Search

Few-shot learning (FSL) aims to classify novel images based on a few labeled samples with the help of meta-knowledge. Most previous works address this problem based on the hypothesis that the training set and testing set are from the same domain, which is not realistic for some real-world applications. Thus, we extend FSL to domain-agnostic few-shot recognition, where the domain of the testing task is unknown. In domain-agnostic few-shot recognition, the model is optimized on data from one domain and evaluated on tasks from different domains. Previous methods for FSL mostly focus on learning general features or adapting to few-shot tasks effectively. They suffer from inappropriate features or complex adaptation in domain-agnostic few-shot recognition. In this brief, we propose meta-prototypical learning to address this problem. In particular, a meta-encoder is optimized to learn the general features. Different from the traditional prototypical learning, the meta encoder can effectively adapt to few-shot tasks from different domains by the traces of the few labeled examples. Experiments on many datasets demonstrate that meta-prototypical learning performs competitively on traditional few-shot tasks, and on few-shot tasks from different domains, meta-prototypical learning outperforms related methods. [ABSTRACT FROM AUTHOR]

In this article, the exponential stability problem for fractional-order complex multi-links networks with aperiodically intermittent control is considered. Using the graph theory and Lyapunov method, two theorems, including a Lyapunov-type theorem and a coefficient-type theorem, are given to ensure the exponential stability of the underlying networks. The theoretical results show that the exponential convergence rate is dependent on the control gain and the order of fractional derivative. To be specific, the larger control gain, the higher the exponential convergence rate. Meanwhile, when aperiodically intermittent control degenerates into periodically intermittent control, a corollary is also provided to ensure the exponential stability of the underlying networks. Furthermore, to show the practicality of theoretical results, as an application, the exponential stability of fractional-order multi-links competitive neural networks with aperiodically intermittent control is investigated and a stability criterion is established. Finally, the effectiveness and feasibility of the theoretical results are demonstrated through a numerical example. [ABSTRACT FROM AUTHOR]

This paper investigates the global bipartite synchronization of fractional-order time-varying coupled signed networks with proportional delays. A systematic approach is proposed by combining graph theory, the Lyapunov method, and the Razumikhin method. A Razumikhin-type theorem and a coefficient-type theorem are given, and thus several significant bipartite synchronization criteria related to both the order of fractional-order derivative and the proportionality factor are obtained. Besides, the results are applied to Chua's circuits to reflect the availability of theoretical analysis. Finally, a numerical example is proposed to substantiate the effectiveness of theoretical results. • An approach is used by combining graph theory, Lyapunov method and Razumikhin method. • Criteria are related to the fractional order and the proportionality factor. • The theoretical results are applied to Chua's circuits. [ABSTRACT FROM AUTHOR]

Highlights • We firstly propose a new periodically intermittent discrete observation control. • Synchronization issue of unified fractional-order chaotic coupled systems is discussed. • The method combined graph theory with Lyapunov method is adopted successfully. • We take nonlinear coupling term into account in our model, which is more general. Abstract In this paper, we propose a novel control technique, named as periodically intermittent discrete observation control (PIDOC), to investigate the synchronization issue of fractional-order coupled systems. Distinguished from periodically intermittent control which is widely applied in control fields, PIDOC proposed in this paper adopts discrete-time state observations in work time during a control period. In this way, PIDOC is more reasonable and available than traditional periodically intermittent control based on continuous-time state observations. Moreover, different from previous work about synchronization of fractional-order coupled systems, coupling term considered in this paper is nonlinear, which is more general. Then, combining graph theory with Lyapunov method, several synchronization criteria are obtained. Next, we successfully employ PIDOC to investigate synchronization of unified fractional-order chaotic coupled systems. Finally, two numerical simulations are presented to verify the validity of our theoretical results and the effectiveness of control scheme. [ABSTRACT FROM AUTHOR]

Conjugate gradient (CG) methods are a class of important methods for solving linear equations and nonlinear optimization problems. In this paper, we propose a new stochastic CG algorithm with variance reduction and we prove its linear convergence with the Fletcher and Reeves method for strongly convex and smooth functions. We experimentally demonstrate that the CG with variance reduction algorithm converges faster than its counterparts for four learning models, which may be convex, nonconvex or nonsmooth. In addition, its area under the curve performance on six large-scale data sets is comparable to that of the LIBLINEAR solver for the $L2$ -regularized $L2$ -loss but with a significant improvement in computational efficiency. 1 CGVR algorithm is available on github: https://github.com/xbjin/cgvr [ABSTRACT FROM AUTHOR]

Highlights • Combining Lyapunov method with graph-theoretic approach, periodically intermittent control is applied to synchronizing the FOCSs with time-varying delays for the first time. And two applications are also given. • Nonlinear coupling, time-varying internal delay and time-varying coupling delay are considered into our model, which is more general no matter in theoretical analysis and real applications. • The synchronization criteria depend on the fractional order, control gain, control rate and control period. Abstract This paper deals with synchronization problem of fractional-order coupled systems (FOCSs) with time-varying delays via periodically intermittent control. Here, nonlinear coupling, time-varying internal delay and time-varying coupling delay are considered when modeling, which makes our model more general in comparison with the most existing fractional-order models. It is the first time that periodically intermittent control is applied to synchronizing FOCSs with time-varying delays. Combining Lyapunov method with graph-theoretic approach, some synchronization criteria are obtained. Moreover, the synchronization criteria we derive depend on the fractional order α , control gain, control rate and control period. Besides, the synchronization issues of fractional-order coupled chaotic systems with time-varying delays and fractional-order coupled Hindmarsh–Rose neuron systems with time-varying delays are also investigated as applications of our theoretical results, and relevant sufficient conditions are derived. Finally, numerical simulations with two examples are provided in order to demonstrate the effectiveness of the theoretical results and the feasibility of control strategy. [ABSTRACT FROM AUTHOR]

