Qian Zhao, Emmanuel J. Candes
Accurate statistical inference in logistic regression models remains a critical challenge when the ratio between the number of parameters and sample size is not negligible. This is because approximations based on either classical asymptotic theory or bootstrap calculations are grossly off the mark. This paper introduces a resized bootstrap method to infer model parameters in arbitrary dimensions. As in the parametric bootstrap, we resample observations from a distribution, which depends on an estimated regression coefficient sequence. The novelty is that this estimate is actually far from the maximum likelihood estimate (MLE). This estimate is informed by recent theory studying properties of the MLE in high dimensions, and is obtained by appropriately shrinking the MLE towards the origin. We demonstrate that the resized bootstrap method yields valid confidence intervals in both simulated and real data examples. Our methods extend to other high-dimensional generalized linear models.
Qian Zhao, Yan-Xi Wu, Mamutjan Ababekri, Zhong-Peng Li, Liang Tang, Jian-Xing Li
The advent of laser-driven high-intensity $γ$-photon beams has opened up new opportunities for designing advanced photon-photon colliders. Such colliders have the potential to produce a large yield of linear Breit-Wheeler (LBW) pairs in a single shot, which offers a unique platform for studying the polarized LBW process. In our recent work [Phys. Rev. D 105, L071902(2022)], we investigated the polarization characteristics of LBW pair production in CP $γ$-photon collisions. To fully clarify the polarization effects involving both CP and LP $γ$-photons, here we further investigate the LBW process using the polarized cross section with explicit azimuthal-angle dependence due to the base rotation of photon polarization vectors. We accomplished this by defining a new spin basis for positrons and electrons, which enables us to decouple the transverse and longitudinal spin components of $e^\pm$. By means of analytical calculations and Monte Carlo simulations, we find that the linear polarization of photon can induce the highly angle-dependent pair yield and polarization distributions. The comprehensive knowledge of the polarized LBW process will also open up avenues for investigating the higher-order photon-photon scattering, the laser-driven quantum electrodynamic plasmas and the high-energy astrophysics.
Hao Qian, Hongting Zhou, Qian Zhao, Hao Chen, Hongxiang Yao, Jingwei Wang, Ziqi Liu, Fei Yu, Zhiqiang Zhang, Jun Zhou
Jan 19, 2024·q-fin.ST·PDF The stock market is a crucial component of the financial system, but predicting the movement of stock prices is challenging due to the dynamic and intricate relations arising from various aspects such as economic indicators, financial reports, global news, and investor sentiment. Traditional sequential methods and graph-based models have been applied in stock movement prediction, but they have limitations in capturing the multifaceted and temporal influences in stock price movements. To address these challenges, the Multi-relational Dynamic Graph Neural Network (MDGNN) framework is proposed, which utilizes a discrete dynamic graph to comprehensively capture multifaceted relations among stocks and their evolution over time. The representation generated from the graph offers a complete perspective on the interrelationships among stocks and associated entities. Additionally, the power of the Transformer structure is leveraged to encode the temporal evolution of multiplex relations, providing a dynamic and effective approach to predicting stock investment. Further, our proposed MDGNN framework achieves the best performance in public datasets compared with state-of-the-art (SOTA) stock investment methods.
Qian Zhao, Siyu Tao, Jie Zhou, Linlin Wang, Xin Lin, Liang He
This paper describes our system for SemEval-2020 Task 4: Commonsense Validation and Explanation (Wang et al., 2020). We propose a novel Knowledge-enhanced Graph Attention Network (KEGAT) architecture for this task, leveraging heterogeneous knowledge from both the structured knowledge base (i.e. ConceptNet) and unstructured text to better improve the ability of a machine in commonsense understanding. This model has a powerful commonsense inference capability via utilizing suitable commonsense incorporation methods and upgraded data augmentation techniques. Besides, an internal sharing mechanism is cooperated to prohibit our model from insufficient and excessive reasoning for commonsense. As a result, this model performs quite well in both validation and explanation. For instance, it achieves state-of-the-art accuracy in the subtask called Commonsense Explanation (Multi-Choice). We officially name the system as ECNU-SenseMaker. Code is publicly available at https://github.com/ECNU-ICA/ECNU-SenseMaker.
Qian Hu, Jianhua Sun, Qian Zhao, Yonggang Meng
The experimentally demonstration of Casimir force transition from attraction to repulsion is still challenging. Herein, the Casimir forces for a sphere above a plate immersed in different liquids were precisely measured using Atomic force microscope, and the long-range repulsive Casimir force in the gold-cyclohexane-PTFE system is observed for the first time. The experimental data are consistent with the calculation by Lifshitz theory, which offers the direct evidence for the system of ε1<ε3<ε2. It further verifies the reasonability of van Zwol et al. dielectric model to describe the intervening fluids. This study is promising for potential applications on quantum levitation and frictionless devices in MEMS and NEMS by Casimir repulsion.
