arXiv subjects
Publications and source records attributed to Jing Xu.
We raise a detuning-dependent loss mechanism to describe the soliton formation dynamics when the lumped filtering operation is manipulated in anomalous group velocity dispersion regime, using stability analysis of generalized Lugiato-Lefever equation.
Optical Coherence Tomography Angiography (OCTA) is a non-invasive and non-contacting imaging technique providing visualization of microvasculature of retina and optic nerve head in human eyes in vivo. The adequate image quality of OCTA is the prerequisite for the subsequent quantification of retinal microvasculature. Traditionally, the image quality score based on signal strength is used for discriminating low quality. However, it is insufficient for identifying artefacts such as motion and off-centration, which rely specialized knowledge and need tedious and time-consuming manual identification. One of the most primary issues in OCTA analysis is to sort out the foveal avascular zone (FAZ) region in the retina, which highly correlates with any visual acuity disease. However, the variations in OCTA visual quality affect the performance of deep learning in any downstream marginally. Moreover, filtering the low-quality OCTA images out is both labor-intensive and time-consuming. To address these issues, we develop an automated computer-aided OCTA image processing system using deep neural networks as the classifier and segmentor to help ophthalmologists in clinical diagnosis and research. This system can be an assistive tool as it can process OCTA images of different formats to assess the quality and segment the FAZ area. The source code is freely available at https://github.com/shanzha09/COIPS.git. Another major contribution is the large-scale OCTA dataset, namely OCTA-25K-IQA-SEG we publicize for performance evaluation. It is comprised of four subsets, namely sOCTA-3$\times$3-10k, sOCTA-6$\times$6-14k, sOCTA-3$\times$3-1.1k-seg, and dOCTA-6$\times$6-1.1k-seg, which contains a total number of 25,665 images. The large-scale OCTA dataset is available at https://doi.org/10.5281/zenodo.5111975, https://doi.org/10.5281/zenodo.5111972.
Despite recent improvements in open-domain dialogue models, state of the art models are trained and evaluated on short conversations with little context. In contrast, the long-term conversation setting has hardly been studied. In this work we collect and release a human-human dataset consisting of multiple chat sessions whereby the speaking partners learn about each other's interests and discuss the things they have learnt from past sessions. We show how existing models trained on existing datasets perform poorly in this long-term conversation setting in both automatic and human evaluations, and we study long-context models that can perform much better. In particular, we find retrieval-augmented methods and methods with an ability to summarize and recall previous conversations outperform the standard encoder-decoder architectures currently considered state of the art.
Backdoor attacks represent a serious threat to neural network models. A backdoored model will misclassify the trigger-embedded inputs into an attacker-chosen target label while performing normally on other benign inputs. There are already numerous works on backdoor attacks on neural networks, but only a few works consider graph neural networks (GNNs). As such, there is no intensive research on explaining the impact of trigger injecting position on the performance of backdoor attacks on GNNs. To bridge this gap, we conduct an experimental investigation on the performance of backdoor attacks on GNNs. We apply two powerful GNN explainability approaches to select the optimal trigger injecting position to achieve two attacker objectives -- high attack success rate and low clean accuracy drop. Our empirical results on benchmark datasets and state-of-the-art neural network models demonstrate the proposed method's effectiveness in selecting trigger injecting position for backdoor attacks on GNNs. For instance, on the node classification task, the backdoor attack with trigger injecting position selected by GraphLIME reaches over $84 \%$ attack success rate with less than $2.5 \%$ accuracy drop
Deep Convolutional Neural Networks (DCNNs) and their variants have been widely used in large scale face recognition(FR) recently. Existing methods have achieved good performance on many FR benchmarks. However, most of them suffer from two major problems. First, these methods converge quite slowly since they optimize the loss functions in a high-dimensional and sparse Gaussian Sphere. Second, the high dimensionality of features, despite the powerful descriptive ability, brings difficulty to the optimization, which may lead to a sub-optimal local optimum. To address these problems, we propose a simple yet efficient training mechanism called MultiFace, where we approximate the original high-dimensional features by the ensemble of low-dimensional features. The proposed mechanism is also generic and can be easily applied to many advanced FR models. Moreover, it brings the benefits of good interpretability to FR models via the clustering effect. In detail, the ensemble of these low-dimensional features can capture complementary yet discriminative information, which can increase the intra-class compactness and inter-class separability. Experimental results show that the proposed mechanism can accelerate 2-3 times with the softmax loss and 1.2-1.5 times with Arcface or Cosface, while achieving state-of-the-art performances in several benchmark datasets. Especially, the significant improvements on large-scale datasets(e.g., IJB and MageFace) demonstrate the flexibility of our new training mechanism.
