arXiv subjects

Kai Zhang

Publications and source records attributed to Kai Zhang.

At least 541 records · Page 30Linked to original sources

arXiv2021

Designing a Practical Degradation Model for Deep Blind Image Super-Resolution

It is widely acknowledged that single image super-resolution (SISR) methods would not perform well if the assumed degradation model deviates from those in real images. Although several degradation models take additional factors into consideration, such as blur, they are still not effective enough to cover the diverse degradations of real images. To address this issue, this paper proposes to design a more complex but practical degradation model that consists of randomly shuffled blur, downsampling and noise degradations. Specifically, the blur is approximated by two convolutions with isotropic and anisotropic Gaussian kernels; the downsampling is randomly chosen from nearest, bilinear and bicubic interpolations; the noise is synthesized by adding Gaussian noise with different noise levels, adopting JPEG compression with different quality factors, and generating processed camera sensor noise via reverse-forward camera image signal processing (ISP) pipeline model and RAW image noise model. To verify the effectiveness of the new degradation model, we have trained a deep blind ESRGAN super-resolver and then applied it to super-resolve both synthetic and real images with diverse degradations. The experimental results demonstrate that the new degradation model can help to significantly improve the practicability of deep super-resolvers, thus providing a powerful alternative solution for real SISR applications.

eess.IV↗

arXiv2021

Towards Flexible Blind JPEG Artifacts Removal

Training a single deep blind model to handle different quality factors for JPEG image artifacts removal has been attracting considerable attention due to its convenience for practical usage. However, existing deep blind methods usually directly reconstruct the image without predicting the quality factor, thus lacking the flexibility to control the output as the non-blind methods. To remedy this problem, in this paper, we propose a flexible blind convolutional neural network, namely FBCNN, that can predict the adjustable quality factor to control the trade-off between artifacts removal and details preservation. Specifically, FBCNN decouples the quality factor from the JPEG image via a decoupler module and then embeds the predicted quality factor into the subsequent reconstructor module through a quality factor attention block for flexible control. Besides, we find existing methods are prone to fail on non-aligned double JPEG images even with only a one-pixel shift, and we thus propose a double JPEG degradation model to augment the training data. Extensive experiments on single JPEG images, more general double JPEG images, and real-world JPEG images demonstrate that our proposed FBCNN achieves favorable performance against state-of-the-art methods in terms of both quantitative metrics and visual quality.

eess.IV↗

arXiv2021

A Multi-Stencil Fast Marching Method with Path Correction for Efficient Reservoir Simulation and Automated History Matching

The efficiency of reservoir simulation is important for automated history matching (AHM) and production optimization, etc. The fast marching marching method (FMM) has been used for efficient reservoir simulation. FMM can be regarded as a generalised streamline method but without the need to construct streamlines. In FMM reservoir simulation, the Eikonal equation for the diffusive time-of-flight (DTOF) is solved by FMM and then the governing equations are computed on the 1D DTOF coordinate. Standard FMM solves the Eikonal equation using a 4-stencil algorithm on 2D Cartesian grids, ignoring the diagonal neighbouring cells. In the current study, we build a 8-stencil algorithm considering all neighbouring cells, and use local analytical propagation speeds. In addition, a local path-correction coefficient is introduced to further increase the accuracy of DTOF solution. Next, a discretisation scheme is built on the 1D DTOF coordinate for efficient reservoir simulation. The algorithm is validated on homogeneous and heterogeneous test cases, and its potential for efficient forward simulation in AHM is demonstrated by two examples of dimensions 2 and 6.

physics.flu-dyn↗

arXiv2021

Low-Reynolds-number aerodynamic characteristics of airfoils with piezocomposite trailing control surfaces

