It is a difficult task to classify images with multiple class labels using only a small number of labeled examples, especially when the label (class) distribution is imbalanced. Emotion classification is such an example of imbalanced label distribution, because some classes of emotions like disgusted are relatively rare comparing to other labels like happy or sad. In this paper, we propose a data augmentation method using generative adversarial networks (GAN). It can complement and complete the data manifold and find better margins between neighboring classes. Specifically, we design a framework using a CNN model as the classifier and a cycle-consistent adversarial networks (CycleGAN) as the generator. In order to avoid gradient vanishing problem, we employ the least-squared loss as adversarial loss. We also propose several evaluation methods on three benchmark datasets to validate GAN's performance. Empirical results show that we can obtain 5%∼10% increase in the classification accuracy after employing the GAN-based data augmentation techniques.

Finite-time stability analysis and controller synthesis for switched linear parameter-varying (LPV) systems are discussed in this paper. A new finite-time stability condition and robust finite-time controller design method are presented for switched LPV systems with two different structured uncertainty modelling assumptions (i.e. affine linear structured uncertainty or polytopic structured uncertainty). On the one hand, by using the piecewise parameter-dependent Lyapunovlike function, a less conservativeness finite-time stability condition is established. On the other hand, the new condition based on linear matrix inequalities relieves the controller design burden of dealing with specific applications. Finally, the provided design method is highly desirable to treat the problem of attitude control of bank-to-turn missiles with different channels coupling, and computer simulations demonstrate the effectiveness and superiority of the theoretical results. IntroductionA switched system is an important class of hybrid systems and has received a considerable attention from many researchers in the last decade. It consists of several subsystems and a switching rule specifying the switches among subsystems. It can be applied into a great number of real-world systems. For example, in flight control, the controller of aircraft switches at different flight operating points along the flight trajectory [1]. (For more details, see [2] and the references therein.) Up to now, most of the existing literatures on stability of switched systems concentrate on Lyapunov asymptotic stability, which is defined over an infinite-time interval [3,4]. However, in many realistic cases, the main concern is the transient performance of the system over a fixed finite-time interval. For the study of above stability problems, the concept of finite-time stability was first introduced in [5]. Specifically, a system is said to be finitetime stable if, given a bound on the initial condition, its states remain within a prescribed bound over a fixed time interval. Then, some works on finite-time stability of linear system were further discussed in [6,7].In recent years, the concepts of finite-time stability have been extended to switched linear systems [8][9][10][11][12][13][14] and then extended further to different switched systems, such as switched positive systems [15, 16], switched delay systems [17], switched discretetime systems [18] and switched stochastic systems [19]. Since the stability of switched systems depends not only on the dynamics of each subsystem but also the properties of switching signals, among the exiting literatures, various switching signals are used, including average dwell-time switching [8], asynchronous switching [9, 10], mode-dependent average dwell-time switching [12], state-dependent switching [14]and so on. However, the problems addressed in most of the above literatures mainly focused on the switched systems without considering model uncertainty and only few papers investigated the finite-time stability of uncertain s...

In this paper, we propose a model to analyze sentiment of online stock forum and use the information to predict the stock volatility in the Chinese market. We have labeled the sentiment of the online financial posts and make the dataset public available for research. By generating a sentimental dictionary based on financial terms, we develop a model to compute the sentimental score of each online post related to a particular stock. Such sentimental information is represented by two sentiment indicators, which are fused to market data for stock volatility prediction by using the Recurrent Neural Networks (RNNs). Empirical study shows that, comparing to using RNN only, the model performs significantly better with sentimental indicators.

No abstract

No abstract