Search

A novel adaptive controller for the half-car active suspension systems (ASSs), which can improve the riding comfortability and handling stability of the driver, is proposed in this paper. By using nonlinear mapping, it is demonstrated that the nonlinear ASSs with partial performance constraints are transformed into the novel pure-feedback systems without constraints. By introducing a modified dynamic surface control (DSC) into the Lyapunov function, the adaptive neural network (NN) controller is discussed. The unknown continuous functions are estimated by the NNs, and the boundedness of all signals in the closed-loop systems is guaranteed by the Lyapunov stability theory. Meanwhile, the performance constraints are not violated. Finally, the simulations are performed to clarify and verify the effectiveness of the proposed scheme. [ABSTRACT FROM AUTHOR]

An adaptive control-based neural network for a n -link robot is studied and the considered robot can be transformed as a class of multi-input–multioutput systems. The position of the robot or the output of the transformed systems is constrained in a time-varying compact set. It is commonly known that the constant constraint belongs to a special case of the time-varying constraint, and thus, it can be more general for handling practical problem as compared with the existing methods for robot. The neural approximation is used to estimate the unknown functions of systems and the time-varying barrier Lyapunov function is used to overcome the violation of constraints. It can prove the stability of the closed-loop systems by using Lyapunov analysis. The feasibility of the approach is demonstrated by performing a simulation example. [ABSTRACT FROM PUBLISHER]

In this paper, a stable reinforcement learning control approach using neural networks (NNs) is developed for the trajectory tracking of an n-link rigid robot manipulator. The considered systems are in discrete time form. The proposed controller design consists of two NNs. One is the critic network that is used to approximate the long-term cost function, whereas other is that the action NN is employed to generate the system input. Then, an optimal control input can be obtained compared with other robot manipulator systems. Using the Lyapunov approach, the tracking error and weight estimates are proven to be semi-global uniformly ultimately bounded. A simulation example is employed to illustrate the effectiveness of the proposed controller. [ABSTRACT FROM PUBLISHER]

In this paper, an adaptive control algorithm is proposed for a class of robot manipulator systems with unknown functions and dead-zone input by using a reinforcement learning scheme. The parameters of the dead zone are supposed to be unknown but bounded. The unknown functions can be approximated based on the neural networks, which is one part of the reinforcement learning scheme, namely an action network. The other part is called critic network which is used to approximate the reinforcement signal. Then, the prominent advantage of the proposed approach is that an optimal control input can be obtained by using two networks compared with the results of robot manipulator with dead zone: an additional term is given to compensate for the effect of the dead zone, and a special design procedure to solve the difficulties in constructing the controllers and adaptation laws. Based on the Lyapunov analysis theory, all the signals of the closed-loop system are proved to be bounded and the system output can track the reference signal to a bounded compact set. Finally, a simulation example is given to illustrate the effectiveness of the approach. [ABSTRACT FROM AUTHOR]

In this article, a direct adaptive neural networks control algorithm is presented for a class of SISO discrete-time systems with non-symmetric dead-zone. The property of the dead-zone is discretized. Mean value theorem is used to transform the systems into a special form. The unknown functions in the input–output model are approximated using the radial basis function neural networks. Compared with the results for the discrete non-symmetric dead-zone, this article presents a new algorithm to reduce the computational burden. Lyapunov analysis method is utilized to prove that all the signals in the closed-loop systems are semi-global uniformly ultimately bounded. The tracking error is proved to converge to a small set around the zero. A simulation example provided to illustrate the effectiveness of the control schemes. [ABSTRACT FROM AUTHOR]

A robust adaptive NN output feedback control is proposed to control a class of uncertain discrete-time nonlinear multi-input-multi-output (MIMO) systems. The high-order neural networks are utilized to approximate the unknown nonlinear functions in the systems. Compared with the previous research for discrete-time MIMO systems, robustness of the proposed adaptive algorithm is obvious improved. Using Lyapunov stability theorem, the results show all the signals in the closed-loop system are semi-globally uniformly ultimately bounded, and the tracking errors converge to a small neighborhood of zero by choosing the design parameters appropriately. [ABSTRACT FROM AUTHOR]