Morphing wings comprised of fixed leading sections with piezocomposite trailing control surfaces have emerged as a novel active control technique for unmanned aerial vehicles. However, the wake dynamics and aerodynamic performance of such hybrid airfoil configuration has not been thoroughly investigated. In this paper, direct numerical simulations of two-dimensional flows over hybrid airfoils comprised of NACA 0012 leading sections with piezocomposite trailing control surfaces are performed at a fixed Reynolds number of 1000. The effects of length and camber of the trailing control surface on the laminar aerodynamic characteristics are studied over a wide range of angle of attack. It is shown that the flow behind the airfoil exhibits different features, including steady flow, periodic vortex shedding, and quasi-periodic vortex shedding for different configurations. The transition between these wake states occurs at slightly smaller angles of attack compared to the nominal NACA 0012 airfoil. While the drag coefficient remains close to each other at a fixed angle of attack, the lift coefficient of the hybrid airfoil is positively affected by the length and camber of the trailing control surface. The mechanism of lift generation is examined by surface pressure distributions and a force element analysis. It is revealed that with increased camber of the trailing control surface, the flow on both the suction and pressure sides of the airfoil are modified in a beneficial way to enhance lift. Increasing the length ratio only significantly modifies the flow near the aft section on the pressure side. The results herein provide a laminar aerodynamic characterization of hybrid airfoils with trailing control surfaces, and could potentially aid the design of control techniques for next-generation small unmanned aerial vehicles.

physics.flu-dyn↗

arXiv2021

Laminar vortex dynamics around forward-swept wings

Forward-swept wings offer unique advantages in the aerodynamic performance of air vehicles. However, the low-Reynolds-number characteristics of such wings have not been explored in the past. In this work, we numerically study laminar separated flows over forward-swept wings with semi aspect ratios $sAR=0.5$ to 2 at a chord-based Reynolds number of 400. Forward-swept wings generate wakes that are significantly different from those of backward-swept wings. For low-aspect-ratio forward wings, the wakes remain steady due to the strong downwash effects induced by the tip vortices. For larger aspect ratio, the downwash effects weaken over the inboard regions of the wing, allowing unsteady vortex shedding to occur. Further larger aspect ratio allows for the formation of streamwise vortices for highly-swept wings, stabilizing the flow. Forward-swept wings can generate enhanced lift at high angles of attack than the unswept and backward-swept wings, with the cost of high drag. We show through force element analysis that the increased lift of forward-swept wings is attributed to the vortical structure that is maintained by the tip-vortex-induced downwash over the outboard region of wing span. The current findings offer a detailed understanding of the sweep effects on laminar separated flows over forward-swept wings, and invite innovative designs of high-lift devices.

physics.flu-dyn↗

arXiv2021

GradTS: A Gradient-Based Automatic Auxiliary Task Selection Method Based on Transformer Networks

A key problem in multi-task learning (MTL) research is how to select high-quality auxiliary tasks automatically. This paper presents GradTS, an automatic auxiliary task selection method based on gradient calculation in Transformer-based models. Compared to AUTOSEM, a strong baseline method, GradTS improves the performance of MT-DNN with a bert-base-cased backend model, from 0.33% to 17.93% on 8 natural language understanding (NLU) tasks in the GLUE benchmarks. GradTS is also time-saving since (1) its gradient calculations are based on single-task experiments and (2) the gradients are re-used without additional experiments when the candidate task set changes. On the 8 GLUE classification tasks, for example, GradTS costs on average 21.32% less time than AUTOSEM with comparable GPU consumption. Further, we show the robustness of GradTS across various task settings and model selections, e.g. mixed objectives among candidate tasks. The efficiency and efficacy of GradTS in these case studies illustrate its general applicability in MTL research without requiring manual task filtering or costly parameter tuning.

cs.LG↗

arXiv2021

A Fast PC Algorithm with Reversed-order Pruning and A Parallelization Strategy

The PC algorithm is the state-of-the-art algorithm for causal structure discovery on observational data. It can be computationally expensive in the worst case due to the conditional independence tests are performed in an exhaustive-searching manner. This makes the algorithm computationally intractable when the task contains several hundred or thousand nodes, particularly when the true underlying causal graph is dense. We propose a critical observation that the conditional set rendering two nodes independent is non-unique, and including certain redundant nodes do not sacrifice result accuracy. Based on this finding, the innovations of our work are two-folds. First, we innovate on a reserve order linkage pruning PC algorithm which significantly increases the algorithm's efficiency. Second, we propose a parallel computing strategy for statistical independence tests by leveraging tensor computation, which brings further speedup. We also prove the proposed algorithm does not induce statistical power loss under mild graph and data dimensionality assumptions. Experimental results show that the single-threaded version of the proposed algorithm can achieve a 6-fold speedup compared to the PC algorithm on a dense 95-node graph, and the parallel version can make a 825-fold speed-up. We also provide proof that the proposed algorithm is consistent under the same set of conditions with conventional PC algorithm.