Federated Learning is a distributed machine learning approach which enables model training without data sharing. In this paper, we propose a new federated learning algorithm, Federated Averaging with Client-level Momentum (FedCM), to tackle problems of partial participation and client heterogeneity in real-world federated learning applications. FedCM aggregates global gradient information in previous communication rounds and modifies client gradient descent with a momentum-like term, which can effectively correct the bias and improve the stability of local SGD. We provide theoretical analysis to highlight the benefits of FedCM. We also perform extensive empirical studies and demonstrate that FedCM achieves superior performance in various tasks and is robust to different levels of client numbers, participation rate and client heterogeneity.
Exploiting the relationships between attributes is a key challenge for improving multiple facial attribute recognition. In this work, we are concerned with two types of correlations that are spatial and non-spatial relationships. For the spatial correlation, we aggregate attributes with spatial similarity into a part-based group and then introduce a Group Attention Learning to generate the group attention and the part-based group feature. On the other hand, to discover the non-spatial relationship, we model a group-based Graph Correlation Learning to explore affinities of predefined part-based groups. We utilize such affinity information to control the communication between all groups and then refine the learned group features. Overall, we propose a unified network called Multi-scale Group and Graph Network. It incorporates these two newly proposed learning strategies and produces coarse-to-fine graph-based group features for improving facial attribute recognition. Comprehensive experiments demonstrate that our approach outperforms the state-of-the-art methods.
Convolutional Neural Networks (CNNs) have achieved tremendous success in a number of learning tasks including image classification. Recent advanced models in CNNs, such as ResNets, mainly focus on the skip connection to avoid gradient vanishing. DenseNet designs suggest creating additional bypasses to transfer features as an alternative strategy in network design. In this paper, we design Attentive Feature Integration (AFI) modules, which are widely applicable to most recent network architectures, leading to new architectures named AFI-Nets. AFI-Nets explicitly model the correlations among different levels of features and selectively transfer features with a little overhead.AFI-ResNet-152 obtains a 1.24% relative improvement on the ImageNet dataset while decreases the FLOPs by about 10% and the number of parameters by about 9.2% compared to ResNet-152.
Spectrum anomaly detection is of great importance in wireless communication to secure safety and improve spectrum efficiency. However, spectrum anomaly detection faces many difficulties, especially in unauthorized frequency bands. For example, the composition of unauthorized frequency bands is very complex and the abnormal usage patterns are unknown in prior. In this paper, a noise attention method is proposed for unsupervised spectrum anomaly detection in unauthorized bands. First of all, we theoretically prove that the anomalies in unauthorized bands will raise the noise floor of spectrogram after VAE reconstruction. Then, we introduce a novel anomaly metric named as noise attention score to more effectively capture spectrum anomaly. The effectiveness of the proposed method is experimentally verified in 2.4 GHz ISM band. Leveraging the noise attention score, the AUC metric of anomaly detection is increased by 0.193. The proposed method is beneficial to reliably detecting abnormal spectrum while keeping low false alarm rate.
We consider a subclass of the Cowen-Douglas class in which the problem of deciding whether two operators are similar becomes more manageable. A similarity criterion for Cowen-Douglas operators is known to be dependent on the trace of the curvatures of the corresponding eigenvector bundles. Unless the given eignvector bundle is a line bundle, the computation of the curvatures, in general, is not so simple as one might hope. By using a structure theorem given in \cite{JW}, we reduce the problem of finding the trace of the curvatures to looking at the curvatures of the associated line bundles. Moreover, several questions related to the similarity problem are also taken into account.
