Springer Science and Business Media LLC

Many applications of light field (LF) imaging have been limited by the spatial-angular resolution problem, hence the need for efficient super-resolution techniques. Recently, learning-based solutions have achieved remarkably better performances than traditional super-resolution (SR) techniques. Unfortunately, the learning or training process relies heavily on the training dataset, which could be limited for most LF imaging applications. In this paper, we propose a novel LF spatial-angular SR algorithm based on zero-shot learning. We suggest learning cross-scale reusable features in the epipolar plane image (EPI) space, and avoiding explicitly modeling scene priors or implicitly learning that from a large number of LFs. Most importantly, without using any external LFs, the proposed algorithm can simultaneously super-resolve a LF in both spatial and angular domains. Moreover, the proposed solution is free of depth or disparity estimation, which is usually employed by existing LF spatial and angular SR. By using a simple 8-layers fully convolutional network, we show that the proposed algorithm can generate comparable results to the state-of-the-art spatial SR. Our algorithm outperforms the existing methods in terms of angular SR on multiple groups of public LF datasets. The experiment results indicate that the cross-scale features can be well learned and be reused for LF SR in the EPI space.

In this paper, we are concerned with the problem of nonlinear inequalities defined on a graph. The feasible solution set to this problem is often infinity and Laplacian eigenmap is used as heuristic information to gain better performance in the solution. A continuous-time projected neural network, and the corresponding discrete-time projected neural network are both given to tackle this problem iteratively. The convergence of the neural networks are proven in theory. The effectiveness of the proposed neural networks are tested and compared with others via its applications in the range-free localization of wireless sensor networks. Simulations demonstrate the effectiveness of the proposed methods.

Homonyms are words that share their spelling but differ in meaning and are a common feature in most languages. Homonyms are a source of noise i most text analyses and are difficult to detect; numerous studies have been conducted in this regard. However, extant methods typically detect homonyms using a rule-based or statistical-based approach, which requires an answer set, with little regard to the semantic meaning of the word. Therefore, we propose a novel approach for the detection of homonyms based on contextual word embedding that allows a word to be understood based on the context in which it appears. In this study, we extracted all contextual word embedding vectors of individual words and clustered those vectors using a spherical k-means clustering to detect pairs of homonyms. In addition, we developed a homonym replacement method to increase the performance of a document embedding technique, based on the word vector value. We replaced the embedding vectors of homonyms with a representative vector based on the respective meaning using the proposed homonym detection method. Experimental results indicate that the proposed method effectively detects homonyms and significantly improves the performance of document embedding.

In recent years, multivariate time series prediction has attracted extensive research interests. However, the dynamic changes of the spatial topology and the temporal evolution of multivariate variables bring great challenges to the spatio-temporal time series prediction. In this paper, a novel Dirichlet graph convolution module is introduced to automatically learn the spatio-temporal representation, and we combine graph attention (GAT) and neural differential equation (NDE) based on nonlinear state transition to model spatio-temporal state evolution of nonlinear systems. Specifically, the spatial topology is revealed by the cosine similarity of node embeddings. The use of multi-layer Dirichlet graph convolution aims to enhance the representation ability of the model while suppressing the phenomenon of over-smoothing or over-separation. The GCN and LSTM-based network is used as the nonlinear operator to model the evolution law of the dynamic system, and the GAT updates the strength of the connection. In addition, the Euler trapezoidal integral method is used to model the temporal dynamics and makes medium and long-term prediction in latent space from the perspective of nonlinear state transition. The proposed model can adaptively mine spatial correlations and discover spatio-temporal dynamic evolution patterns through the coupled NDE, which makes the modeling process more interpretable. Experiment results demonstrate the effectiveness of spatio-temporal dynamic discovery on predictive performance.

In this paper, a class of Bidirectional Associative Memory neural networks with time-varying weights and continuously distributed delays is discussed. Sufficient conditions are obtained for the existence and uniqueness of weighted pseudo-almost periodic solution of the considered model and numerical examples are given to show the effectiveness of the obtained results.

In this paper, we propose hybrid reaction–diffusion Cohen–Grossberg neural networks (RDCGNNs) with variable coefficients and mixed time delays. By using the Lyapunov–Krasovkii functional approach, stochastic analysis technique and Hardy inequality, some novel sufficient conditions are derived to ensure the pth moment exponential stability of hybrid RDCGNNs with mixed time delays. The obtained sufficient conditions are relevant to the diffusion terms. The results of this paper are novel and improve some of the previously known results. Finally, two numerical examples are provided to verify the usefulness of the obtained results.

Image captioning aims to describe the content of an image with a complete and natural sentence. Recently, the image captioning methods with encoder-decoder architecture has made great progress, in which LSTM became a dominant decoder to generate word sequence. However, in the decoder stage, the input vector keep same and there is much uncorrelated with previously visual parts or generated words. In this paper, we propose an attentional LSTM (ALSTM) and show how to integrate it within state-of-the-art automatic image captioning framework. Instead of traditional LSTM in existing models, ALSTM learns to refine input vector from network hidden states and sequential context information. Thus ALSTM can attend more relevant features such as spatial attention, visual relations and pay more attention on the most relevant context words. Moreover, ALSTM is utilized as the decoder in some classical frameworks and shows how to get effective visual/context attention to update input vector. Extensive quantitative and qualitative evaluations on the Flickr30K and MSCOCO image datasets with modified network illustrate the superiority of ALSTM. ALSTM based methods can generate high quality descriptions by combining sequence context and relations.

This sequel is concerned with the analysis of robust synchronization for a multi-weighted complex structure on fractional-order coupled neural networks (MWCFCNNs) with linear coupling delays via state feedback controller. Firstly, by means of fractional order comparison principle, suitable Lyapunov method, Kronecker product technique, some famous inequality techniques about fractional order calculus and the basis of interval parameter method, two improved robust asymptotical synchronization analysis, both algebraic method and LMI method, respectively are established via state feedback controller. Secondly, when the parameter uncertainties are ignored, several synchronization criterion are also given to ensure the global asymptotical synchronization of considered MWCFCNNs. Moreover, two type of special cases for global asymptotical synchronization MWCFCNNs with and without linear coupling delays, respectively are investigated. Ultimately, the accuracy and feasibility of obtained synchronization criteria are supported by the given two numerical computer simulations.