arXiv subjects

Xu Zheng

Publications and source records attributed to Xu Zheng.

At least 145 records · Page 8Linked to original sources

arXiv2023

Look at the Neighbor: Distortion-aware Unsupervised Domain Adaptation for Panoramic Semantic Segmentation

Endeavors have been recently made to transfer knowledge from the labeled pinhole image domain to the unlabeled panoramic image domain via Unsupervised Domain Adaptation (UDA). The aim is to tackle the domain gaps caused by the style disparities and distortion problem from the non-uniformly distributed pixels of equirectangular projection (ERP). Previous works typically focus on transferring knowledge based on geometric priors with specially designed multi-branch network architectures. As a result, considerable computational costs are induced, and meanwhile, their generalization abilities are profoundly hindered by the variation of distortion among pixels. In this paper, we find that the pixels' neighborhood regions of the ERP indeed introduce less distortion. Intuitively, we propose a novel UDA framework that can effectively address the distortion problems for panoramic semantic segmentation. In comparison, our method is simpler, easier to implement, and more computationally efficient. Specifically, we propose distortion-aware attention (DA) capturing the neighboring pixel distribution without using any geometric constraints. Moreover, we propose a class-wise feature aggregation (CFA) module to iteratively update the feature representations with a memory bank. As such, the feature similarity between two domains can be consistently optimized. Extensive experiments show that our method achieves new state-of-the-art performance while remarkably reducing 80% parameters.

cs.CV↗

arXiv2023

Priors in Deep Image Restoration and Enhancement: A Survey

Image restoration and enhancement is a process of improving the image quality by removing degradations, such as noise, blur, and resolution degradation. Deep learning (DL) has recently been applied to image restoration and enhancement. Due to its ill-posed property, plenty of works have been explored priors to facilitate training deep neural networks (DNNs). However, the importance of priors has not been systematically studied and analyzed by far in the research community. Therefore, this paper serves as the first study that provides a comprehensive overview of recent advancements in priors for deep image restoration and enhancement. Our work covers five primary contents: (1) A theoretical analysis of priors for deep image restoration and enhancement; (2) A hierarchical and structural taxonomy of priors commonly used in the DL-based methods; (3) An insightful discussion on each prior regarding its principle, potential, and applications; (4) A summary of crucial problems by highlighting the potential future directions, especially adopting the large-scale foundation models as prior, to spark more research in the community; (5) An open-source repository that provides a taxonomy of all mentioned works and code links.

cs.CV↗

arXiv2023

A Good Student is Cooperative and Reliable: CNN-Transformer Collaborative Learning for Semantic Segmentation

In this paper, we strive to answer the question "how to collaboratively learn convolutional neural network (CNN)-based and vision transformer (ViT)-based models by selecting and exchanging the reliable knowledge between them for semantic segmentation?" Accordingly, we propose an online knowledge distillation (KD) framework that can simultaneously learn compact yet effective CNN-based and ViT-based models with two key technical breakthroughs to take full advantage of CNNs and ViT while compensating their limitations. Firstly, we propose heterogeneous feature distillation (HFD) to improve students' consistency in low-layer feature space by mimicking heterogeneous features between CNNs and ViT. Secondly, to facilitate the two students to learn reliable knowledge from each other, we propose bidirectional selective distillation (BSD) that can dynamically transfer selective knowledge. This is achieved by 1) region-wise BSD determining the directions of knowledge transferred between the corresponding regions in the feature space and 2) pixel-wise BSD discerning which of the prediction knowledge to be transferred in the logit space. Extensive experiments on three benchmark datasets demonstrate that our proposed framework outperforms the state-of-the-art online distillation methods by a large margin, and shows its efficacy in learning collaboratively between ViT-based and CNN-based models.

cs.CV↗

arXiv2023

Both Style and Distortion Matter: Dual-Path Unsupervised Domain Adaptation for Panoramic Semantic Segmentation