Qian Zhao, Liang Tang, Feng Wan, Bo-Chao Liu, Ruo-Yu Liu, Rui-Zhi Yang, Jin-Qing Yu, Xue-Guang Ren, Zhong-Feng Xu, Yong-Tao Zhao, Yong-Sheng Huang, Jian-Xing Li
Laser-driven brilliant controllable polarized $γ$-photon sources open the way for designing compact $γγ$ collider, which enable the large yield of linear Breit-Wheeler (LBW) pairs in a single shot and thus provide an opportunity for the investigation of polarized LBW process. In this work we investigate the polarization characteristics of LBW pair production via our developed spin-resolved binary collision simulation method. Polarization of $γ$-photons modifies the kinematics of scattering particles and induces the correlated energy-angle shift of LBW pairs, and the latter's polarization characteristic depends on the helicity configures of scattering particles. Our method confirms that the polarized $γγ$ collider with an asymmetric setup can be performed with currently achievable laser facilities to produce abundant polarized LBW pairs, fulfilling the detection power of polarimetries. The precise knowledge of polarized LBW process is in favor of the calibration and monitor of polarized $γγ$ collider, and could enhance the opacity of $γ$-photons in high-energy astrophysical objects to exacerbate the inconsistency between some observations and standard models.
Hui Wang, Zongsheng Yue, Qian Zhao, Deyu Meng
Blind image deblurring is an important yet very challenging problem in low-level vision. Traditional optimization based methods generally formulate this task as a maximum-a-posteriori estimation or variational inference problem, whose performance highly relies on the handcraft priors for both the latent image and the blur kernel. In contrast, recent deep learning methods generally learn, from a large collection of training images, deep neural networks (DNNs) directly mapping the blurry image to the clean one or to the blur kernel, paying less attention to the physical degradation process of the blurry image. In this paper, we present a deep variational Bayesian framework for blind image deblurring. Under this framework, the posterior of the latent clean image and blur kernel can be jointly estimated in an amortized inference fashion with DNNs, and the involved inference DNNs can be trained by fully considering the physical blur model, together with the supervision of data driven priors for the clean image and blur kernel, which is naturally led to by the evidence lower bound objective. Comprehensive experiments are conducted to substantiate the effectiveness of the proposed framework. The results show that it can not only achieve a promising performance with relatively simple networks, but also enhance the performance of existing DNNs for deblurring.
Zhi-Wei Lu, Qian Zhao, Feng Wan, Bo-Chao Liu, Yong-Sheng Huang, Zhong-Feng Xu, Jian-Xing Li
Generation of arbitrarily spin polarized muon pairs is investigated via polarized $e^-e^+$ collision. We calculate the fully spin-resolved cross section ${\rm d}σ_{e^-e^+\rightarrow μ^-μ^+}$ and utilize the Monte Carlo method of binary collision to describe the production and polarization processes of muon pairs. We find that, due to the dependence of mixed helicities on the scattering angle, arbitrarily polarized muon pairs with both of the longitudinal and transverse spin components can be produced. The collision of tightly collimated electron and positron beams with highly longitudinal polarization and nC charge can generate about $40\%$ muon pairs with longitudinal polarization and about $60\%$ muon pairs with transverse polarization. The compact high-flux $e^-e^+\rightarrowμ^-μ^+$ muon source could be implemented through the next-generation laser-plasma linear collider, and would be essential to facilitate the investigation of fundamental physics and the measurement technology in broad areas.
Xinyi Liu, Qi Xie, Qian Zhao, Hong Wang, Deyu Meng
Motivated by their recent advances, deep learning techniques have been widely applied to low-light image enhancement (LIE) problem. Among which, Retinex theory based ones, mostly following a decomposition-adjustment pipeline, have taken an important place due to its physical interpretation and promising performance. However, current investigations on Retinex based deep learning are still not sufficient, ignoring many useful experiences from traditional methods. Besides, the adjustment step is either performed with simple image processing techniques, or by complicated networks, both of which are unsatisfactory in practice. To address these issues, we propose a new deep learning framework for the LIE problem. The proposed framework contains a decomposition network inspired by algorithm unrolling, and adjustment networks considering both global brightness and local brightness sensitivity. By virtue of algorithm unrolling, both implicit priors learned from data and explicit priors borrowed from traditional methods can be embedded in the network, facilitate to better decomposition. Meanwhile, the consideration of global and local brightness can guide designing simple yet effective network modules for adjustment. Besides, to avoid manually parameter tuning, we also propose a self-supervised fine-tuning strategy, which can always guarantee a promising performance. Experiments on a series of typical LIE datasets demonstrated the effectiveness of the proposed method, both quantitatively and visually, as compared with existing methods.
