Calibration of ϵ−insensitive loss in support vector machines regression

Support vector machines regression (SVMR) is an important tool in many machine learning applications. In this paper, we focus on the theoretical understanding of SVMR based on the $??$insensitive loss. For fixed ??≥?0 and general data generating distributions, we show that the minimizer of the expected risk for $??$insensitive loss used in SVMR is a set-valued function called conditional $??$median. We then establish a calibration inequality of $??$insensitive loss under a noise condition on the conditional distributions. This inequality also ensures us to present a nontrivial variance-expectation bound for $??$insensitive loss, and which is known to be important in statistical analysis of the regularized learning algorithms. With the help of the calibration inequality and variance-expectation bound, we finally derive an explicit learning rate for SVMR in some $L^r?$space.     

KPI relevant and irrelevant fault monitoring with neighborhood component analysis and two-level PLS

Generally, the efficiency of key performance indicator (KPI) relevant and irrelevant fault monitoring is highly related to the definition of KPI relevant and irrelevant variations in the input space. Therefore, a novel KPI relevant and irrelevant fault monitoring scheme is proposed, the offline modeling phase of which incorporates a neighborhood component analysis (NCA)-based KPI relevant variable selection method and a two-level partial least square (PLS) modeling strategy. With the utilization of NCA, the KPI relevant and irrelevant input variables could be determined, respectively. To obtain an explicit decomposition of KPI relevant and irrelevant variations from the input, a two-level PLS modeling strategy is proposed to avoid the potential loss of KPI relevant information that hidden in KPI irrelevant variables and the potential loss of KPI irrelevant information that resulted from the first level PLS model. It is thus expected to achieve superior performance than the methods that considered in the current work. The effectiveness and superiority of the proposed method in monitoring KPI relevant and irrelevant faults have also be demonstrated by implementing comparisons with its counterparts.     

The stability analysis of time-varying delayed systems based on new augmented vector method

This paper focuses on the stability analysis of systems with interval time-varying delay. A new augmented vector containing single and double integral terms is constructed and the corresponding Lyapunov functional with triple integral terms is introduced. In order to improve the estimating accuracy of the derivatives of the constructed Lyapunov functional, single integral inequalities and double integral inequalities via auxiliary functions are employed on the first step, then an extended relaxed integral inequality and reciprocally convex approach are further utilized to narrow the scaling room of the functional derivatives. As a result, some novel delay-dependent stability criteria with less conservatism are derived. Finally, numerical examples are provided to check the effectiveness of the theoretical results and the improvement of the proposed method over the existing works.     

Learning soft mask with DNN and DNN-SVM for multi-speaker DOA estimation using an acoustic vector sensor

Using an acoustic vector sensor (AVS), an efficient method has been presented recently for direction of arrival (DOA) estimation of multiple speech sources via the clustering of the inter-sensor data ratio (AVS-ISDR). Through extensive experiments on simulated and recorded data, we observed that the performance of the AVS-DOA method is largely dependent on the reliable extraction of the target speech dominated time–frequency points (TD-TFPs) which, however, may be degraded with the increase in the level of additive noise and room reverberation in the background. In this paper, inspired by the great success of deep learning in speech recognition, we design two new soft mask learners, namely deep neural network (DNN) and DNN cascaded with a support vector machine (DNN-SVM), for multi-source DOA estimation, where a novel feature, namely, the tandem local spectrogram block (TLSB) is used as the input to the system. Using our proposed soft mask learners, the TD-TFPs can be accurately extracted under different noisy and reverberant conditions. Additionally, the generated soft masks can be used to calculate the weighted centers of the ISDR-clusters for better DOA estimation as compared to the original center used in our previously proposed AVS-ISDR. Extensive experiments on simulated and recorded data have been presented to show the improved performance of our proposed methods over two baseline AVS-DOA methods in presence of noise and reverberation.     

A new approach for constrained chaos synchronization with application to secure data communication

In this paper, a new technique is introduced for chaos secure data communication. In this approach, in addition to the usually used techniques for data encryption, the concept of carrier encryption is introduced to increase the security level of the secure communication scheme. To fulfill this objective, at the transmitting end, two chaotic oscillators are coupled, and a set of inequality time dependent constraints with time dependent bounds is imposed on the generated chaotic signals. Moreover, to increase system complexity and its security level, the imposed set of constraints and their bounds are allowed to be changeable from one time period to another during the transmission process. As a result, the patterns of the generated chaotic signals are completely changed and the chaotic oscillator is completely encrypted. At the receiving end, the newly developed Constrained Smoothed Regularized Least Square (CSRLS) observer is used to synchronize the received constrained chaotic signals and hence retrieve the transmitted data. Using such an approach, the quality of the received information, measured by the Bit Error Rate (BER), is highly improved due to the superior performance of the developed CSRLS observer. The stability of the observer is analyzed, and simulation results are presented to show the efficiency and effectiveness of the proposed secure communication scheme.     