The ability of scene understanding has sparked active research for panoramic image semantic segmentation. However, the performance is hampered by distortion of the equirectangular projection (ERP) and a lack of pixel-wise annotations. For this reason, some works treat the ERP and pinhole images equally and transfer knowledge from the pinhole to ERP images via unsupervised domain adaptation (UDA). However, they fail to handle the domain gaps caused by: 1) the inherent differences between camera sensors and captured scenes; 2) the distinct image formats (e.g., ERP and pinhole images). In this paper, we propose a novel yet flexible dual-path UDA framework, DPPASS, taking ERP and tangent projection (TP) images as inputs. To reduce the domain gaps, we propose cross-projection and intra-projection training. The cross-projection training includes tangent-wise feature contrastive training and prediction consistency training. That is, the former formulates the features with the same projection locations as positive examples and vice versa, for the models' awareness of distortion, while the latter ensures the consistency of cross-model predictions between the ERP and TP. Moreover, adversarial intra-projection training is proposed to reduce the inherent gap, between the features of the pinhole images and those of the ERP and TP images, respectively. Importantly, the TP path can be freely removed after training, leading to no additional inference cost. Extensive experiments on two benchmarks show that our DPPASS achieves +1.06$\%$ mIoU increment than the state-of-the-art approaches.

cs.CV↗

arXiv2023

Continuum of Bound States in a Non-Hermitian Model

In a Hermitian system, bound states must have quantized energies, whereas extended states can form a continuum. We demonstrate how this principle fails for non-Hermitian systems, by analyzing non-Hermitian continuous Hamiltonians with an imaginary momentum and Landau-type vector potential. The eigenstates, which we call ``continuum Landau modes'' (CLMs), have gaussian spatial envelopes and form a continuum filling the complex energy plane. We present experimentally-realizable 1D and 2D lattice models that can be used to study CLMs; the lattice eigenstates are localized and have other features that are the same as in the continuous model. One of these lattices can serve as a rainbow trap, whereby the response to an excitation is concentrated at a position proportional to the frequency. Another lattice can act a wave funnel, concentrating an input excitation onto a boundary over a wide frequency bandwidth. Unlike recent funneling schemes based on the non-Hermitian skin effect, this requires only a simple lattice design without nonreciprocal couplings.

quant-ph↗

arXiv2022

Evaluation of the Synthetic Electronic Health Records

Generative models have been found effective for data synthesis due to their ability to capture complex underlying data distributions. The quality of generated data from these models is commonly evaluated by visual inspection for image datasets or downstream analytical tasks for tabular datasets. These evaluation methods neither measure the implicit data distribution nor consider the data privacy issues, and it remains an open question of how to compare and rank different generative models. Medical data can be sensitive, so it is of great importance to draw privacy concerns of patients while maintaining the data utility of the synthetic dataset. Beyond the utility evaluation, this work outlines two metrics called Similarity and Uniqueness for sample-wise assessment of synthetic datasets. We demonstrate the proposed notions with several state-of-the-art generative models to synthesise Cystic Fibrosis (CF) patients' electronic health records (EHRs), observing that the proposed metrics are suitable for synthetic data evaluation and generative model comparison.

cs.LG↗

arXiv2022

Equivalence of fluctuation-dissipation and Edwards' temperature in cyclically sheared granular systems

Using particle trajectory data obtained from x-ray tomography, we determine two kinds of effective temperatures in a cyclically sheared granular system. The first one is obtained from the fluctuation-dissipation theorem which relates the diffusion and mobility of lighter tracer particles immersed in the system. The second is the Edwards compactivity defined via the packing volume fluctuations. We find robust excellent agreement between these two temperatures, independent of the type of the tracers, cyclic shear amplitudes, and particle surface roughness. We further elucidate that in granular systems the viscous-like drag force is due to the broken symmetry of the local contact geometry.

cond-mat.soft↗

arXiv2022

Synthesising Electronic Health Records: Cystic Fibrosis Patient Group

Class imbalance can often degrade predictive performance of supervised learning algorithms. Balanced classes can be obtained by oversampling exact copies, with noise, or interpolation between nearest neighbours (as in traditional SMOTE methods). Oversampling tabular data using augmentation, as is typical in computer vision tasks, can be achieved with deep generative models. Deep generative models are effective data synthesisers due to their ability to capture complex underlying distributions. Synthetic data in healthcare can enhance interoperability between healthcare providers by ensuring patient privacy. Equipped with large synthetic datasets which do well to represent small patient groups, machine learning in healthcare can address the current challenges of bias and generalisability. This paper evaluates synthetic data generators ability to synthesise patient electronic health records. We test the utility of synthetic data for patient outcome classification, observing increased predictive performance when augmenting imbalanced datasets with synthetic data.

cs.LG↗

arXiv2021

Uncertainty-Aware Deep Co-training for Semi-supervised Medical Image Segmentation

