Image annotation aims to jointly predict multiple tags for an image. Although significant progress has been achieved, existing approaches usually overlook aligning specific labels and their corresponding regions due to the weak supervised information (i.e., "bag of labels" for regions), thus failing to explicitly exploit the discrimination from different classes. In this article, we propose the deep label-specific feature (Deep-LIFT) learning model to build the explicit and exact correspondence between the label and the local visual region, which improves the effectiveness of feature learning and enhances the interpretability of the model itself. Deep-LIFT extracts features for each label by aligning each label and its region. Specifically, Deep-LIFTs are achieved through learning multiple correlation maps between image convolutional features and label embeddings. Moreover, we construct two variant graph convolutional networks (GCNs) to further capture the interdependency among labels. Empirical studies on benchmark datasets validate that the proposed model achieves superior performance on multilabel classification over other existing state-of-the-art methods.

Multi-label classification has been recognized to be challenging since the complex correlations among labels and the combinatorial nature of the output label space. Motivated by the label dimension reduction and ensemble learning, in this paper, we propose a novel multi-label classification algorithm based on ensemble learning, termed PMAE: Pseudo Multi-Label Attention Ensemble. Different from existing landmark-based methods which reduce the predicted label space by selecting a few real labels as landmarks in a "hard" manner, the proposed PMAE develops the encoder-decoder mechanism to effectively utilize the low-dimensional pseudo-label representation space in a "soft" manner, which can capture the underlying correlations among labels and lead to more robust performance in complex scenarios. Moreover, we conduct the feature embedding based on multiple attention networks integrated, then, exponent and hinge diversity regularization are introduced to ensure the diversity and complementarity of multiple learners. Experimental results on multiple bench-mark datasets show the superiority of our proposed method over other state-of-the-arts.

In this paper, a new class of soft-input soft-output block decision feedback equalizer (BDFE) for turbo equalization over inter-symbol interference (ISI) channels under minimum mean-square error (MMSE) criterion is proposed. First, the BDFE which employs both the a posteriori moments of estimated symbols and the a priori moments of unestimated symbols is investigated, and two recursive schemes are derived to reduce the complexity. Then, to mitigate error propagation, the bidirectional BDFE (Bi-BDFE) consisting of a forward BDFE and a backward BDFE is proposed. Furthermore, a threshold is employed to reduce the number of inverse matrix update process for the BDFE (T-BDFE), which can attain a flexible trade-off between complexity and performance. Finally, the bidirectional equalization is introduced into a low-complexity approximate solution of the BDFE (Bi-LC-BDFE) to further reduce the complexity and obtain time diversity gain, which utilizes a time-invariant covariance matrix. Simulation results show that the proposed equalizers significantly outperform the block linear equalizer (BLE), the exact MMSE-based linear equalizer (Exact-MMSE-LE) and the soft-decision feedback equalizers (SDFE), while keeping a similar complexity as the BLE. In addition, the Bi-BDFE can achieve better performance than the Max-Log-MAP equalizer while its complexity is on the same order as the BDFE.

Stripe noise removal is a fundamental problem in remote sensing image processing. Many efforts have been made to resolve this problem. Recently, a state-of-the-art method was proposed from image-decomposition perspective. This method argued that the stripe and clear image can be simultaneously estimated by modeling the directional structure of stripes and the local smoothness of remote sensing images. However, the potential of this method cannot be fully delivered when confronting with dense stripes with high intensity. In this letter, we further consider the nonlocal self-similarity of image patches in the spatiospectral volume in terms of nonlocal total variation and propose a method of better robustness to dense stripes. Experimental results on both synthetic and real multispectral data show that the proposed method outperforms other competing methods in the remote sensing image destriping task.

The sensor-target placement plays a pivotal role in improving the localization accuracy; the commonly used evaluation function for this problem is built by Cramer-Rao lower bound (CRLB) or Fisher information matrix (FIM). Frame theory, which was developed recently, can provide a more easily tractable evaluation function, especially for some complex cases. In this letter, we derive an evaluation function by frame theory and prove its equivalence with that by CRLB and FIM. For angle-of-arrival based localization system, we present the optimal sensor augmentation problem with several fixed sensors and an arbitrary number of new sensors, and for arbitrary but fixed distances between new sensors and target. The conclusions are verified by several simulations.

The relative sensor-target formation configuration can significantly affect the potential performance of any particular localization algorithm. An evaluation measure based on the one-step prediction error covariance in extended Kalman filter process is derived in this letter. We determine the optimal sensor placement for the range-only localization considering the prior target location information. The conclusions are different from that in the static estimation problem.