arXiv subjects

Pingping Zhang

Publications and source records attributed to Pingping Zhang.

96 records · Page 6Linked to original sources

arXiv2018

Unsupervised Band Selection of Hyperspectral Images via Multi-dictionary Sparse Representation

Hyperspectral images have far more spectral bands than ordinary multispectral images. Rich band information provides more favorable conditions for the tremendous applications. However, significant increase in the dimensionality of spectral bands may lead to the curse of dimensionality, especially for classification applications. Furthermore, there are a large amount of redundant information among the raw image cubes due to water absorptions, sensor noises and other influence factors. Band selection is a direct and effective method to remove redundant information and reduce the spectral dimension for decreasing computational complexity and avoiding the curse of dimensionality. In this paper, we present a novel learning framework for band selection based on the idea of sparse representation. More specifically, first each band is approximately represented by the linear combination of other bands, then the original band image can be represented by a multi-dictionary learning mechanism. As a result, a group of weights can be obtained by sparse optimization for all bands. Finally, the specific bands will be selected, if they get higher weights than other bands in the representation of the original image. Experimental results on three widely used hyperspectral datasets show that our proposed algorithm achieves better performance in hyperspectral image classification, when compared with other state-of-art band selection methods.

cs.CV↗

arXiv2018

Robustness of classification ability of spiking neural networks

It is well-known that the robustness of artificial neural networks (ANNs) is important for their wide ranges of applications. In this paper, we focus on the robustness of the classification ability of a spiking neural network which receives perturbed inputs. Actually, the perturbation is allowed to be arbitrary styles. However, Gaussian perturbation and other regular ones have been rarely investigated. For classification problems, the closer to the desired point, the more perturbed points there are in the input space. In addition, the perturbation may be periodic. Based on these facts, we only consider sinusoidal and Gaussian perturbations in this paper. With the SpikeProp algorithm, we perform extensive experiments on the classical XOR problem and other three benchmark datasets. The numerical results show that there is not significant reduction in the classification ability of the network if the input signals are subject to sinusoidal and Gaussian perturbations.

stat.ML↗

arXiv2017

Amulet: Aggregating Multi-level Convolutional Features for Salient Object Detection

Fully convolutional neural networks (FCNs) have shown outstanding performance in many dense labeling problems. One key pillar of these successes is mining relevant information from features in convolutional layers. However, how to better aggregate multi-level convolutional feature maps for salient object detection is underexplored. In this work, we present Amulet, a generic aggregating multi-level convolutional feature framework for salient object detection. Our framework first integrates multi-level feature maps into multiple resolutions, which simultaneously incorporate coarse semantics and fine details. Then it adaptively learns to combine these feature maps at each resolution and predict saliency maps with the combined features. Finally, the predicted results are efficiently fused to generate the final saliency map. In addition, to achieve accurate boundary inference and semantic enhancement, edge-aware feature maps in low-level layers and the predicted results of low resolution features are recursively embedded into the learning framework. By aggregating multi-level convolutional features in this efficient and flexible manner, the proposed saliency model provides accurate salient object labeling. Comprehensive experiments demonstrate that our method performs favorably against state-of-the art approaches in terms of near all compared evaluation metrics.

cs.CV↗

arXiv2017

Learning Uncertain Convolutional Features for Accurate Saliency Detection

Deep convolutional neural networks (CNNs) have delivered superior performance in many computer vision tasks. In this paper, we propose a novel deep fully convolutional network model for accurate salient object detection. The key contribution of this work is to learn deep uncertain convolutional features (UCF), which encourage the robustness and accuracy of saliency detection. We achieve this via introducing a reformulated dropout (R-dropout) after specific convolutional layers to construct an uncertain ensemble of internal feature units. In addition, we propose an effective hybrid upsampling method to reduce the checkerboard artifacts of deconvolution operators in our decoder network. The proposed methods can also be applied to other deep convolutional networks. Compared with existing saliency detection methods, the proposed UCF model is able to incorporate uncertainties for more accurate object boundary inference. Extensive experiments demonstrate that our proposed saliency model performs favorably against state-of-the-art approaches. The uncertain feature learning mechanism as well as the upsampling method can significantly improve performance on other pixel-wise vision tasks.

cs.CV↗

arXiv2016

Weak log majorization and determinantal inequalities

Denote by $¶_n$ the set of $n\times n$ positive definite matrices. Let $D = D_1\oplus \dots \oplus D_k$, where $D_1\in ¶_{n_1}, \dots, D_k \in ¶_{n_k}$ with $n_1+\cdots + n_k=n$. Partition $C\in ¶_n$ according to $(n_1, \dots, n_k)$ so that $\Diag C = C_1\oplus \dots \oplus C_k$. We prove the following weak log majorization result: \begin{equation*} λ(C^{-1}_1D_1\oplus \cdots \oplus C^{-1}_kD_k)\prec_{w \,\log} λ(C^{-1}D), \end{equation*} where $λ(A)$ denotes the vector of eigenvalues of $A\in \Cnn$. The inequality does not hold if one replaces the vectors of eigenvalues by the vectors of singular values, i.e., \begin{equation*} s(C^{-1}_1D_1\oplus \cdots \oplus C^{-1}_kD_k)\prec_{w \,\log} s(C^{-1}D) \end{equation*} is not true. As an application, we provide a generalization of a determinantal inequality of Matic \cite[Theorem 1.1]{M}. In addition, we obtain a weak majorization result which is complementary to a determinantal inequality of Choi \cite[Theorem 2]{C} and give a weak log majorization open question.

math.FA↗

arXiv2015

Remarks on an operator Wielandt inequality

Let $A$ be a positive operator on a Hilbert space $\mathcal{H}$ with $0 0.$$ We consider several upper bounds for $\frac{1}{2}|Γ+Γ^{*}|$. These bounds complement a recent result on operator Wielandt inequality.

math.FA↗