arXiv subjects

Ping Zhong

Publications and source records attributed to Ping Zhong.

At least 37 records · Page 2Linked to original sources

arXiv2025

Nearshore Underwater Target Detection Meets UAV-borne Hyperspectral Remote Sensing: A Novel Hybrid-level Contrastive Learning Framework and Benchmark Dataset

UAV-borne hyperspectral remote sensing has emerged as a promising approach for underwater target detection (UTD). However, its effectiveness is hindered by spectral distortions in nearshore environments, which compromise the accuracy of traditional hyperspectral UTD (HUTD) methods that rely on bathymetric model. These distortions lead to significant uncertainty in target and background spectra, challenging the detection process. To address this, we propose the Hyperspectral Underwater Contrastive Learning Network (HUCLNet), a novel framework that integrates contrastive learning with a self-paced learning paradigm for robust HUTD in nearshore regions. HUCLNet extracts discriminative features from distorted hyperspectral data through contrastive learning, while the self-paced learning strategy selectively prioritizes the most informative samples. Additionally, a reliability-guided clustering strategy enhances the robustness of learned representations.To evaluate the method effectiveness, we conduct a novel nearshore HUTD benchmark dataset, ATR2-HUTD, covering three diverse scenarios with varying water types and turbidity, and target types. Extensive experiments demonstrate that HUCLNet significantly outperforms state-of-the-art methods. The dataset and code will be publicly available at: https://github.com/qjh1996/HUTD

cs.CV↗

arXiv2025

The Brown measure of a sum of two free nonselfadjoint random variables, one of which is R-diagonal

Suppose that $X_{1}$ and $X_{2}$ are two $*$-free (generally unbounded) random variables with Brown measures $μ_{X_{1}}$ and $μ_{X_{2}}$, respectively. Using properties of classical free additive convolutions, we develop a method for calculating $μ_{X_{1}+X_{2}}$when $X_{2}$ is $R$-diagonal. This method determines a density relative to Lebesgue measure on an open set whose closure contains the support of $μ_{X_{1}+X_{2}}$. Effective calculations are possible in important cases. Biane and Lehner were the first to make significant progress on the problem we consider, even in some cases in which neither $X_{1}$ nor $X_{2}$ is $R$-diagonal. Our examples overlap with theirs, but we emphasize the use of subordination functions. When $X_{2}$ is circular, $μ_{X_{1}+X_{2}}$ was studied earlier using two different approaches, one involving Hamilton-Jacobi equations, and another using standard free probability techniques. Our work extends the second approach.

math.PR↗

arXiv2024

Deformed single ring theorems

Given a sequence of deterministic matrices $A = A_N$ and a sequence of deterministic nonnegative matrices $Σ=Σ_N$ such that $A\to a$ and $Σ\to σ$ in $\ast$-distribution for some operators $a$ and $σ$ in a finite von Neumann algebra $\mathcal{A}$. Let $U =U_N$ and $V=V_N$ be independent Haar-distributed unitary matrices. We use free probability techniques to prove that, under mild assumptions, the empirical eigenvalue distribution of $UΣV^*+A$ converges to the Brown measure of $T+a$, where $T\in\mathcal{A}$ is an $R$-diagonal operator freely independent from $a$ and $\vert T\vert$ has the same distribution as $σ$. The assumptions can be removed if $A$ is Hermitian or unitary. By putting $A= 0$, our result removes a regularity assumption in the single ring theorem by Guionnet, Krishnapur and Zeitouni. We also prove a local convergence on optimal scale, extending the local single ring theorem of Bao, Erdős and Schnelli.

math.PR↗

arXiv2024

Lighten CARAFE: Dynamic Lightweight Upsampling with Guided Reassemble Kernels

As a fundamental operation in modern machine vision models, feature upsampling has been widely used and investigated in the literatures. An ideal upsampling operation should be lightweight, with low computational complexity. That is, it can not only improve the overall performance but also not affect the model complexity. Content-aware Reassembly of Features (CARAFE) is a well-designed learnable operation to achieve feature upsampling. Albeit encouraging performance achieved, this method requires generating large-scale kernels, which brings a mass of extra redundant parameters, and inherently has limited scalability. To this end, we propose a lightweight upsampling operation, termed Dynamic Lightweight Upsampling (DLU) in this paper. In particular, it first constructs a small-scale source kernel space, and then samples the large-scale kernels from the kernel space by introducing learnable guidance offsets, hence avoiding introducing a large collection of trainable parameters in upsampling. Experiments on several mainstream vision tasks show that our DLU achieves comparable and even better performance to the original CARAFE, but with much lower complexity, e.g., DLU requires 91% fewer parameters and at least 63% fewer FLOPs (Floating Point Operations) than CARAFE in the case of 16x upsampling, but outperforms the CARAFE by 0.3% mAP in object detection. Code is available at https://github.com/Fu0511/Dynamic-Lightweight-Upsampling.