In this brief, we propose a new margin scalable discriminative least squares regression (MSDLSR) model for multicategory classification. The main motivation behind the MSDLSR is to explicitly control the margin of DLSR model. We first prove that the DLSR is a relaxation of the traditional L2 -support vector machine. Based on this fact, we further provide a theorem on the margin of DLSR. With this theorem, we add an explicit constraint on DLSR to restrict the number of zeros of dragging values, so as to control the margin of DLSR. The new model is called MSDLSR. Theoretically, we analyze the determination of the margin and support vectors of MSDLSR. Extensive experiments illustrate that our method outperforms the current state-of-the-art approaches on various machine leaning and real-world data sets. [ABSTRACT FROM PUBLISHER]

This brief presents a framework of retargeted least squares regression (ReLSR) for multicategory classification. The core idea is to directly learn the regression targets from data other than using the traditional zero–one matrix as regression targets. The learned target matrix can guarantee a large margin constraint for the requirement of correct classification for each data point. Compared with the traditional least squares regression (LSR) and a recently proposed discriminative LSR models, ReLSR is much more accurate in measuring the classification error of the regression model. Furthermore, ReLSR is a single and compact model, hence there is no need to train two-class (binary) machines that are independent of each other. The convex optimization problem of ReLSR is solved elegantly and efficiently with an alternating procedure including regression and retargeting as substeps. The experimental evaluation over a range of databases identifies the validity of our method. [ABSTRACT FROM AUTHOR]

In this paper, the H ∞ synchronization problem of semi-Markovian jump neural networks (s-MJNNs) based on adaptive event-triggered scheme (AETS) output feedback control is studied. Also, the time-varying delay and leakage delay are considered in the system model. In order to solve the problems that the state of the system is not completely observable and limited network resources, an output feedback controller with AETS is designed. At the same time, in order to describe the performance of the sensor in the feedback link, the random nonlinear phenomenon of the sensor is described by the variable complying with Bernoulli probability distribution. A suitable Lyapunov–Krasovskii functional (LKF) is constructed and the bounds of integral terms are estimated by affine Bessel–Legendre inequality. Finally, sufficient conditions for asymptotic stability of the synchronization error system are obtained. And, two numerical examples show the feasibility of the research work. [ABSTRACT FROM AUTHOR]

• Multiple weights are considered into FCDNs as the initial attempt, which makes the model more realistic in practical applications. • A finite-time synchronization criterion based on adaptive control for FCDNs is derived for the first time. • Settling time of synchronization depends on the order of fractional derivative, the parameters of control and topologica structure of networks. In this paper, finite-time synchronization issue of fractional-order complex-valued dynamical networks with multiple weights is addressed. Multiple weights are pulled into consideration, which makes our model more practical and general than single-weight one. Combining Lyapunov method with graph-theoretic method and making use of the theory of complex functions, finite-time synchronization criteria for fractional-order complex-valued dynamical networks with multiple weights are established by using feedback control and adaptive control, respectively. Furthermore, the settling time of synchronization is effectively estimated, which is connected with the order of fractional derivative, the parameters of control and topological structure of networks. Eventually, numerical simulations are exhibited to demonstrate the validity of theoretical results and analyse the relationship between the settling time of synchronization and the order of fractional derivative with the parameters of control. [ABSTRACT FROM AUTHOR]

Feature representation learning, an emerging topic in recent years, has achieved great progress. Powerful learned features can lead to excellent classification accuracy. In this article, a selective and robust feature representation framework with a supervised constraint (SRSC) is presented. SRSC seeks a selective, robust, and discriminative subspace by transforming the original feature space into the category space. Particularly, we add a selective constraint to the transformation matrix (or classifier parameter) that can select discriminative dimensions of the input samples. Moreover, a supervised regularization is tailored to further enhance the discriminability of the subspace. To relax the hard zero-one label matrix in the category space, an additional error term is also incorporated into the framework, which can lead to a more robust transformation matrix. SRSC is formulated as a constrained least square learning (feature transforming) problem. For the SRSC problem, an inexact augmented Lagrange multiplier method (ALM) is utilized to solve it. Extensive experiments on several benchmark data sets adequately demonstrate the effectiveness and superiority of the proposed method. The proposed SRSC approach has achieved better performances than the compared counterpart methods. [ABSTRACT FROM AUTHOR]

A three-phase three-level Vienna rectifier has been widely used in the front-end PFC modules of electric vehicle and data center DC power supply system with the advantages of without dead-time, low switch voltage stress and small grid current harmonic. However, to achieve unity power factor and alleviate specific harmonic pollution, its controller design becomes more complex. Therefore, a novel finite control set model predictive control strategy is presented. The behavior of the output voltage is firstly predicted for each possible switching state in each sampling interval based on the model of a Vienna rectifier. Besides, the midpoint voltage imbalance and time delay caused by the calculation may deteriorate the grid current or even damage the device seriously. Thus, a cost function is first built to balance the neutral-point-potential voltage. After then, the cost function is updated by considering the delay caused by the calculation, sampling and gate driver. Compared with the conventional PI-SVM control strategy, the effectiveness of the proposed model predictive control strategy is validated by both the simulation and experimental results. [ABSTRACT FROM AUTHOR]