begins with the determination of initial cluster centers via a linear assignment method using the least similar group representatives in its first phase. A fuzzy C-means clustering method is followed in its second phase for part-family and machine-cell formation using the obtained initial cluster centers. The two-phase algorithm can remedy the problem of clustering inconsistency resulting from the fuzzy C-means method with random initializations. Experimental results on many benchmark data sets based on multiple performance criteria substantiate the effectiveness of the proposed algorithm.
Yan, Z & Wang, J 2013, 'Stochastic model predictive control of Markov jump linear systems based on a two-layer recurrent neural network', 2013 IEEE International Conference on Information and Automation, ICIA 2013, IEEE International Conference on Information and Automation, IEEE, Yinchuan, China, pp. 564-569.View/Download from: Publisher's site
This paper presents a stochastic model predictive control approach to constrained Markov jump linear systems based on neurodynamic optimization. The stochastic model predictive control problem is formulated as a nonlinear convex optimization problem, which is iteratively solved by using a two-layer recurrent neural network in real-time. The applied neural network can globally converge to the exact optimal solution of the optimization problem. Simulation results are provided to demonstrate the effectiveness and characteristics of the proposed approach. © 2013 IEEE.
Han, WT, Yan, Y & Yan, Z 2012, 'Control method of 4WS based on neural network', World Automation Congress Proceedings.
This paper introduces an active 4WS control method based on neural network. The method takes nonlinear dynamic characteristic of vehicle and tire into account. And discerns them by neural network method according to those actual survey data come from real vehicle. It shows that it has a good control property and can improve the safety and handling stability of vehicle effectively. © 2012 TSI Press.
Yan, Z & Wang, J 2012, 'A neurodynamic approach to bicriteria model predictive control of nonlinear affine systems based on a Goal Programming formulation', Proceedings of the International Joint Conference on Neural Networks.View/Download from: Publisher's site
This paper presents a neurodynamic approach to bicriteria model predictive control (MPC) of nonlinear affine systems based on a goal programming formulation. Bicriteria MPC refers to finding optimal control inputs that minimizes two performance indexes corresponding to tracking errors and control efforts. The bicriteria MPC is formulated as the solution to a nonlinear optimization problem via goal programming technique and is solved by using a two-layer recurrent neural network. Simulation results are included to illustrate the effectiveness of the proposed approach. © 2012 IEEE.
Yan, Z & Wang, J 2012, 'A neurodynamic approach to model predictive control of piecewise linear systems', ICICIP 2012 - 2012 3rd International Conference on Intelligent Control and Information Processing, pp. 463-468.View/Download from: Publisher's site
This paper presents a neurodynamic approach to model predictive control (MPC) of constrained piecewise linear systems. A novel procedure for estimating uncertain system parameters of piecewise linear systems is proposed. The model predictive control problem is then formulated as a quadratic optimization problem. To realize the real-time optimization in MPC, a one-layer recurrent neural network is employed for solving the quadratic optimization problem during each sampling interval. The overall MPC approach is of low computational complexity. Simulation results are included to substantiate the effectiveness and usefulness of the proposed approach. © 2012 IEEE.
Yan, Z, Chung, SF & Wang, J 2012, 'Model predictive control of autonomous underwater vehicles based on the simplified dual neural network', Conference Proceedings - IEEE International Conference on Systems, Man and Cybernetics, IEEE International Conference on Systems, Man and Cybernetics, IEEE, Institute of Electrical and Electronics Engineers, Seoul, South Korea, pp. 2551-2556.View/Download from: Publisher's site
Based on a recurrent neural network, a model predictive control (MPC) method for control of a class of autonomous underwater vehicles (AUVs) is presented. A coupled nonlinear kinematic model with constrains is considered. The model predictive control problem of AUVs is formulated as a time-varying quadratic programming problem, and a one-layer recurrent neural network called the simplified dual network is applied for real-time optimization. It is able to converge to the global optimal solution of the constrained optimization problem. Simulation results are discussed to demonstrate the effectiveness and characteristics of the proposed model predictive control method. © 2012 IEEE.
Yan, Z & Wang, J 2011, 'A neural network approach to nonlinear model predictive control', IECON Proceedings (Industrial Electronics Conference), Annual Conference of the IEEE Industrial Electronics Society, IEEE, Melbourne, VIC, Australia, pp. 2305-2310.View/Download from: Publisher's site
This paper proposes a neural network approach to nonlinear model predictive control (NMPC). The NMPC problem is formulated as a convex programming problem via Jacobain linearization. The unknown high-order term associated with the linearization is estimated by using a feedforward neural network via supervised learning. The convex optimization problem involved in MPC is solved by using a recurrent neural network. Simulation results are provided to demonstrate the performance of the approach. © 2011 IEEE.
Yan, Z & Wang, J 2011, 'Model predictive control of nonlinear affine systems based on the general projection neural network and its application to a continuous stirred tank reactor', 2011 International Conference on Information Science and Technology, ICIST 2011, International Conference on Information Science and Technology, IEEE, Nanjing, China, pp. 1011-1015.View/Download from: Publisher's site
Model predictive control (MPC) is an advanced technique for process control. It is based on iterative, finite horizon optimization of a cost function associated with a plant model. Neural network is an effective approach for on-line optimization problems. In this paper, we apply the general projection neural network for MPC of nonlinear affine systems. Continuous stirred tank reactor (CSTR) system is a typical chemical reactor widely used in chemical industry and can be characterized as a nonlinear affine system. The general projection neural network based MPC is applied to the CSTR problem with input and output constraints. This application demonstrates the usefulness and effectiveness of proposed MPC approach to industrial problems. © 2011 IEEE.
Yan, Z & Wang, J 2011, 'Robust model predictive control of nonlinear affine systems based on a two-layer recurrent neural network', 2014 International Joint Conference on Neural Networks (IJCNN), IEEE International Joint Conference on Neural Networks, IEEE, San Jose, CA, USA, pp. 24-29.View/Download from: Publisher's site
A robust model predictive control (MPC) method is proposed for nonlinear affine systems with bounded disturbances. The robust MPC technique requires on-line solution of a minimax optimal control problem. The minimax strategy means that worst-case performance with respect to uncertainties is optimized. The minimax optimization problem involved in robust MPC is reformulated to a minimization problem and then is solved by using a two-layer recurrent neural network. Simulation examples are included to illustrate the effectiveness of the proposed method. © 2011 IEEE.