cs.CV↗

arXiv2024

The Brown measure of a sum of two free random variables, one of which is triangular elliptic

The triangular elliptic operators are natural extensions of the elliptic deformation of circular operators. We obtain a Brown measure formula for the sum of a triangular elliptic operator $g_{_{α, β, γ}}$ with a random variable $x_0$, which is $*$-free from $g_{_{α, β, γ}}$ with amalgamation over certain unital subalgebra. Let $c_t$ be a circular operator. We prove that the Brown measure of $x_0 + g_{_{α, β, γ}}$ is the push-forward measure of the Brown measure of $x_0 + c_t$ by an explicitly defined map on $\mathbb{C}$ for some suitable $t$. We show that the Brown measure of $x_0+c_t$ is absolutely continuous with respect to the Lebesgue measure on $\mathbb{C}$ and its density is bounded by $1/(π{t})$. This work generalizes earlier results on the addition with a circular operator, semicircular operator, or elliptic operator to a larger class of operators. We extend operator-valued subordination functions, due to Biane and Voiculescu, to certain unbounded operators. This allows us to extend our results to unbounded operators.

math.OA↗

arXiv2023

Spectrum of Laplacian matrices associated with large random elliptic matrices

A Laplacian matrix is a square matrix whose row sums are zero. We study the limiting eigenvalue distribution of a Laplacian matrix formed by taking a random elliptic matrix and subtracting the diagonal matrix containing its row sums. Under some mild assumptions, we show that the empirical spectral distribution of the Laplacian matrix converges to a deterministic probability distribution as the size of the matrix tends to infinity. The limiting measure can be interpreted as the Brown measure of the sum of an elliptic operator and a freely independent normal operator with a Gaussian distribution.

math.PR↗

arXiv2023

HyperDID: Hyperspectral Intrinsic Image Decomposition with Deep Feature Embedding

The dissection of hyperspectral images into intrinsic components through hyperspectral intrinsic image decomposition (HIID) enhances the interpretability of hyperspectral data, providing a foundation for more accurate classification outcomes. However, the classification performance of HIID is constrained by the model's representational ability. To address this limitation, this study rethinks hyperspectral intrinsic image decomposition for classification tasks by introducing deep feature embedding. The proposed framework, HyperDID, incorporates the Environmental Feature Module (EFM) and Categorical Feature Module (CFM) to extract intrinsic features. Additionally, a Feature Discrimination Module (FDM) is introduced to separate environment-related and category-related features. Experimental results across three commonly used datasets validate the effectiveness of HyperDID in improving hyperspectral image classification performance. This novel approach holds promise for advancing the capabilities of hyperspectral image analysis by leveraging deep feature embedding principles. The implementation of the proposed method could be accessed soon at https://github.com/shendu-sw/HyperDID for the sake of reproducibility.

cs.CV↗

arXiv2023

Beyond Sharing Weights in Decoupling Feature Learning Network for UAV RGB-Infrared Vehicle Re-Identification

Owing to the capacity of performing full-time target search, cross-modality vehicle re-identification (Re-ID) based on unmanned aerial vehicle (UAV) is gaining more attention in both video surveillance and public security. However, this promising and innovative research has not been studied sufficiently due to the data inadequacy issue. Meanwhile, the cross-modality discrepancy and orientation discrepancy challenges further aggravate the difficulty of this task. To this end, we pioneer a cross-modality vehicle Re-ID benchmark named UAV Cross-Modality Vehicle Re-ID (UCM-VeID), containing 753 identities with 16015 RGB and 13913 infrared images. Moreover, to meet cross-modality discrepancy and orientation discrepancy challenges, we present a hybrid weights decoupling network (HWDNet) to learn the shared discriminative orientation-invariant features. For the first challenge, we proposed a hybrid weights siamese network with a well-designed weight restrainer and its corresponding objective function to learn both modality-specific and modality shared information. In terms of the second challenge, three effective decoupling structures with two pretext tasks are investigated to learn orientation-invariant feature. Comprehensive experiments are carried out to validate the effectiveness of the proposed method. The dataset and codes will be released at https://github.com/moonstarL/UAV-CM-VeID.

