No abstract

No abstract

Due to the curse of dimensionality of search space, it is extremely difficult for evolutionary algorithms to approximate the optimal solutions of large-scale multiobjective optimization problems (LMOPs) by using a limited budget of evaluations. If the Pareto optimal subspace is approximated during the evolutionary process, the search space can be reduced and the difficulty encountered by evolutionary algorithms can be highly alleviated. Following the above idea, this paper proposes an evolutionary algorithm to solve sparse LMOPs by learning the Pareto optimal subspace. The proposed algorithm uses two unsupervised neural networks, a restricted Boltzmann machine and a denoising autoencoder to learn a sparse distribution and a compact representation of the decision variables, where the combination of the learnt sparse distribution and compact representation is regarded as an approximation of the Pareto optimal subspace. The genetic operators are conducted in the learnt subspace, and the resultant offspring solutions then can be mapped back to the original search space by the two neural networks. According to the experimental results on eight benchmark problems and eight real-world problems, the proposed algorithm can effectively solve sparse LMOPs with 10,000 decision variables by only 100,000 evaluations.

The global lunar image of the first phase of Chinese Lunar Exploration Program is the first image that covered all over the surface of the Moon. It will serve as a critical foundation for succeeding exploration and scientific research. In this paper, the acquisition, characteristics, and data quality of Chang'E-1 CCD image data are described in detail. Also described are the methodology and procedure of data processing. According to rule of planetary cartography, the image data have been processed, geometrically corrected, and then mosaicked and merged in a scale of 1:2.5 million. The results of data processing and charting show that the image data of Chang'E-1 CCD and their geometric precision meet the demand of charting a map in the scale of 1:2.5 million. The relative geometric positioning precision of the global image is better than 240 m, and the absolute geometric positioning precision is slightly better than that of the ULCN2005 and Clementine lunar basemap (V2.0). The plane positioning precision is approximately 100-1500 m. This global image proves to be the best global image of the Moon so far in terms of space coverage, image quality, and positioning precision.Chang'E-1, lunar CCD data processing, lunar image position, lunar global image Citation:Li C L, Liu J J, Ren X, et al. The global image of the Moon obtained by the Chang'E-1: Data processing and lunar cartography.

The Laser AltiMeter (LAM), as one of the main payloads of Chang'E-1 probe, is used to measure the topography of the lunar surface. It performed the first measurement at 02:22 on November 28th, 2007. Up to December 4th 2008, the total number of measurements was approximately 9.12 million, covering the whole surface of the Moon. Using the LAM data, we constructed a global lunar Digtal Elevation Model (DEM) with 3 km spatial resolution. The model shows pronounced morphological characteristics, legible and vivid details of the lunar surface. The plane positioning accuracy of the DEM is 445 m (1σ), and the vertical accuracy is 60 m (1σ). From this DEM model, we measured the full range of the altitude difference on the lunar surface, which is about 19.807 km. The highest point is 10.629 km high, on a peak between crater Korolev and crater Dirichlet-Jackson at (158.656°W, 5.441°N) and the lowest point is −9.178 km in height, inside crater Antoniadi (172.413°W, 70.368°S) in the South Pole-Aitken Basin. By comparison, the DEM model of Chang'E-1 is better than the USA ULCN2005 in accuracy and resolution and is probably identical to the DEM of Japan SELENE, but the DEM of Chang'E-1 reveals a new lowest point, clearly lower than that of SELENE.Chang'E-1, laser altimetry, lunar DEM, topographic tops of the Moon Citation:Li C L, Ren X, Liu J J, et al. Laser altimetry data of Chang'E-1 and the global lunar DEM model.

No abstract

We investigate the scalable image classification problem with a large number of categories. Hierarchical visual data structures are helpful for improving the efficiency and performance of large-scale multi-class classification. We propose a novel image classification method based on learning hierarchical inter-class structures. Specifically, we first design a fast algorithm to compute the similarity metric between categories, based on which a visual tree is constructed by hierarchical spectral clustering. Using the learned visual tree, a test sample label is efficiently predicted by searching for the best path over the entire tree. The proposed method is extensively evaluated on the ILSVRC2010 and Caltech 256 benchmark datasets. The experimental results show that our method obtains significantly better category hierarchies than other state-of-the-art visual tree-based methods and, therefore, much more accurate classification.

Identification of B-cell epitopes in target antigens is one of the most crucial steps for epitopebased vaccine development, immunodiagnostic tests, antibody production, and disease diagnosis and therapy. Experimental methods for B-cell epitope mapping are time consuming, costly and labor intensive; in the meantime, various in-silico methods are proposed to predict both linear and conformational B-cell epitopes. The accurate identification of B-cell epitopes presents major challenges for immunoinformaticians. In this paper, we have comprehensively reviewed in-silico methods for B-cell epitope identification. The aim of this review is to stimulate the development of better tools which could improve the identification of B-cell epitopes, and further for the development of therapeutic antibodies and diagnostic tools.