cs.LG↗

arXiv2021

On the Fundamental Limits of Coded Caching Systems with Restricted Demand Types

Caching is a technique to reduce the communication load in peak hours by prefetching contents during off-peak hours. An information-theoretic framework for coded caching was introduced by Maddah-Ali and Niesen in a recent work, where it was shown that significant improvement can be obtained compared to uncoded caching. Considerable efforts have been devoted to identify the precise information-theoretic fundamental limits of the coded caching systems, however the difficulty of this task has also become clear. One of the reasons for this difficulty is that the original coded caching setting allows all possible multiple demand types during delivery, which in fact introduces tension in the coding strategy. In this paper, we seek to develop a better understanding of the fundamental limits of coded caching by investigating systems with certain demand type restrictions. We first consider the canonical three-user three-file system, and show that, contrary to popular beliefs, the worst demand type is not the one in which all three files are requested. Motivated by these findings, we focus on coded caching systems where every file must be requested by at least one user. A novel coding scheme is proposed, which can provide new operating points that are not covered by any previously known schemes.

cs.IT↗

arXiv2021

Nanoscale spin detection of copper ions using double electron-electron resonance at room temperature

We report the nanoscale spin detection and electron paramagnetic resonance (EPR) spectrum of copper (Cu$^{2+}$) ions via double electron-electron resonance with single spins in diamond at room temperature and low magnetic fields. We measure unexpectedly narrow EPR resonances with linewidths $\sim 2-3$ MHz from copper-chloride molecules dissolved in poly-lysine. We also observe coherent Rabi oscillations and hyperfine splitting from single Cu$^{2+}$ ions, which could be used for dynamic nuclear spin polarization and higher sensitivity of spin detection. We interpret and analyze these observations using both spin hamiltonian modeling of the copper-chloride molecules and numerical simulations of the predicted DEER response, and obtain a sensing volume $\sim (250 \text{nm})^3$. This work will open the door for copper-labeled EPR measurements under ambient conditions in bio-molecules and nano-materials.

quant-ph↗

arXiv2021

Federated Variational Learning for Anomaly Detection in Multivariate Time Series

Anomaly detection has been a challenging task given high-dimensional multivariate time series data generated by networked sensors and actuators in Cyber-Physical Systems (CPS). Besides the highly nonlinear, complex, and dynamic natures of such time series, the lack of labeled data impedes data exploitation in a supervised manner and thus prevents an accurate detection of abnormal phenomenons. On the other hand, the collected data at the edge of the network is often privacy sensitive and large in quantity, which may hinder the centralized training at the main server. To tackle these issues, we propose an unsupervised time series anomaly detection framework in a federated fashion to continuously monitor the behaviors of interconnected devices within a network and alerts for abnormal incidents so that countermeasures can be taken before undesired consequences occur. To be specific, we leave the training data distributed at the edge to learn a shared Variational Autoencoder (VAE) based on Convolutional Gated Recurrent Unit (ConvGRU) model, which jointly captures feature and temporal dependencies in the multivariate time series data for representation learning and downstream anomaly detection tasks. Experiments on three real-world networked sensor datasets illustrate the advantage of our approach over other state-of-the-art models. We also conduct extensive experiments to demonstrate the effectiveness of our detection framework under non-federated and federated settings in terms of overall performance and detection latency.

cs.LG↗

arXiv2021

Learning Effective and Efficient Embedding via an Adaptively-Masked Twins-based Layer

Embedding learning for categorical features is crucial for the deep learning-based recommendation models (DLRMs). Each feature value is mapped to an embedding vector via an embedding learning process. Conventional methods configure a fixed and uniform embedding size to all feature values from the same feature field. However, such a configuration is not only sub-optimal for embedding learning but also memory costly. Existing methods that attempt to resolve these problems, either rule-based or neural architecture search (NAS)-based, need extensive efforts on the human design or network training. They are also not flexible in embedding size selection or in warm-start-based applications. In this paper, we propose a novel and effective embedding size selection scheme. Specifically, we design an Adaptively-Masked Twins-based Layer (AMTL) behind the standard embedding layer. AMTL generates a mask vector to mask the undesired dimensions for each embedding vector. The mask vector brings flexibility in selecting the dimensions and the proposed layer can be easily added to either untrained or trained DLRMs. Extensive experimental evaluations show that the proposed scheme outperforms competitive baselines on all the benchmark tasks, and is also memory-efficient, saving 60\% memory usage without compromising any performance metrics.