Semi-supervised learning has made significant strides in the medical domain since it alleviates the heavy burden of collecting abundant pixel-wise annotated data for semantic segmentation tasks. Existing semi-supervised approaches enhance the ability to extract features from unlabeled data with prior knowledge obtained from limited labeled data. However, due to the scarcity of labeled data, the features extracted by the models are limited in supervised learning, and the quality of predictions for unlabeled data also cannot be guaranteed. Both will impede consistency training. To this end, we proposed a novel uncertainty-aware scheme to make models learn regions purposefully. Specifically, we employ Monte Carlo Sampling as an estimation method to attain an uncertainty map, which can serve as a weight for losses to force the models to focus on the valuable region according to the characteristics of supervised learning and unsupervised learning. Simultaneously, in the backward process, we joint unsupervised and supervised losses to accelerate the convergence of the network via enhancing the gradient flow between different tasks. Quantitatively, we conduct extensive experiments on three challenging medical datasets. Experimental results show desirable improvements to state-of-the-art counterparts.

cs.CV↗

arXiv2021

A Graph-based Imputation Method for Sparse Medical Records

Electronic Medical Records (EHR) are extremely sparse. Only a small proportion of events (symptoms, diagnoses, and treatments) are observed in the lifetime of an individual. The high degree of missingness of EHR can be attributed to a large number of factors, including device failure, privacy concerns, or other unexpected reasons. Unfortunately, many traditional imputation methods are not well suited for highly sparse data and scale poorly to high dimensional datasets. In this paper, we propose a graph-based imputation method that is both robust to sparsity and to unreliable unmeasured events. Our approach compares favourably to several standard and state-of-the-art imputation methods in terms of performance and runtime. Moreover, results indicate that the model learns to embed different event types in a clinically meaningful way. Our work can facilitate the diagnosis of novel diseases based on the clinical history of past events, with the potential to increase our understanding of the landscape of comorbidities.

cs.AI↗

arXiv2021

Network Generation with Differential Privacy

We consider the problem of generating private synthetic versions of real-world graphs containing private information while maintaining the utility of generated graphs. Differential privacy is a gold standard for data privacy, and the introduction of the differentially private stochastic gradient descent (DP-SGD) algorithm has facilitated the training of private neural models in a number of domains. Recent advances in graph generation via deep generative networks have produced several high performing models. We evaluate and compare state-of-the-art models including adjacency matrix based models and edge based models, and show a practical implementation that favours the edge-list approach utilizing the Gaussian noise mechanism when evaluated on commonly used graph datasets. Based on our findings, we propose a generative model that can reproduce the properties of real-world networks while maintaining edge-differential privacy. The proposed model is based on a stochastic neural network that generates discrete edge-list samples and is trained using the Wasserstein GAN objective with the DP-SGD optimizer. Being the first approach to combine these beneficial properties, our model contributes to further research on graph data privacy.

cs.LG↗

arXiv2021

Fröhlich Condensate of Phonons in Optomechanical Systems

We propose that the optomechanical systems can be potential platforms to implement the Fröhlich condensate of phonons. We consider a one-dimensional array of membranes coupled to the cavity field via a quadratic interaction, and the cavity is pumped by an external laser. Analytical and numerical results predict that the high phonon occupancy of the lowest or highest mechanical mode is achievable depending on the detuning of the driving laser, the optomechnical strength, and the temperature. The decoherence of the Fröhlich condensate can be largely suppressed by the large number of membranes. Our results shed light on narrow-linewidth phonon laser, energy conversion/transfer, and efficient multimode cooling.

quant-ph↗

arXiv2021

Reciprocity of thermal diffusion in time-modulated systems

The reciprocity principle governs the symmetry in transmission of electromagnetic and acoustic waves, as well as the diffusion of heat between two points in space, with important consequences for thermal management and energy harvesting. There has been significant recent interest in materials with time-modulated properties, which have been shown to efficiently break reciprocity for light, sound, and even charge diffusion. Quite surprisingly, here we show that, from a practical point of view, time modulation cannot generally be used to break reciprocity for thermal diffusion. We establish a theoretical framework to accurately describe the behavior of diffusive processes under time modulation, and prove that thermal reciprocity in dynamic materials is generally preserved by the continuity equation, unless some external bias or special material is considered. We then experimentally demonstrate reciprocal heat transfer in a time-modulated device. Our findings correct previous misconceptions regarding reciprocity breaking for thermal diffusion, revealing the generality of symmetry constraints in heat transfer, and clarifying its differences from other transport processes in what concerns the principles of reciprocity and microscopic reversibility.