Resilient sampled-data control design for singular networked systems with random missing data

This paper proposes an active resilient control strategy for singular networked control systems with external disturbances and missing data scenario based on sampled-data scheme. To characterize the missing data scenario, a stochastic variable satisfying Bernoulli distributed white sequence is introduced. Based on this scenario, in this paper, two different models are proposed. For both the models, by using Lyapunov–Krasovskii functional approach, which fully uses the available information about the actual sampling pattern, some sufficient conditions in terms of linear matrix inequalities (LMIs) are separately obtained to guarantee that the resulting closed-loop system is admissible and strictly dissipative with a prescribed performance index. In particular, Jensen’s and Wirtinger based integral inequalities are employed to simplify the integral terms which appeared in the derivation of stabilization results. Then, if the obtained LMIs are feasible, the corresponding parameters of the designed resilient sampled-data controller are determined. Finally, two numerical examples are presented to demonstrate the effectiveness of the proposed control design technique.     

Parameter identification of time-delayed nonlinear systems: An integrated method with adaptive noise correction

This paper proposes a novel method called the adaptive-noise-correction integrated parameter identification (ANCPI) for time-delayed nonlinear systems. Compared with the existing de-noising techniques, the significance of the proposed method is the use of the system itself to correct the noise-polluted components so that the accuracy of parameter identification is enhanced. To achieve the goal of adaptive noise correction, this study starts from the case of periodic response and then parameterizes the noise correction as the coefficient correction of harmonic basis. In this way, the parameter identification integrated with noise correction can be performed as the parameter optimization of the error function. For the convenience of application, a user-friendly program package is further provided and a detailed tutorial is presented in the supplementary material.     

Non-fragile sampled-data stabilization analysis for linear systems with probabilistic time-varying delays

This paper deals with the problem of non-fragile sampled-data stabilization analysis for a class of linear systems with probabilistic time-varying delays via new double integral inequality approach. Based on the auxiliary function-based integral inequality (AFBII) and with the help of some mathematical approaches, a new double integral inequality (NDII) is developed. Then, to demonstrate the merits of the proposed inequality, an appropriate Lyapunov–Krasovskii functional (LKF) is constructed with some augmented delay-dependent terms. By employing integral inequalities, an enhanced stability criterion for the concerned system model is derived in terms of linear matrix inequalities (LMIs). Finally, three benchmark illustrative examples are given to validate the effectiveness and advantages of the proposed results.     

Leader-following consensus of multi-agent systems with limited data rate

This paper investigates the leader-following consensus problem of time-invariant linear multi-agent systems with limited data rate. Based on the idea of assigning a priority level for each agent of the concerned multi-agent system, a novel distributed control law has been proposed. The proposed control law has two distinctive advantages. That is, it is fully distributed in the sense that it does not rely on the eigenvalues of the Laplace matrix associated with the topology. Moreover, the required data rate is independent of the number of agents and remains small even if the number of the agents in multi-agent systems is large. An example is finally given to illustrate the effectiveness of the proposed controller.     

Robust observer design by sign-stability for the monitoring of population systems

Monitoring problem in population ecology can be formalized as observer design for the population system in question: Supposing that we observe only certain species considered indicators, we want to recover or estimate the whole state process of the population system, where the state vector is usually composed from the biomasses of the single populations. In the present paper, for stably coexisting population systems, a new approach to the design of the corresponding observer system is proposed which, from the knowledge of the observed indicator(s), estimates the state process with exponential convergence. In the usual observer design, an auxiliary matrix, the so-called gain matrix must be found that guarantees the mentioned exponential convergence. The novelty is in that due to the required sign-stability (or qualitative stability) of the interaction pattern, the designed observer system (i.e. the gain matrix) is robust against quantitative changes in the inter- and intra-specific interactions. (Here the interaction pattern is described by a matrix having the signs as entries, indicating the quality of the interactions within and between the considered species.) In other words, under sign-stability conditions, in the observer design the same gain matrix can be used even if, due to environmental changes, the intensities of certain interactions suffer a quantitative change in the meanwhile. The requirement of sign-stability of the interaction pattern can be considered rather natural, since in a stably coexisting population system, it means for example that a predator–prey relation does not change into a prey–predator interaction, and interactions neither appear nor disappear within the system. Our approach to robust observer design is illustrated on model population systems, such as trophic chains of type resource-producer-primary consumer-secondary consumer and Lotka–Volterra system with vertical structure. For the latter system a Lyapunov function is also constructed that guarantees the global asymptotic stability of the positive equilibrium of the considered model.     

Quantized consensus in second-order multi-agent systems via event-trigger control with sampled data

This paper addresses the quantized consensus problem of second-order multi-agent systems (MASs) where the topology has a directed spanning tree. An event-trigger control protocol (ETCP) is proposed by designing a combined threshold. The combined threshold not only reduces more event triggers than the state-dependent threshold, but also is more practical than the time-dependent threshold. For further reducing computation resources and transmission cost, the sampled data, self-trigger scheme and data quantization are employed together. Under the proposed ETCP, the sufficient condition is derived to guarantee the quantized consensus of second-order MASs. Finally, the comparison experiments are conducted to demonstrate the superiority of ETCP based on the combined threshold.     