In this paper, a direct adaptive neural network control algorithm based on the backstepping technique is proposed for a class of uncertain nonlinear discrete-time systems in the strict-feedback form. The neural networks are utilized to approximate unknown functions, and a stable adaptive neural network controller is synthesized. The fact that all the signals in the closed-loop system are semi-globally uniformly ultimately bounded is proven and the tracking error can converge to a small neighborhood of zero by choosing the design parameters appropriately. Compared with the previous research for discrete-time systems, the proposed algorithm improves the robustness of the systems. A simulation example is employed to illustrate the effectiveness of the proposed algorithm. [ABSTRACT FROM AUTHOR]

An adaptive output feedback control is studied for uncertain nonlinear single-input–single-output systems with partial unmeasured states. In the scheme, a reduced-order observer (ROO) is designed to estimate those unmeasured states. By employing radial basis function neural networks and incorporating the ROO into a new backstepping design, an adaptive output feedback controller is constructively developed. A prominent advantage is its ability to balance the control action between the state feedback and the output feedback. In addition, the scheme can be still implemented when all the states are not available. The stability of the closed-loop system is guaranteed in the sense that all the signals are semiglobal uniformly ultimately bounded and the system output tracks the reference signal to a bounded compact set. A simulation example is given to validate the effectiveness of the proposed scheme. [ABSTRACT FROM AUTHOR]

This brief studies an adaptive neural output feedback tracking control of uncertain nonlinear multi-input–multi-output (MIMO) systems in the discrete-time form. The considered MIMO systems are composed of n subsystems with the couplings of inputs and states among subsystems. In order to solve the noncausal problem and decouple the couplings, it needs to transform the systems into a predictor form. The higher order neural networks are utilized to approximate the desired controllers. By using Lyapunov analysis, it is proven that all the signals in the closed-loop system is the semi-globally uniformly ultimately bounded and the output errors converge to a compact set. In contrast to the existing results, the advantage of the scheme is that the number of the adjustable parameters is highly reduced. The effectiveness of the scheme is verified by a simulation example. [ABSTRACT FROM AUTHOR]

Abstract: Based on the backstepping technique, a direct adaptive neural network control algorithm is proposed for a class of uncertain nonlinear discrete-time systems in the strict-feedback form. Neural networks are utilized to approximate unknown functions, and a stable adaptive neural backstepping controller is synthesized. It is proven that all the signals in the closed-loop system are semi-globally uniformly ultimately bounded (SGUUB) and the tracking error converges to a small neighborhood of zero by choosing the design parameters appropriately. Compared with the existing results for discrete-time systems, the proposed algorithm needs only less parameters to be adjusted online, therefore, it can reduce online computation burden. A simulation example is employed to illustrate the effectiveness of the proposed algorithm. [ABSTRACT FROM AUTHOR]

This article extends the application of the adaptive neural network control to a new class of uncertain MIMO non-linear systems, which are composed of interconnected subsystems where each interconnected subsystem is in the non-affine pure-feedback form. Because both the variables which are used as virtual controllers and the actual controllers appear non-linearly in unknown functions of the MIMO systems, thus, this class of systems is difficult to control. The radial basis function neural networks are utilised to approximate the desired virtual controllers and the desired actual controllers which are obtained by using implicit function theorem. The salient property of the proposed approach is that the number of the adjustable parameters is less than the numerous alternative approaches existing in the literature. It is proven that, under appropriate assumptions, all the signals in the closed-loop system are uniformly bounded and the tracking errors converge to a small neighbourhood of the origin by appropriately choosing design parameters. The feasibility of the developed approach is verified by two simulation examples. [ABSTRACT FROM AUTHOR]

Abstract: We present a systematic design procedure for synchronization of neural networks subject to time-varying delays and sector nonlinearity in the control input. Based on the drive-response concept and the Lyapunov stability theorem, a memoryless decentralized control law is proposed which guarantees exponential synchronization even when input nonlinearity is present. The supplementary requirement that the time-derivative of time-varying delays must be smaller than one is released for the proposed control scheme. A four-dimensional Hopfield neural network with time-varying delays is presented as the illustrative example to demonstrate the effectiveness of the proposed synchronization scheme. [Copyright &y& Elsevier]