cs.LG↗

arXiv2021

SwinIR: Image Restoration Using Swin Transformer

Image restoration is a long-standing low-level vision problem that aims to restore high-quality images from low-quality images (e.g., downscaled, noisy and compressed images). While state-of-the-art image restoration methods are based on convolutional neural networks, few attempts have been made with Transformers which show impressive performance on high-level vision tasks. In this paper, we propose a strong baseline model SwinIR for image restoration based on the Swin Transformer. SwinIR consists of three parts: shallow feature extraction, deep feature extraction and high-quality image reconstruction. In particular, the deep feature extraction module is composed of several residual Swin Transformer blocks (RSTB), each of which has several Swin Transformer layers together with a residual connection. We conduct experiments on three representative tasks: image super-resolution (including classical, lightweight and real-world image super-resolution), image denoising (including grayscale and color image denoising) and JPEG compression artifact reduction. Experimental results demonstrate that SwinIR outperforms state-of-the-art methods on different tasks by $\textbf{up to 0.14$\sim$0.45dB}$, while the total number of parameters can be reduced by $\textbf{up to 67%}$.

eess.IV↗

arXiv2021

Learning to Detect: A Data-driven Approach for Network Intrusion Detection

With massive data being generated daily and the ever-increasing interconnectivity of the world's Internet infrastructures, a machine learning based intrusion detection system (IDS) has become a vital component to protect our economic and national security. In this paper, we perform a comprehensive study on NSL-KDD, a network traffic dataset, by visualizing patterns and employing different learning-based models to detect cyber attacks. Unlike previous shallow learning and deep learning models that use the single learning model approach for intrusion detection, we adopt a hierarchy strategy, in which the intrusion and normal behavior are classified firstly, and then the specific types of attacks are classified. We demonstrate the advantage of the unsupervised representation learning model in binary intrusion detection tasks. Besides, we alleviate the data imbalance problem with SVM-SMOTE oversampling technique in 4-class classification and further demonstrate the effectiveness and the drawback of the oversampling mechanism with a deep neural network as a base model.

cs.CR↗

arXiv2021

SIFN: A Sentiment-aware Interactive Fusion Network for Review-based Item Recommendation

Recent studies in recommender systems have managed to achieve significantly improved performance by leveraging reviews for rating prediction. However, despite being extensively studied, these methods still suffer from some limitations. First, previous studies either encode the document or extract latent sentiment via neural networks, which are difficult to interpret the sentiment of reviewers intuitively. Second, they neglect the personalized interaction of reviews with user/item, i.e., each review has different contributions when modeling the sentiment preference of user/item. To remedy these issues, we propose a Sentiment-aware Interactive Fusion Network (SIFN) for review-based item recommendation. Specifically, we first encode user/item reviews via BERT and propose a light-weighted sentiment learner to extract semantic features of each review. Then, we propose a sentiment prediction task that guides the sentiment learner to extract sentiment-aware features via explicit sentiment labels. Finally, we design a rating prediction task that contains a rating learner with an interactive and fusion module to fuse the identity (i.e., user and item ID) and each review representation so that various interactive features can synergistically influence the final rating score. Experimental results on five real-world datasets demonstrate that the proposed model is superior to state-of-the-art models.

cs.IR↗

arXiv2021

Contributions of Transformer Attention Heads in Multi- and Cross-lingual Tasks

This paper studies the relative importance of attention heads in Transformer-based models to aid their interpretability in cross-lingual and multi-lingual tasks. Prior research has found that only a few attention heads are important in each mono-lingual Natural Language Processing (NLP) task and pruning the remaining heads leads to comparable or improved performance of the model. However, the impact of pruning attention heads is not yet clear in cross-lingual and multi-lingual tasks. Through extensive experiments, we show that (1) pruning a number of attention heads in a multi-lingual Transformer-based model has, in general, positive effects on its performance in cross-lingual and multi-lingual tasks and (2) the attention heads to be pruned can be ranked using gradients and identified with a few trial experiments. Our experiments focus on sequence labeling tasks, with potential applicability on other cross-lingual and multi-lingual tasks. For comprehensiveness, we examine two pre-trained multi-lingual models, namely multi-lingual BERT (mBERT) and XLM-R, on three tasks across 9 languages each. We also discuss the validity of our findings and their extensibility to truly resource-scarce languages and other task settings.