cs.CV↗

arXiv2022

Brown measure of $R$-diagonal operators, revisited

We use subordination functions perspective to reformulate Haagerup--Schultz's approach for the Brown measure of $R$-diagonal operators. This allows us to simplify the original argument and find a connection with the other approach due to Belinschi-Śniady-Speicher. The Brown measure formula can be rewritten in terms of subordination functions.

math.OA↗

arXiv2022

The spectrum of local random Hamiltonians

The spectrum of a local random Hamiltonian can be represented generically by the so-called $ε$-free convolution of its local terms' probability distributions. We establish an isomorphism between the set of $ε$-noncrossing partitions and permutations to study its spectrum. Moreover, we derive some lower and upper bounds for the largest eigenvalue of the Hamiltonian.

math-ph↗

arXiv2022

Masked Spatial-Spectral Autoencoders Are Excellent Hyperspectral Defenders

Deep learning methodology contributes a lot to the development of hyperspectral image (HSI) analysis community. However, it also makes HSI analysis systems vulnerable to adversarial attacks. To this end, we propose a masked spatial-spectral autoencoder (MSSA) in this paper under self-supervised learning theory, for enhancing the robustness of HSI analysis systems. First, a masked sequence attention learning module is conducted to promote the inherent robustness of HSI analysis systems along spectral channel. Then, we develop a graph convolutional network with learnable graph structure to establish global pixel-wise combinations.In this way, the attack effect would be dispersed by all the related pixels among each combination, and a better defense performance is achievable in spatial aspect.Finally, to improve the defense transferability and address the problem of limited labelled samples, MSSA employs spectra reconstruction as a pretext task and fits the datasets in a self-supervised manner.Comprehensive experiments over three benchmarks verify the effectiveness of MSSA in comparison with the state-of-the-art hyperspectral classification methods and representative adversarial defense strategies.

cs.CV↗

arXiv2022

A CNN with Noise Inclined Module and Denoise Framework for Hyperspectral Image Classification

Deep Neural Networks have been successfully applied in hyperspectral image classification. However, most of prior works adopt general deep architectures while ignore the intrinsic structure of the hyperspectral image, such as the physical noise generation. This would make these deep models unable to generate discriminative features and provide impressive classification performance. To leverage such intrinsic information, this work develops a novel deep learning framework with the noise inclined module and denoise framework for hyperspectral image classification. First, we model the spectral signature of hyperspectral image with the physical noise model to describe the high intraclass variance of each class and great overlapping between different classes in the image. Then, a noise inclined module is developed to capture the physical noise within each object and a denoise framework is then followed to remove such noise from the object. Finally, the CNN with noise inclined module and the denoise framework is developed to obtain discriminative features and provides good classification performance of hyperspectral image. Experiments are conducted over two commonly used real-world datasets and the experimental results show the effectiveness of the proposed method. The implementation of the proposed method and other compared methods could be accessed at https://github.com/shendu-sw/noise-physical-framework.

cs.CV↗

arXiv2022

Transferable Physical Attack against Object Detection with Separable Attention

Transferable adversarial attack is always in the spotlight since deep learning models have been demonstrated to be vulnerable to adversarial samples. However, existing physical attack methods do not pay enough attention on transferability to unseen models, thus leading to the poor performance of black-box attack.In this paper, we put forward a novel method of generating physically realizable adversarial camouflage to achieve transferable attack against detection models. More specifically, we first introduce multi-scale attention maps based on detection models to capture features of objects with various resolutions. Meanwhile, we adopt a sequence of composite transformations to obtain the averaged attention maps, which could curb model-specific noise in the attention and thus further boost transferability. Unlike the general visualization interpretation methods where model attention should be put on the foreground object as much as possible, we carry out attack on separable attention from the opposite perspective, i.e. suppressing attention of the foreground and enhancing that of the background. Consequently, transferable adversarial camouflage could be yielded efficiently with our novel attention-based loss function. Extensive comparison experiments verify the superiority of our method to state-of-the-art methods.

cs.CV↗

arXiv2022

Superconvergence in free probability limit theorems for arbitrary triangular arrays

It is known that limit theorems for triangular arrays with identically distributed rows yields convergence of densities rather than just convergence in distribution. We show that this superconvergence result holds -- at least at points at which the limit density is nonzero -- even if the rows of the array are not identically distributed.

math.PR↗

arXiv2022

Self-aligned Spatial Feature Extraction Network for UAV Vehicle Re-identification