Qian Zhao, Zhengwei Wu, Zhiqiang Zhang, Jun Zhou
Graph Convolutional Networks (GCNs) has demonstrated promising results for recommender systems, as they can effectively leverage high-order relationship. However, these methods usually encounter data sparsity issue in real-world scenarios. To address this issue, GCN-based recommendation methods employ contrastive learning to introduce self-supervised signals. Despite their effectiveness, these methods lack consideration of the significant degree disparity between head and tail nodes. This can lead to non-uniform representation distribution, which is a crucial factor for the performance of contrastive learning methods. To tackle the above issue, we propose a novel Long-tail Augmented Graph Contrastive Learning (LAGCL) method for recommendation. Specifically, we introduce a learnable long-tail augmentation approach to enhance tail nodes by supplementing predicted neighbor information, and generate contrastive views based on the resulting augmented graph. To make the data augmentation schema learnable, we design an auto drop module to generate pseudo-tail nodes from head nodes and a knowledge transfer module to reconstruct the head nodes from pseudo-tail nodes. Additionally, we employ generative adversarial networks to ensure that the distribution of the generated tail/head nodes matches that of the original tail/head nodes. Extensive experiments conducted on three benchmark datasets demonstrate the significant improvement in performance of our model over the state-of-the-arts. Further analyses demonstrate the uniformity of learned representations and the superiority of LAGCL on long-tail performance. Code is publicly available at https://github.com/im0qianqian/LAGCL
Qian Zhao, Hao Qian, Ziqi Liu, Gong-Duo Zhang, Lihong Gu
Recommendation systems are widely used in e-commerce websites and online platforms to address information overload. However, existing systems primarily rely on historical data and user feedback, making it difficult to capture user intent transitions. Recently, Knowledge Base (KB)-based models are proposed to incorporate expert knowledge, but it struggle to adapt to new items and the evolving e-commerce environment. To address these challenges, we propose a novel Large Language Model based Complementary Knowledge Enhanced Recommendation System (LLM-KERec). It introduces an entity extractor that extracts unified concept terms from item and user information. To provide cost-effective and reliable prior knowledge, entity pairs are generated based on entity popularity and specific strategies. The large language model determines complementary relationships in each entity pair, constructing a complementary knowledge graph. Furthermore, a new complementary recall module and an Entity-Entity-Item (E-E-I) weight decision model refine the scoring of the ranking model using real complementary exposure-click samples. Extensive experiments conducted on three industry datasets demonstrate the significant performance improvement of our model compared to existing approaches. Additionally, detailed analysis shows that LLM-KERec enhances users' enthusiasm for consumption by recommending complementary items. In summary, LLM-KERec addresses the limitations of traditional recommendation systems by incorporating complementary knowledge and utilizing a large language model to capture user intent transitions, adapt to new items, and enhance recommendation efficiency in the evolving e-commerce landscape.
Qian Zhao, Ting Sun, Kun Xue, Feng Wan, Jian-Xing Li
Cascaded Compton scattering and Breit-Wheeler (BW) processes play fundamental roles in high-energy astrophysical sources and laser-driven quantum electrodynamics (QED) plasmas. A thorough comprehension of the polarization transfer in these cascaded processes is essential for elucidating the polarization mechanism of high-energy cosmic gamma rays and laser-driven QED plasmas. In this study, we employ analytical cross-sectional calculations and Monte Carlo (MC) numerical simulations to investigate the polarization transfer in the cascade of electron-seeded inverse Compton scattering (ICS) and BW process. Theoretical analysis indicates that the polarization of background photons can effectively transfer to final-state particles in the first-generation cascade due to helicity transfer. Through MC simulations involving polarized background photons and non-polarized seed electrons, we reveal the characteristic polarization curves as a function of particle energy produced by the cascaded processes of ICS and BW pair production. Our results demonstrate that the first-generation photons from ICS exhibit the non-decayed stair-shape polarization curves, in contrast to the linearly decayed ones of the first-generation electrons. Interestingly, this polarization curve trend can be reversed in the second-generation cascade, facilitated by the presence of polarized first-generation BW pairs with fluctuant polarization curves. The cascade culminates with the production of second-generation BW pairs, due to diminished energy of second-generation photons below the threshold of BW process. Our findings provide crucial insights into the cascaded processes of Compton scattering and BW process, significantly contributing to the understanding and further exploration of laser-driven QED plasma creation in laboratory settings and high-energy astrophysics research.
