Traffic demand prediction is of great importance for traffic management, yet it is also a challenging problem since the traffic data are usually with complex spatial-temporal dependencies and nonlinear relationships. In this paper, to better characterize and utilize the spatial and temporal features, we propose a Multi-dimensional Graph Convolutional Network (M-GCN) to capture dynamic spatial-temporal dependence of traffic data. M-GCN captures the explicit spatio-temporal dependencies by establishing a spatial adjacency graph and a temporal adjacency graph, and captures the hidden spatiotemporal dependencies by establishing a spatial adaptive graph and a temporal adaptive graph. We leverage graph convolutional networks for complex spatial and temporal dependencies modeling and design a gated fusion module to obtain the interactive spatial-temporal dependence. Besides, an attention mechanism is applied to alleviate the error propagation problem in long-term traffic prediction. Experimental results on two real-world traffic datasets demonstrate the superiority of M-GCN. The proposed M-GCN outperforms baseline methods by up to 4.6% improvement in MAE on the dataset TaxiNYC, and the training time and predicting time of $\mathrm{M}-\mathrm{GCN}$ are reduced by nearly half.

The recent end-to-end co-saliency models have good performance, however, they cannot express the semantic consistency among a group of images well and usually require many co-saliency labels. To this end, a two-stream encoder generative adversarial network (TSE-GAN) with progressive training is proposed in this paper. In the pre-training stage, the salient object detection generative adversarial networks (SOD-GAN) and classification network (CN) are separately trained by the salient object detection (SOD) datasets and co-saliency datasets with only category labels to learn the intra-saliency and preliminary inter-saliency cues and alleviate the problem of insufficient co-saliency labels. In the second training stage, the backbone of TSE-GAN is inherited from the trained SOD-GAN, the encoder of trained SOD-GAN (SOD-Encoder) is used to extract intra-saliency features, the group-wise semantic encoder (GS-Encoder) is constructed by the multi-level group-wise category features extracted from CN for extracting inter-saliency features with better semantic consistency, the TSE-GAN constructed by incorporating the GS-Encoder into SOD-GAN is trained on co-saliency datasets for co-saliency detection. The comprehensive comparisons with 13 state-of-the-art methods demonstrate the effectiveness of proposed method.

BACKGROUND: Comparative genomics has become an essential approach for identifying homologous gene candidates and their functions, and for studying genome evolution. There are many tools available for genome comparisons. Unfortunately, most of them are not applicable for the identification of unique genes and the inference of phylogenetic relationships in a given set of genomes. RESULTS: GenomeBlast is a Web tool developed for comparative analysis of multiple small genomes. A new parameter called "coverage" was introduced and used along with sequence identity to evaluate global similarity between genes. With GenomeBlast, the following results can be obtained: (1) unique genes in each genome; (2) homologous gene candidates among compared genomes; (3) 2D plots of homologous gene candidates along the all pairwise genome comparisons; and (4) a table of gene presence/absence information and a genome phylogeny. We demonstrated the functions in GenomeBlast with an example of multiple herpesviral genome analysis and illustrated how GenomeBlast is useful for small genome comparison. CONCLUSION: We developed a Web tool for comparative analysis of small genomes, which allows the user not only to identify unique genes and homologous gene candidates among multiple genomes, but also to view their graphical distributions on genomes, and to reconstruct genome phylogeny. GenomeBlast runs on a Linux server with 4 CPUs and 4 GB memory. The online version of GenomeBlast is available to public by using a Web browser with the URL http://bioinfo-srv1.awh.unomaha.edu/genomeblast/.

Formal concept analysis (FCA) is a method for deriving conceptual structures out of data that are represented as objects with features. FCA discovers dependencies within the data based on the relation among objects and features. However, not every pair of objects and features defines a concept. Concept approximation is to find the best or closest concept(s) to approximate a pair of objects and features. Concept approximation is significant in that under the circumstances that we can not find a concept, using concept approximation will give the best or most possible solution. In this paper, we evaluate three approaches through experiments in the application of document retrieval. We provide analysis of these approaches and give our concluding remarks.

The methods for batch processes monitoring based on multivariate techniques always have two assumptions. One is that batches must have equal duration. The other one is the normal distribution of batch process data. However, those are not always satisfied in practice. In order to overcome those problems, a novel monitoring method based on the dissimilarity index for batch processes is proposed without any assumptions on the lengths of batches and distribution of process data. The proposed method is applied to a simulation fed-batch penicillin production. The application results show that this method is effective.

Due to starting conditions and exotic environment of each batch run are different, their lengths are unequal. Moreover, batch data of measured parameters are not complete until the end of its operation. The shortcomings of conventional multiway Fisher discriminant analysis (MFDA) are that all batch lengths should be equal and that future trajectory of the current batch must be estimated to allow on-line fault diagnosis. Therefore, conventional MFDA easily leads to false fault diagnosis. In order to overcome those drawbacks and enhance the diagnostic performance, an improved method of fault diagnosis, improved multiway Fisher discriminant analysis (IMFDA), is proposed. The diagnostic ability of proposed method is demonstrated by application in the simulated fed-batch penicillin fermentation. Application results show that this method is very efficient.