The leader‐following consensus problem of a class of non‐strict feedback multi‐agent systems with unknown disturbances, this paper develops a novel adaptive event‐triggered control strategy ground on the consideration of the full state constraints. Combining the dynamic surface control and event triggering mechanism, the unknown external disturbances are estimated by designing a disturbance observer. In order to deal with unknown functions, this paper employs the neural networks. On purpose of keeping the output variable within a constraint boundary, the barrier Lyapunov function is chosen during the design course. Then, through the backstepping method and the Lyapunov stability theorem, it is proved that all signals in the closed‐loop systems are bounded, and the control performance of the closed‐loop system is ensured by selecting design parameters reasonably. Eventually, the availability of the control strategy is guaranteed through experimental example. [ABSTRACT FROM AUTHOR]

As one of the most commonly used acoustic systems in seabed surveys, the altitude of the side scan sonar from the seafloor is always difficult to determine, especially when raw signal levels and gain information are unavailable. The inaccurate sonar altitudes would limit the applications of sonar image geocoding, target detection, and sediment classification. The sonar altitude can be obtained by using bottom tracking methods, but traditional methods often require manual thresholds or complex post-processing procedures, which cannot ensure accurate and real-time bottom tracking. In this paper, a real-time bottom tracking method of side scan data is proposed based on a one-dimensional convolution neural network. First, according to the characteristics of side scan backscatter strength sequences, positive (bottom sequences) and negative (water column and seabed sequences) samples are extracted to establish the sample sets. Second, a one-dimensional convolution neural network is carefully designed and trained by using the sample set to recognize the bottom sequences. Third, a complete processing procedure of the real-time bottom tracking method is established by traversing each side scan ping data and recognizing the bottom sequences. The auxiliary methods for improving real-time performance and sample data augmentation are also explained in detail. The proposed method is implemented on the measured side scan data from the marine area in Meizhou Bay. The trained network model achieves a 100% recognition of the initial sample set as well as 100% bottom tracking accuracy of the training survey line. The average bottom tracking accuracy of the testing survey lines excluding missed pings reaches 99.2%. By comparison with multi-beam bathymetric data and the statistical analysis of real-time performance, the experimental results prove the validity and accuracy of the proposed real-time bottom tracking method. [ABSTRACT FROM AUTHOR]

In this paper, we design an adaptive neural network (NN) controller of uncertain ${n}$ -joint robotic systems with time-varying state constraints. By proposing a nonlinear mapping, the robotic systems are transformed into the multiple-input, multiple-output systems. Compared with constant constraints, the time-varying state constraints are more general in the real systems. To overcome the design challenge, the time-varying barrier Lyapunov function is introduced to ensure that the states of the robotic systems are bounded within the predetermined time-varying range. The NN approximations are employed to approximate the uncertain parametric and unknown functions in the robotic systems. Based on the Lyapunov analysis, it can be proved that all signals of robotic systems are bounded; the tracking errors of system output converge on a small neighborhood of zero and the time-varying state constraints are never violated. Finally, a simulation example is performed to demonstrate the feasibility of the proposed approach. [ABSTRACT FROM AUTHOR]

This paper proposes a novel intelligent fault diagnosis method to automatically identify different health conditions of wind turbine (WT) gearbox. Unlike traditional approaches, where feature extraction and classification are separately designed and performed, this paper aims to automatically learn effective fault features directly from raw vibration signals while classify the type of faults in a single framework, thus providing an end-to-end learning-based fault diagnosis system for WT gearbox without additional signal processing and diagnostic expertise. Considering the multiscale characteristics inherent in vibration signals of a gearbox, a new multiscale convolutional neural network (MSCNN) architecture is proposed to perform multiscale feature extraction and classification simultaneously. The proposed MSCNN incorporates multiscale learning into the traditional CNN architecture, which has two merits: 1) high-level fault features can be effectively learned by the hierarchical learning structure with multiple pairs of convolutional and pooling layers; and 2) multiscale learning scheme can capture complementary and rich diagnosis information at different scales. This greatly improves the feature learning ability and enables better diagnosis performance. The proposed MSCNN approach is evaluated through experiments on a WT gearbox test rig. Experimental results and comprehensive comparison analysis with respect to the traditional CNN and traditional multiscale feature extractors have demonstrated the superiority of the proposed method. [ABSTRACT FROM AUTHOR]