The rapid spread of the new pandemic, i.e., COVID-19, has severely threatened global health. Deep-learning-based computer-aided screening, e.g., COVID-19 infected CT area segmentation, has attracted much attention. However, the publicly available COVID-19 training data are limited, easily causing overfitting for traditional deep learning methods that are usually data-hungry with millions of parameters. On the other hand, fast training/testing and low computational cost are also necessary for quick deployment and development of COVID-19 screening systems, but traditional deep learning methods are usually computationally intensive. To address the above problems, we propose MiniSeg, a lightweight deep learning model for efficient COVID-19 segmentation. Compared with traditional segmentation methods, MiniSeg has several significant strengths: i) it only has 83K parameters and is thus not easy to overfit; ii) it has high computational efficiency and is thus convenient for practical deployment; iii) it can be fast retrained by other users using their private COVID-19 data for further improving performance. In addition, we build a comprehensive COVID-19 segmentation benchmark for comparing MiniSeg to traditional methods.
Scattering immune propagation of light in topological photonic systems may revolutionarize the design of integrated photonic circuits for information processing and communications. In optics, various photonic topological circuits have been developed, which were based on classical emulation of either quantum spin Hall effect or quantum valley Hall effect. On the other hand, the combination of both the valley and spin degrees of freedom can lead to a new kind of topological transport phenomenon, dubbed quantum spin valley Hall effect (QSVH), which can further expand the number of topologically protected edge channels and would be useful for information multiplexing. However, it is challenging to realize QSVH in most known material platforms, due to the requirement of breaking both the (pseudo-)fermionic time-reversal (T) and parity symmetries (P) individually, but leaving the combined symmetry S=TP intact. Here, we propose an experimentally feasible platform to realize QSVH for light, based on coupled ring resonators mediated by optical Kerr nonlinearity. Thanks to the inherent flexibility of cross-mode modulation (XMM), the coupling between the probe light can be engineered in a controllable way such that spin-dependent staggered sublattice potential emerges in the effective Hamiltonian. With delicate yet experimentally feasible pump conditions, we show the existence of spin valley Hall induced topological edge states. We further demonstrate that both degrees of freedom, i.e., spin and valley, can be manipulated simultaneously in a reconfigurable manner to realize spin-valley photonics, doubling the degrees of freedom for enhancing the information capacity in optical communication systems.
Negative longitudinal magnetoresistance (NLMR) has been reported in a variety of materials and has attracted extensive attention as an electrotransport hallmark of topological Weyl semimetals. However, its origin is still under debate. Here, we demonstrate that the NLMR in a two dimensional electron gas can be influenced by the measurement current. While the NLMR persists up to 130 K, its magnitude and magnetic field response become dependent on the applied current below 60 K. The tunable NLMR at low and high currents can be best attributed to quantum interference and disorder scattering effects, respectively. This work uncovers non-Ohmic NLMR in a non-Weyl material and highlights potential effects of the measurement current in elucidating electrotransport phenomena. We also demonstrate that NLMRs can be a valuable phenomenon in revealing the origins of other properties, such as negative MRs in perpendicular magnetic fields.
The ResNet and its variants have achieved remarkable successes in various computer vision tasks. Despite its success in making gradient flow through building blocks, the simple shortcut connection mechanism limits the ability of re-exploring new potentially complementary features due to the additive function. To address this issue, in this paper, we propose to introduce a regulator module as a memory mechanism to extract complementary features, which are further fed to the ResNet. In particular, the regulator module is composed of convolutional RNNs (e.g., Convolutional LSTMs or Convolutional GRUs), which are shown to be good at extracting Spatio-temporal information. We named the new regulated networks as RegNet. The regulator module can be easily implemented and appended to any ResNet architecture. We also apply the regulator module for improving the Squeeze-and-Excitation ResNet to show the generalization ability of our method. Experimental results on three image classification datasets have demonstrated the promising performance of the proposed architecture compared with the standard ResNet, SE-ResNet, and other state-of-the-art architectures.