Jiahong Fu, Hong Wang, Qi Xie, Qian Zhao, Deyu Meng, Zongben Xu
Although current deep learning-based methods have gained promising performance in the blind single image super-resolution (SISR) task, most of them mainly focus on heuristically constructing diverse network architectures and put less emphasis on the explicit embedding of the physical generation mechanism between blur kernels and high-resolution (HR) images. To alleviate this issue, we propose a model-driven deep neural network, called KXNet, for blind SISR. Specifically, to solve the classical SISR model, we propose a simple-yet-effective iterative algorithm. Then by unfolding the involved iterative steps into the corresponding network module, we naturally construct the KXNet. The main specificity of the proposed KXNet is that the entire learning process is fully and explicitly integrated with the inherent physical mechanism underlying this SISR task. Thus, the learned blur kernel has clear physical patterns and the mutually iterative process between blur kernel and HR image can soundly guide the KXNet to be evolved in the right direction. Extensive experiments on synthetic and real data finely demonstrate the superior accuracy and generality of our method beyond the current representative state-of-the-art blind SISR methods. Code is available at: https://github.com/jiahong-fu/KXNet.
Qi Xie, Qian Zhao, Zongben Xu, Deyu Meng
It has been shown that equivariant convolution is very helpful for many types of computer vision tasks. Recently, the 2D filter parametrization technique plays an important role when designing equivariant convolutions. However, the current filter parametrization method still has its evident drawbacks, where the most critical one lies in the accuracy problem of filter representation. Against this issue, in this paper we modify the classical Fourier series expansion for 2D filters, and propose a new set of atomic basis functions for filter parametrization. The proposed filter parametrization method not only finely represents 2D filters with zero error when the filter is not rotated, but also substantially alleviates the fence-effect-caused quality degradation when the filter is rotated. Accordingly, we construct a new equivariant convolution method based on the proposed filter parametrization method, named F-Conv. We prove that the equivariance of the proposed F-Conv is exact in the continuous domain, which becomes approximate only after discretization. Extensive experiments show the superiority of the proposed method. Particularly, we adopt rotation equivariant convolution methods to image super-resolution task, and F-Conv evidently outperforms previous filter parametrization based method in this task, reflecting its intrinsic capability of faithfully preserving rotation symmetries in local image features.
Hong Wang, Qi Xie, Qian Zhao, Yuexiang Li, Yong Liang, Yefeng Zheng, Deyu Meng
As a common weather, rain streaks adversely degrade the image quality. Hence, removing rains from an image has become an important issue in the field. To handle such an ill-posed single image deraining task, in this paper, we specifically build a novel deep architecture, called rain convolutional dictionary network (RCDNet), which embeds the intrinsic priors of rain streaks and has clear interpretability. In specific, we first establish a RCD model for representing rain streaks and utilize the proximal gradient descent technique to design an iterative algorithm only containing simple operators for solving the model. By unfolding it, we then build the RCDNet in which every network module has clear physical meanings and corresponds to each operation involved in the algorithm. This good interpretability greatly facilitates an easy visualization and analysis on what happens inside the network and why it works well in inference process. Moreover, taking into account the domain gap issue in real scenarios, we further design a novel dynamic RCDNet, where the rain kernels can be dynamically inferred corresponding to input rainy images and then help shrink the space for rain layer estimation with few rain maps so as to ensure a fine generalization performance in the inconsistent scenarios of rain types between training and testing data. By end-to-end training such an interpretable network, all involved rain kernels and proximal operators can be automatically extracted, faithfully characterizing the features of both rain and clean background layers, and thus naturally lead to better deraining performance. Comprehensive experiments substantiate the superiority of our method, especially on its well generality to diverse testing scenarios and good interpretability for all its modules. Code is available in \emph{\url{https://github.com/hongwang01/DRCDNet}}.