Observer based leader-following consensus of second-order multi-agent systems with nonuniform sampled position data

This paper deals with the leader-following consensus problem of multi-agent systems with the consideration that each agent can only transmit its position state to the neighbors at irregular discrete sampling times. In the proposed algorithm, a continuous-discrete time observer is designed for the continuous estimation of both position and velocity from the discrete position information of the neighbors. These estimated states are then used for designing a continuous control law which solves the leader-following consensus problem. Moreover, the dynamics of the leader is not fixed and can be controlled through an external input. The stability analysis has been carried out by employing the Lyapunov approach which provides sufficient conditions to tune the parameters according to the maximum allowable sampling period. The developed algorithm has been simulated and then tested on an actual multi-robot system consisting of three differential drive wheeled robots. Both simulation and hardware results validate the effectiveness of the control algorithm.     

Delays compensation for an atmospheric sliced tomatoes dehydration process via state predictors

In this paper the experimental application of static and dynamic predictors based control of the temperature of a sliced tomatoes dehydration process is presented. The control schemes concerning the case corresponding to the process under consideration, where the state delay is greater than the input delay, are developed. The number of subdivisions in the approximation of the distributed delays of the dynamic predictors by finite sums is chosen to insure loop system stability. The two schemes are experimentally tested, and a comparison of their performance is presented.     

Enhanced efficiency in vector control of a Surface-mounted PMSM drive

In this paper, emphasis has been put on providing the result of a detailed simulation study on the dynamics and efficiency of a vector-controlled voltage source inverter (VSI)-fed surface-mounted permanent magnet synchronous motor (SPMSM) drive; which employs firstly a model-based loss-minimization algorithm (MLMA) and secondly a binary search-based online loss-minimization algorithm (BSOLMA), so as to reduce the total controllable electrical losses without much affecting the dynamic performance of the drive. The steady-state SPMSM model taking into account the core loss has been considered. The dynamic performances of the SPMSM drive employing individually the abovementioned loss minimization algorithms (LMAs) have been compared with that employing conventional zero d-axis current control (ZDCC) through extensive digital computer simulations. The simulation results show that efficiency of the drive enhances with the employment of the LMA.     

A clustering algorithm based TS fuzzy model for tracking dynamical system data

This paper presents an optimal fuzzy partition based Takagi Sugeno Fuzzy Model (TSFM) in which a novel clustering algorithm, known as Modified Fuzzy C-Regression Model (MFCRM), has been proposed. The objective function of MFCRM algorithm has been developed by considering of geometrical structure of input data and linear functional relation between input–output data. The MFCRM partitions the data space to create fuzzy subspaces (rules). A new validation criterion has been developed for detecting the right number of rules (subspaces) in a given data set. The obtained fuzzy partition is used to build the fuzzy structure and identify the premise parameters. Once, right number of rules and premise parameters have been identified, then consequent parameters have been identified by orthogonal least square (OLS) approach. The cluster validation index has been tested on synthetic data set. The effectiveness of MFCRM based TSFM has been validated on benchmark examples, such as Boiler Turbine system, Mackey–Glass time series data and Box–Jenkins model. The model performance is also validated through high-dimensional data such as Auto-MPG data and Boston Housing data.     

Optimizing between data transformation and parametric weighting for stable binary classification

In this paper, an optimization problem is formulated for stable binary classification. Essentially, the objective function seeks to optimize a full data transformation matrix along with the learning of a linear parametric model. The data transformation matrix and the weight parameter vector are alternatingly optimized based on the area above the receiver operating characteristic curve criterion. The proposed method improves the existing means via an optimal data transformation rather than that based on the diagonal, random and ad-hoc settings. This optimal transformation stretches beyond the fixed settings of known optimization methods. Extensive experiments using 34 binary classification data sets show that the proposed method can be more stable than competing classifiers. Specifically, the proposed method shows robustness to imbalanced and small training data sizes in terms of classification accuracy with statistical evidence.     

An integrated data-driven Markov parameters sequence identification and adaptive dynamic programming method to design fault-tolerant optimal tracking control for completely unknown model systems

In this paper, an integrated design of data-driven fault-tolerant tracking control is addressed relying on the Markov parameters sequence identification and adaptive dynamic programming techniques. For the unknown model systems, the sequence of Markov parameters together with the covariance of innovation signal is firstly estimated by least square method. After a transformation of value function from stochastic to deterministic, a policy iteration adaptive dynamic programming algorithm is then formulated to find the optimal tracking control law. In order to eliminate the influence of unpredicted faults, an active fault-tolerant supervisory control strategy is further constructed by synthesizing fault detection, isolation, estimation and compensation. All these involved designs are performed in the data-driven manner, and thus avoid the information requirement about system drift dynamics. From the perspective of system operation management, the above integrated control scheme provides a framework to achieve the tracking performance optimization, monitoring and maintaining simultaneously. The effectiveness of these conclusions is finally verified via two case studies.