Compared with existing vehicle re-identification (ReID) tasks conducted with datasets collected by fixed surveillance cameras, vehicle ReID for unmanned aerial vehicle (UAV) is still under-explored and could be more challenging. Vehicles with the same color and type show extremely similar appearance from the UAV's perspective so that mining fine-grained characteristics becomes necessary. Recent works tend to extract distinguishing information by regional features and component features. The former requires input images to be aligned and the latter entails detailed annotations, both of which are difficult to meet in UAV application. In order to extract efficient fine-grained features and avoid tedious annotating work, this letter develops an unsupervised self-aligned network consisting of three branches. The network introduced a self-alignment module to convert the input images with variable orientations to a uniform orientation, which is implemented under the constraint of triple loss function designed with spatial features. On this basis, spatial features, obtained by vertical and horizontal segmentation methods, and global features are integrated to improve the representation ability in embedded space. Extensive experiments are conducted on UAV-VeID dataset, and our method achieves the best performance compared with recent ReID works.

cs.CV↗

arXiv2021

Brown Measures of Free Circular and Multiplicative Brownian Motions with Self-Adjoint and Unitary Initial Conditions

Let $Z_N$ be a Ginibre ensemble and let $A_N$ be a Hermitian random matrix independent from $Z_N$ such that $A_N$ converges in distribution to a self-adjoint random variable $x_0$. For each $t>0$, the random matrix $A_N+\sqrt{t}Z_N$ converges in $\ast$-distribution to $x_0+c_t$, where $c_t$ is the circular variable of variance $t$, free from $x_0$. We use the Hamilton-Jacobi method to compute the Brown measure $ρ_t$ of $x_0+c_t$. The Brown measure has a density that is constant along the vertical direction inside the support. The support of the Brown measure of $x_0+c_t$ is related to the subordination function of the free additive convolution of $x_0+s_t$, where $s_t$ is the semicircular variable of variance $t$, free from $x_0$. Furthermore, the push-forward of $ρ_t$ by a natural map is the law of $x_0+s_t$. Let $G_N(t)$ be the Brownian motion on the general linear group and let $U_N$ be a unitary random matrix independent from $G_N(t)$ such that $U_N$ converges in distribution to a unitary random variable $u$. The random matrix $U_NG_N(t)$ converges in $\ast$-distribution to $ub_t$ where $b_t$ is the free multiplicative Brownian motion, free from $u$. We compute the Brown measure $μ_t$ of $ub_t$, extending the recent work by Driver-Hall-Kemp, which corresponds to the case $u=I$. The measure has a density of the special form \[\frac{1}{r^2}w_t(θ)\] in polar coordinates in its support. The support of $μ_t$ is related to the subordination function of the free multiplicative convolution of $uu_t$ where $u_t$ is the free unitary Brownian motion, free from $u$. The push-forward of $μ_t$ by a natural map is the law of $uu_t$. We compute the explicit formula for the special case where $u$ is Haar unitary. The support of the Brown measure of $ub_t$ is an annulus; in its support, the density in polar coordinates is given by \[\frac{1}{2πt}\frac{1}{r^2}.\]

math.OA↗

arXiv2021

Deep Manifold Embedding for Hyperspectral Image Classification

Deep learning methods have played a more and more important role in hyperspectral image classification. However, the general deep learning methods mainly take advantage of the information of sample itself or the pairwise information between samples while ignore the intrinsic data structure within the whole data. To tackle this problem, this work develops a novel deep manifold embedding method(DMEM) for hyperspectral image classification. First, each class in the image is modelled as a specific nonlinear manifold and the geodesic distance is used to measure the correlation between the samples. Then, based on the hierarchical clustering, the manifold structure of the data can be captured and each nonlinear data manifold can be divided into several sub-classes. Finally, considering the distribution of each sub-class and the correlation between different subclasses, the DMEM is constructed to preserve the estimated geodesic distances on the data manifold between the learned low dimensional features of different samples. Experiments over three real-world hyperspectral image datasets have demonstrated the effectiveness of the proposed method.

cs.CV↗

arXiv2021

Superconvergence and regularity of densities in free probability

The superconvergence phenomenon is shown for products of free, identically distributed random variables. We also show that a certain Holder regularity, first demonstrated by Biane for the density of a free additive convolution with a semicircular law, extends to free additive and multiplicative convolutions with arbitrary freely infinitely divisible laws and to free convolution semigroups.

math.FA↗