In this paper, an adaptive neural network (NN)-based tracking control algorithm is proposed for the wheeled mobile robotic (WMR) system with full state constraints. It is the first time to design an adaptive NN-based control algorithm for the dynamic WMR system with full state constraints. The constraints come from the limitations of the wheels’ forward speed and steering angular velocity, which depends on the motors’ driving performance. By employing adaptive NNs and a barrier Lyapunov function with error variables, then, the unknown functions in the systems are estimated, and the constraints are not violated. Based on the assumptions and lemmas given in this paper and the references, while the design and the system parameters chose properly, our proposed scheme can guarantee the uniform ultimate boundedness for all signals in the WMR system, and the tracking error converge to a bounded compact set to zero. The numerical experiment of a WMR system is presented to illustrate the good performance of the proposed control algorithm. [ABSTRACT FROM PUBLISHER]

In this paper, based on the discrete Wirtinger inequality, a novel summation inequality is established which extends the Jensen inequality. By the technique of the novel inequality, a sufficient criterion on asymptotic stability of discrete neural networks with time-varying delays is obtained. A numerical example is given to demonstrate the effectiveness of the inequality. [ABSTRACT FROM AUTHOR]

We investigate state estimation for a class of discrete-time recurrent neural networks with leakage delay and time-varying delay. The design method for the state estimator to estimate the neuron states through available output measurements is given. A novel delay-dependent sufficient condition is obtained for the existence of state estimator such that the estimation error system is globally asymptotically stable. Based a novel double summation inequality and reciprocally convex approach, an improved stability criterion is obtained for the error-state system. Two numerical examples are given to demonstrate the effectiveness of the proposed design methods. The simulation results show that the leakage delay has a destabilizing influence on a neural network system. [ABSTRACT FROM AUTHOR]

In this paper, an adaptive neural network tracking control is studied for a class of uncertain nonlinear systems. The studied systems are in discrete-time form and unknown time-delay is considered here. Up to now, the research works on nonlinear discrete-time main focus on systems without time-delay, so the problem of the unknown time-delay will be solved in this paper. Based on the Lipschitz or norm-boundedness assumption of the unknown nonlinearities, the mean-value theorem is utilized to solve the unknown time-delay problem. In order to overcome the noncausal problem, the strict-feedback systems will be transformed into a special form. The radial basis functions neural networks (RBFNN) are utilized to approximate the unknown functions of the systems, the adaptation laws and the controllers are designed based on the transformed systems. By using the Lyapunov analysis, it is proven that the closed-loop system is stable in the sense that semi-globally uniformly ultimately bounded (SGUUB) and the output tracking errors converge to a bounded compact set. A simulation example is used to illustrate the effectiveness of the proposed algorithm. [ABSTRACT FROM AUTHOR]

Realizing accurate measurement for thermal contact conductance (TCC) is difficult and needs many additional measurements. In this paper, we design a three-layer back propagation artificial neural network (ANN) model to retrieve TCC of two contact solid specimens. The model can retrieve TCC only according to the measured temperatures and loading pressures, and the effects of loading pressure, temperature and surface roughness on TCC are considered. Besides, thermal conductivity, heat flux and the parameter of the TCC empirical fitting formula can be obtained simultaneously. The retrieved results are validated by experimental values. The results show that for two contact pairs with different surface roughness, the retrieved parameters of TCC empirical fitting formula are clear reasonable and can be explained conceptually compared with the published studies. [ABSTRACT FROM AUTHOR]

In this paper, an adaptive critic neural network controller is designed for a class of discrete-time chaotic system. The critic neural network is used to approximate the long-term function. In contrast with the existing results for discrete-time chaotic systems, in this paper, a near optimal control input can be generated when the long-term function is minimized. It is proven that the tracking error, the adaptation laws and the control input are uniformly bounded. A simulation example is employed to illustrate the effectiveness of the proposed algorithm. [ABSTRACT FROM AUTHOR]

