In this paper, we investigate the cross-media retrieval between images and text, i.e., using image to search text (I2T) and using text to search images (T2I). Existing cross-media retrieval methods usually learn one couple of projections, by which the original features of images and text can be projected into a common latent space to measure the content similarity. However, using the same projections for the two different retrieval tasks (I2T and T2I) may lead to a tradeoff between their respective performances, rather than their best performances. Different from previous works, we propose a modality-dependent cross-media retrieval (MDCR) model, where two couples of projections are learned for different cross-media retrieval tasks instead of one couple of projections. Specifically, by jointly optimizing the correlation between images and text and the linear regression from one modal space (image or text) to the semantic space, two couples of mappings are learned to project images and text from their original feature spaces into two common latent subspaces (one for I2T and the other for T2I). Extensive experiments show the superiority of the proposed MDCR compared with other methods. In particular, based the 4,096 dimensional convolutional neural network (CNN) visual feature and 100 dimensional LDA textual feature, the mAP of the proposed method achieves 41.5%, which is a new state-of-the-art performance on the Wikipedia dataset.
KaVA (KVN and VERA Array) is a new combined VLBI array with KVN (Korean VLBI Network) and VERA (VLBI Exploration of Radio Astrometry). First, we briefly review the imaging capabilities of KaVA array which actually achieves more than three times better dynamic range than that achieved by VERA alone. The KaVA images clearly show detailed structures of extended radio jets in AGNs. Next, we represent the key science program to be led by KaVA AGN sub working group. We will conduct the monitoring observations of Sgr A* and M87 because of the largeness of their central super-massive black hole angular sizes. The main science goals of the program are (i) testing magnetically-driven-jet paradigm by mapping velocity fields of the M87 jet, and (ii) obtaining tight constraints on physical properties of radio emitting region in Sgr A*.
Link prediction in complex network based on solely topological information is a challenging problem. In this paper, we propose a novel similarity index, which is efficient and parameter free, based on clustering ability. Here clustering ability is defined as average clustering coefficient of nodes with the same degree. The motivation of our idea is that common-neighbors are able to contribute to the likelihood of forming a link because they own some ability of clustering their neighbors together, and then clustering ability defined here is a measure for this capacity. Experimental numerical simulations on both real-world networks and modeled networks demonstrated the high accuracy and high efficiency of the new similarity index compared with three well-known common-neighbor based similarity indices: CN, AA and RA.
We present the observational results of the Gamma-ray blazar, 3C 66A, at 2.3, 8.4, and 22 GHz at 4 epochs during 2004-05 with the VLBA. The resulting images show an overall core-jet structure extending roughly to the south with two intermediate breaks occurring in the region near the core. By model-fitting to the visibility data, the northmost component, which is also the brightest, is identified as the core according to its relatively flat spectrum and its compactness. As combined with some previous results to investigate the proper motions of the jet components, it is found the kinematics of 3C 66A is quite complicated with components of inward and outward, subluminal and superluminal motions all detected in the radio structure. The superluminal motions indicate strong Doppler boosting exists in the jet. The apparent inward motions of the innermost components last for at least 10 years and could not be caused by new-born components. The possible reason could be non-stationarity of the core due to opacity change.
Independent Component Analysis (ICA) is an effective unsupervised tool to learn statistically independent representation. However, ICA is not only sensitive to whitening but also difficult to learn an over-complete basis. Consequently, ICA with soft Reconstruction cost(RICA) was presented to learn sparse representations with over-complete basis even on unwhitened data. Whereas RICA is infeasible to represent the data with nonlinear structure due to its intrinsic linearity. In addition, RICA is essentially an unsupervised method and can not utilize the class information. In this paper, we propose a kernel ICA model with reconstruction constraint (kRICA) to capture the nonlinear features. To bring in the class information, we further extend the unsupervised kRICA to a supervised one by introducing a discrimination constraint, namely d-kRICA. This constraint leads to learn a structured basis consisted of basis vectors from different basis subsets corresponding to different class labels. Then each subset will sparsely represent well for its own class but not for the others. Furthermore, data samples belonging to the same class will have similar representations, and thereby the learned sparse representations can take more discriminative power. Experimental results validate the effectiveness of kRICA and d-kRICA for image classification.