Qian Zhao, Kaitong Sun, Si Wu, Hai-Feng Li
We synthesized the ferromagnetic EuAgP semiconductor and conducted a comprehensive study of its crystalline, magnetic, heat capacity, band gap, and magnetoresistance properties. Our investigation utilized a combination of X-ray diffraction, optical, and PPMS DynaCool measurements. EuAgP adopts a hexagonal structure with the $P6_3/mmc$ space group. As the temperature decreases, it undergoes a magnetic phase transition from high-temperature paramagnetism to low-temperature ferromagnetism. We determined the ferromagnetic transition temperature to be $T_{\textrm{C}} =$ 16.45(1) K by fitting the measured magnetic susceptibility using a Curie-Weiss law. Heat capacity analysis of EuAgP considered contributions from electrons, phonons, and magnons, revealing $η$ = 0.03 J/mol/$\textrm{K}^\textrm{2}$, indicative of semiconducting behavior. Additionally, we calculated a band gap of $\sim$ 1.324(4) eV based on absorption spectrum measurements. The resistivity versus temperature of EuAgP measured in the absence of an applied magnetic field shows a pronounced peak around $T_{\textrm{C}}$, which diminishes rapidly with increasing applied magnetic fields, ranging from 1 to 14 T. An intriguing phenomenon emerges in the form of a distinct magnetoresistance transition, shifting from positive (e.g., 1.95\% at 300 K and 14 T) to negative (e.g., -30.73\% at 14.25 K and 14 T) as the temperature decreases. This behavior could be attributed to spin-disordered scattering.
Qian Zhao, Jiaqin Wei, Rongming Wang
Apr 30, 2013·q-fin.PM·PDF In this paper, we study the dividend strategies for a shareholder with non-constant discount rate in a diffusion risk model. We assume that the dividends can only be paid at a bounded rate and restrict ourselves to the Markov strategies. This is a time inconsistent control problem. The extended HJB equation is given and the verification theorem is proved for a general discount function. Considering the pseudo-exponential discount functions (Type I and Type II), we get the equilibrium dividend strategies and the equilibrium value functions by solving the extended HJB equations.
Yingqi Ye, Qian Hu, Qian Zhao, Yonggang Meng
We study the repulsive-attractive transition, regarded as stable equilibrium, between gold and dielectric metamaterial based on Mie resonance immersed in various fluids due to the interplay of gravity, buoyancy and Casimir force among different geometries consisting of parallel plates and spheres levitated over substrates. A wider range of separation distance of stable equilibrium is obtained with Mie metamaterial than natural materials. We investigate the relationship between separation distance of stable equilibrium and constructive parameters of Mie metamaterial and geometric parameters of the system, and provide simple rules to tune the equilibrium position by modifying constructive parameters of Mie metamaterial. Particularly, the effect of permeability of Mie metamaterial on equilibrium separation is also considered. Our work is promising for potential applications in frictionless suspension in micro/nanofabrication technologies.
Qian Zhao, Shuo Yang, Binbin Hu, Zhiqiang Zhang, Yakun Wang, Yusong Chen, Jun Zhou, Chuan Shi
Temporal link prediction, as one of the most crucial work in temporal graphs, has attracted lots of attention from the research area. The WSDM Cup 2022 seeks for solutions that predict the existence probabilities of edges within time spans over temporal graph. This paper introduces the solution of AntGraph, which wins the 1st place in the competition. We first analysis the theoretical upper-bound of the performance by removing temporal information, which implies that only structure and attribute information on the graph could achieve great performance. Based on this hypothesis, then we introduce several well-designed features. Finally, experiments conducted on the competition datasets show the superiority of our proposal, which achieved AUC score of 0.666 on dataset A and 0.902 on dataset B, the ablation studies also prove the efficiency of each feature. Code is publicly available at https://github.com/im0qianqian/WSDM2022TGP-AntGraph.
Qian Zhao, Yinghao Zhu, Si Wu, Junchao Xia, Pengfei Zhou, Kaitong Sun, Hai-Feng Li
We have grown a YCrO$_3$ single crystal by the floating-zone method and studied its temperature-dependent crystalline structure and magnetization by X-ray powder diffraction and PPMS DynaCool measurements. All diffraction patterns were well indexed by an orthorhombic structure with space group of $Pbnm$ (No. 62). From 36 to 300 K, no structural phase transition occurs in the pulverized YCrO$_3$ single crystal. The antiferromagnetic phase transition temperature was determined as $T_\textrm{N} =$ 141.58(5) K by the magnetization versus temperature measurements. We found weak ferromagnetic behavior in the magnetic hysteresis loops below $T_\textrm{N}$. Especially, we demonstrated that the antiferromagnetism and weak ferromagnetism appear simutaniously upon cooling. The lattice parameters ($a$, $b$, $c$, and $V$) deviate downward from the Gr$\ddot{\textrm{u}}$neisen law, displaying an anisotropic magnetostriction effect. We extracted temperature variation of the local distortion parameter $Δ$. Compared to the $Δ$ value of Cr ions, Y, O1, and O2 ions show one order of magnitude larger $Δ$ values indicative of much stronger local lattice distortions. Moreover, the calculated bond valence states of Y and O2 ions have obvious subduction charges.