The ability to control the potential landscape in a medium of interacting particles could lead to intriguing collective behavior and innovative functionalities. Here, we utilize spatially reconfigurable magnetic potentials of a pinwheel artificial spin ice structure to tailor the motion of superconducting vortices. The reconstituted chain structures of the magnetic charges in the artificial pinwheel spin ice and the strong interaction between magnetic charges and superconducting vortices allow significant modification of the transport properties of the underlying superconducting thin film, resulting in a reprogrammable resistance state that enables a reversible and switchable vortex Hall effect. Our results highlight an effective and simple method of using artificial spin ice as an in-situ reconfigurable nanoscale energy landscape to design reprogrammable superconducting electronics, which could also be applied to the in-situ control of properties and functionalities in other magnetic particle systems, such as magnetic skyrmions.
Hybrid magnonics has recently attracted intensive attentions as a promising platform for coherent information processing. In spite of its rapid development, on-demand control over the interaction of magnons with other information carriers, in particular microwave photons in electromagnonic systems, has been long missing, significantly limiting the broad applications of hybrid magnonics. Here, we show that by introducing Floquet engineering into cavity electromagnonics, coherent control on the magnon-microwave photon coupling can be realized. Leveraging the periodic temporal modulation from a Floquet drive, our first-of-its-kind Floquet cavity electromagnonic system can manipulate the interaction between hybridized cavity electromagnonic modes on demand. Moreover, we demonstrate a new coupling regime in such systems: the Floquet ultrastrong coupling, where the Floquet splitting is comparable with or even larger than the level spacing of the two interacting modes, resulting in the breakdown of the rotating wave approximation. Our findings open up new directions for magnon-based coherent signal processing.
Maximum target coverage by adjusting the orientation of distributed sensors is an important problem in directional sensor networks (DSNs). This problem is challenging as the targets usually move randomly but the coverage range of sensors is limited in angle and distance. Thus, it is required to coordinate sensors to get ideal target coverage with low power consumption, e.g. no missing targets or reducing redundant coverage. To realize this, we propose a Hierarchical Target-oriented Multi-Agent Coordination (HiT-MAC), which decomposes the target coverage problem into two-level tasks: targets assignment by a coordinator and tracking assigned targets by executors. Specifically, the coordinator periodically monitors the environment globally and allocates targets to each executor. In turn, the executor only needs to track its assigned targets. To effectively learn the HiT-MAC by reinforcement learning, we further introduce a bunch of practical methods, including a self-attention module, marginal contribution approximation for the coordinator, goal-conditional observation filter for the executor, etc. Empirical results demonstrate the advantage of HiT-MAC in coverage rate, learning efficiency,and scalability, comparing to baselines. We also conduct an ablative analysis on the effectiveness of the introduced components in the framework.
As the fifth-generation (5G) mobile communication system is being commercialized, extensive studies on the evolution of 5G and sixth-generation mobile communication systems have been conducted. Future mobile communication systems are evidently evolving towards a more intelligent and software-reconfigurable functionality paradigm that can provide ubiquitous communication and also sense, control, and optimize wireless environments. Thus, integrating communication and localization by utilizing the highly directional transmission characteristics of millimeter-wave (mmWave) is a promising route. This approach not only expands the localization capabilities of a communication system but also provides new concepts and opportunities to enhance communication. In this paper, we explain the integrated communication and localization in mmWave systems, in which these processes share the same set of hardware architecture and algorithm. We also perform an overview of the key enabling technologies and the basic knowledge on localization. Then, we provide two promising directions for studies on localization with extremely large antenna array and model-based neural networks. We also discuss a comprehensive guidance for location-assisted mmWave communications in terms of channel estimation, channel state information feedback, beam tracking, synchronization, interference control, resource allocation, and user selection. Finally, this paper outlines the future trends on the mutual assistance and enhancement of communication and localization in integrated systems.