• Designs workflow for personalized prediction of response to first-line chemotherapy. • Performs systematic metabolomic analysis of lung cancer fresh tumor core biopsies. • Machine learning analyses differentiate between treatment responses and disease stage. • Predicts response and identifies metabolites associated with patient classifications. Despite extensive effort, the search for clinically-relevant metabolite biomarkers for early detection, disease monitoring, and outcome prediction in lung cancer remains unfulfilled. Although biofluid evaluation has been explored, the complexity inherent in metabolite data and the dynamic discrepancy between metabolites in biofluids vs. tumor tissue have prevented conclusive results. This proof-of-concept study explored models predictive of staging and chemotherapy response based on metabolomic analysis of fresh, patient-derived non-small cell lung cancer (NSCLC) core biopsies. Samples (n = 36) were evaluated with high-resolution 2DLC-MS/MS and 13C-glucose enrichment, and the data were comprehensively analyzed with machine learning techniques. Patients were categorized as Disease-Control (DC) [encompassing complete-response (CR), partial-response (PR), and stable-disease (SD)] and Progressive-Disease (PD) in terms of first-line chemotherapy. Four major types of learning methods (partial least squares discriminant analysis (PLS-DA), support vector machines (SVM), artificial neural networks, and random forests (RF)) were applied to differentiate between positive (DC and CR/PR) and poor (PD and SD/PD) responses, and between stage I/II/III and stage IV disease. Models were trained with forward feature selection based on variable importance and tested on validation subsets. The models predicted patient classifications in the validation subsets with AUC (95 % CI): DC vs. PD (SVM), 0.970(0.961–0.979); CR/PR vs. SD/PD (PLS-DA), 0.880(0.865–0.895); stage I/II/III vs. IV (SVM), 0.902(0.880–0.924). Highest performing model was SVM for DC vs. PD (balanced accuracy = 0.92; kappa = 0.74). This study illustrates a comprehensive evaluation of patient tumor-specific metabolic profiles, with the potential to identify disease stage and predict response to first-line chemotherapy. [ABSTRACT FROM AUTHOR]

The order parameters and synchronization are numerically investigated in the time delay small-world connected FitzHugh–Nagumo excitable systems. The simulations show that the order parameter continuously decreases with increasing D, the quality of the synchronization worsens for large noise intensity. As the coupling intensity goes up, the quality of the synchronization worsens also, and find that the larger rewiring probability becomes, the larger-order parameter. At the same time, the order parameter goes up, the time delay declines. We obtained the complete phase diagram for a wide range of values of noise intensity D and control parameter g. [ABSTRACT FROM AUTHOR]

Abstract: This Letter deals with the synchronization problem of a class of neural networks with/without time delays. Based on the drive–response concept and the Lyapunov stability theorem, a delay-independent and decentralized control law is derived to achieve the exponential synchronization. Two illustrative examples are given to demonstrate the effectiveness of the presented synchronization scheme. [Copyright &y& Elsevier]

Background: Semantic textual similarity (STS) is a fundamental natural language processing (NLP) task which can be widely used in many NLP applications such as Question Answer (QA), Information Retrieval (IR), etc. It is a typical regression problem, and almost all STS systems either use distributed representation or one-hot representation to model sentence pairs.Methods: In this paper, we proposed a novel framework based on a gated network to fuse distributed representation and one-hot representation of sentence pairs. Some current state-of-the-art distributed representation methods, including Convolutional Neural Network (CNN), Bi-directional Long Short Term Memory networks (Bi-LSTM) and Bidirectional Encoder Representations from Transformers (BERT), were used in our framework, and a system based on this framework was developed for a shared task regarding clinical STS organized by BioCreative/OHNLP in 2018.Results: Compared with the systems only using distributed representation or one-hot representation, our method achieved much higher Pearson correlation. Among all distributed representations, BERT performed best. The highest Person correlation of our system was 0.8541, higher than the best official one of the BioCreative/OHNLP clinical STS shared task in 2018 (0.8328) by 0.0213.Conclusions: Distributed representation and one-hot representation are complementary to each other and can be fused by gated network. [ABSTRACT FROM AUTHOR]

The goal of this study was to recognize human standing postures in human-robot collaborations such that the robot can serve the human operator better. An intelligent sensing floor was developed based on a thin-film pressure sensor and a human standing posture dataset was obtained by transforming the pressure data into a pressure image. A human standing posture recognition method based on an improved convolutional neural network is proposed. The results of the experiments demonstrate that a convolutional neural network can be used in the field of pressure images. The proposed method returned a recognition rate of 96.6%. Compared to the traditional neural network, the improved convolutional neural network model has better performance. The study results are expected to be used in standing posture monitoring to provide additional data for a robot in a human-robot collaboration system. [ABSTRACT FROM AUTHOR]