cs.CL↗

arXiv2021

Hierarchical Conditional Flow: A Unified Framework for Image Super-Resolution and Image Rescaling

Normalizing flows have recently demonstrated promising results for low-level vision tasks. For image super-resolution (SR), it learns to predict diverse photo-realistic high-resolution (HR) images from the low-resolution (LR) image rather than learning a deterministic mapping. For image rescaling, it achieves high accuracy by jointly modelling the downscaling and upscaling processes. While existing approaches employ specialized techniques for these two tasks, we set out to unify them in a single formulation. In this paper, we propose the hierarchical conditional flow (HCFlow) as a unified framework for image SR and image rescaling. More specifically, HCFlow learns a bijective mapping between HR and LR image pairs by modelling the distribution of the LR image and the rest high-frequency component simultaneously. In particular, the high-frequency component is conditional on the LR image in a hierarchical manner. To further enhance the performance, other losses such as perceptual loss and GAN loss are combined with the commonly used negative log-likelihood loss in training. Extensive experiments on general image SR, face image SR and image rescaling have demonstrated that the proposed HCFlow achieves state-of-the-art performance in terms of both quantitative metrics and visual quality.

eess.IV↗

arXiv2021

Mutual Affine Network for Spatially Variant Kernel Estimation in Blind Image Super-Resolution

Existing blind image super-resolution (SR) methods mostly assume blur kernels are spatially invariant across the whole image. However, such an assumption is rarely applicable for real images whose blur kernels are usually spatially variant due to factors such as object motion and out-of-focus. Hence, existing blind SR methods would inevitably give rise to poor performance in real applications. To address this issue, this paper proposes a mutual affine network (MANet) for spatially variant kernel estimation. Specifically, MANet has two distinctive features. First, it has a moderate receptive field so as to keep the locality of degradation. Second, it involves a new mutual affine convolution (MAConv) layer that enhances feature expressiveness without increasing receptive field, model size and computation burden. This is made possible through exploiting channel interdependence, which applies each channel split with an affine transformation module whose input are the rest channel splits. Extensive experiments on synthetic and real images show that the proposed MANet not only performs favorably for both spatially variant and invariant kernel estimation, but also leads to state-of-the-art blind SR performance when combined with non-blind SR methods.

cs.CV↗

arXiv2021

Sparse identification of nonlinear dynamics with low-dimensionalized flow representations

We perform a sparse identification of nonlinear dynamics (SINDy) for low-dimensionalized complex flow phenomena. We first apply the SINDy with two regression methods, the thresholded least square algorithm (TLSA) and the adaptive Lasso (Alasso) which show reasonable ability with a wide range of sparsity constant in our preliminary tests, to a two-dimensional single cylinder wake at $Re_D=100$, its transient process, and a wake of two-parallel cylinders, as examples of high-dimensional fluid data. To handle these high dimensional data with SINDy whose library matrix is suitable for low-dimensional variable combinations, a convolutional neural network-based autoencoder (CNN-AE) is utilized. The CNN-AE is employed to map a high-dimensional dynamics into a low-dimensional latent space. The SINDy then seeks a governing equation of the mapped low-dimensional latent vector. Temporal evolution of high-dimensional dynamics can be provided by combining the predicted latent vector by SINDy with the CNN decoder which can remap the low-dimensional latent vector to the original dimension. The SINDy can provide a stable solution as the governing equation of the latent dynamics and the CNN-SINDy based modeling can reproduce high-dimensional flow fields successfully, although more terms are required to represent the transient flow and the two-parallel cylinder wake than the periodic shedding. A nine-equation turbulent shear flow model is finally considered to examine the applicability of SINDy to turbulence, although without using CNN-AE. The present results suggest that the proposed scheme with an appropriate parameter choice enables us to analyze high-dimensional nonlinear dynamics with interpretable low-dimensional manifolds.

physics.flu-dyn↗