physics.app-ph↗

arXiv2020

Transforming heat transfer with thermal metamaterials and devices

The demand for sophisticated tools and approaches in heat management and control has triggered fast development of emerging fields including conductive thermal metamaterials, nanophononics, far-field and near-field radiative thermal management, etc. In this review, we cast a unified perspective on the control of heat transfer, based on which the related studies can be considered as complementary paradigms toward manipulating physical parameters and realizing unprecedented phenomena in heat transfer using artificial structures, such as thermal conductivity in heat conduction, thermal emissivity in radiation, and properties related to multi-physical effects. The review is divided into three parts that focus on the three main categories of heat flow control, respectively. Thermal conduction and radiation are emphasized in the first and second parts at both macro- and micro-scale. The third part discusses the efforts to actively introduce heat sources or tune the material parameters with multi-physical effects in both conduction and radiation, including works using thermal convection. We conclude the review with challenges in this research topic and new possibilities about topological thermal effects, heat waves, and quantum thermal effects.

physics.app-ph↗

arXiv2020

A Systematic Literature Review of Modern Software Visualization

We report on the state-of-the-art of software visualization. To ensure reproducibility, we adopted the Systematic Literature Review methodology. That is, we analyzed 1440 entries from IEEE Xplore and ACM Digital Library databases. We selected 105 relevant full papers published in 2013-2019, which we classified based on the aspect of the software system that is supported (i.e., structure, behavior, and evolution). For each paper, we extracted main dimensions that characterize software visualizations, such as software engineering tasks, roles of users, information visualization techniques, and media used to display visualizations. We provide researchers in the field an overview of the state-of-the-art in software visualization and highlight research opportunities. We also help developers to identify suitable visualizations for their particular context by matching software visualizations to development concerns and concrete details to obtain available visualization tools.

cs.SE↗

arXiv2020

Effect of Interfacial Thermal Resistance in Thermal Cloak

When heat transfers through interface between two different materials, it will encounter an interfacial thermal resistance (ITR) that makes the temperature discontinuous. This effect has been totally neglected so far in the research of thermal cloak, in particular when the thermal cloak is built with multilayer structures. In this paper, we investigate the effect of ITR on the performance of the thermal cloak by using both analytical and numerical method. Our results show that the existence of ITR will distort the external field, thus destroy the cloak. Moreover, we found that the effect of ITR can be quantified by a parameter called characteristic length.

physics.app-ph↗

arXiv2019

STaDA: Style Transfer as Data Augmentation

The success of training deep Convolutional Neural Networks (CNNs) heavily depends on a significant amount of labelled data. Recent research has found that neural style transfer algorithms can apply the artistic style of one image to another image without changing the latter's high-level semantic content, which makes it feasible to employ neural style transfer as a data augmentation method to add more variation to the training dataset. The contribution of this paper is a thorough evaluation of the effectiveness of the neural style transfer as a data augmentation method for image classification tasks. We explore the state-of-the-art neural style transfer algorithms and apply them as a data augmentation method on Caltech 101 and Caltech 256 dataset, where we found around 2% improvement from 83% to 85% of the image classification accuracy with VGG16, compared with traditional data augmentation strategies. We also combine this new method with conventional data augmentation approaches to further improve the performance of image classification. This work shows the potential of neural style transfer in computer vision field, such as helping us to reduce the difficulty of collecting sufficient labelled data and improve the performance of generic image-based deep learning algorithms.

cs.CV↗

arXiv2019

Observation of Complete Photonic Bandgap in Low Refractive Index Contrast Inverse Rod-Connected Diamond Structured Chalcogenides

Three-dimensional complete photonic bandgap materials or photonic crystals block light propagation in all directions. The rod-connected diamond structure exhibits the largest photonic bandgap known to date and supports a complete bandgap for the lowest refractive index contrast ratio down to $n_{high}/n_{low} \sim 1.9$. We confirm this threshold by measuring a complete photonic bandgap in the infrared region in Sn--S--O $(n\sim1.9)$ and Ge--Sb--S--O $(n\sim2)$ inverse rod-connected diamond structures. The structures were fabricated using a low-temperature chemical vapor deposition process via a single-inversion technique. This provides a reliable fabrication technique of complete photonic bandgap materials and expands the library of backfilling materials, leading to a wide range of future photonic applications.

physics.optics↗