Background: Clinical entity recognition as a fundamental task of clinical text processing has been attracted a great deal of attention during the last decade. However, most studies focus on clinical text in English rather than other languages. Recently, a few researchers have began to study entity recognition in Chinese clinical text.Methods: In this paper, a novel deep neural network, called attention-based CNN-LSTM-CRF, is proposed to recognize entities in Chinese clinical text. Attention-based CNN-LSTM-CRF is an extension of LSTM-CRF by introducing a CNN (convolutional neural network) layer after the input layer to capture local context information of words of interest and an attention layer before the CRF layer to select relevant words in the same sentence.Results: In order to evaluate the proposed method, we compare it with other two currently popular methods, CRF (conditional random field) and LSTM-CRF, on two benchmark datasets. One of the datasets is publically available and only contains contiguous clinical entities, and the other one is constructed by us and contains contiguous and discontiguous clinical entities. Experimental results show that attention-based CNN-LSTM-CRF outperforms CRF and LSTM-CRF.Conclusions: CNN and attention mechanism are individually beneficial to LSTM-CRF-based Chinese clinical entity recognition system, no matter whether contiguous clinical entities are considered. The conribution of attention mechanism is greater than CNN. [ABSTRACT FROM AUTHOR]

This paper focuses on the issue of robust stability analysis for recurrent neural networks (RNNs) with leakage delay. By constructing a novel Lyapunov-Krasovskii functional together with the reciprocally convex approach and the free-weighting matrix technique, some less conservative stability criteria in terms of linear matrix inequalities for RNNs are derived. The new contribution of this paper is that a novel delay-partitioning method is proposed, and some new zero equalities are introduced. Finally, several examples are given to demonstrate the effectiveness of the proposed methods. The simulated results reveal that the leakage delay has great influence on the dynamical systems, and it cannot be neglected. [ABSTRACT FROM AUTHOR]

A Long Short-Term Memory (LSTM) neural network model is developed to provide a real-time calculation tool for monitoring the mooring line responses under the operating condition by using the vessel motion as the only input. A feature extraction method based on first order moment and second order center moment is proposed in the data pre-processing to improve the characteristics of fluctuation information of input data. The effective judging standard of the scope of prediction accuracy regarding the proposed neural network model is defined. The impact of normalization method, neural number, and training sets length on the predicting accuracy are studied. The feasibility of the established LSTM neural network model is verified considering different data relevance between the training sets and validating sets. The results indicate that the mooring line responses can be predicted with high precision based on the vessel motion as input by using the established LSTM neural network model. •A LSTM neural network model is developed topredict the mooring line responses only using the vessel motion as input. • A feature extraction method based on first order moment and second order center moment is proposed in the data pre-processing. • The influence of normalization method, neural number and training sets length on the predicting accuracy of LSTM neural network model are studied. [ABSTRACT FROM AUTHOR]

During human–robot collaborations (HRC), robot systems must accurately perceive the actions and intentions of humans. The present study proposes the classification of standing postures from standing-pressure images, by which a robot system can predict the intended actions of human workers in an HRC environment. To this end, it explores deep learning based on standing-posture recognition and a multi-recognition algorithm fusion method for HRC. To acquire the pressure-distribution data, ten experimental participants stood on a pressure-sensing floor embedded with thin-film pressure sensors. The pressure data of nine standing postures were obtained from each participant. The human standing postures were discriminated by seven classification algorithms. The results of the best three algorithms were fused using the Dempster–Shafer evidence theory to improve the accuracy and robustness. In a cross-validation test, the best method achieved an average accuracy of 99.96%. The convolutional neural network classifier and data-fusion algorithm can feasibly classify the standing postures of human workers. [ABSTRACT FROM AUTHOR]

Background: Extracting relationships between chemicals and diseases from unstructured literature have attracted plenty of attention since the relationships are very useful for a large number of biomedical applications such as drug repositioning and pharmacovigilance. A number of machine learning methods have been proposed for chemical-induced disease (CID) extraction due to some publicly available annotated corpora. Most of them suffer from time-consuming feature engineering except deep learning methods. In this paper, we propose a novel document-level deep learning method, called recurrent piecewise convolutional neural networks (RPCNN), for CID extraction.Results: Experimental results on a benchmark dataset, the CDR (Chemical-induced Disease Relation) dataset of the BioCreative V challenge for CID extraction show that the highest precision, recall and F-score of our RPCNN-based CID extraction system are 65.24, 77.21 and 70.77%, which is competitive with other state-of-the-art systems.Conclusions: A novel deep learning method is proposed for document-level CID extraction, where domain knowledge, piecewise strategy, attention mechanism, and multi-instance learning are combined together. The effectiveness of the method is proved by experiments conducted on a benchmark dataset. [ABSTRACT FROM AUTHOR]