arXiv subjects

Hui Li

Publications and source records attributed to Hui Li.

At least 613 records · Page 34Linked to original sources

arXiv2021

Improving Code Summarization with Block-wise Abstract Syntax Tree Splitting

Automatic code summarization frees software developers from the heavy burden of manual commenting and benefits software development and maintenance. Abstract Syntax Tree (AST), which depicts the source code's syntactic structure, has been incorporated to guide the generation of code summaries. However, existing AST based methods suffer from the difficulty of training and generate inadequate code summaries. In this paper, we present the Block-wise Abstract Syntax Tree Splitting method (BASTS for short), which fully utilizes the rich tree-form syntax structure in ASTs, for improving code summarization. BASTS splits the code of a method based on the blocks in the dominator tree of the Control Flow Graph, and generates a split AST for each code split. Each split AST is then modeled by a Tree-LSTM using a pre-training strategy to capture local non-linear syntax encoding. The learned syntax encoding is combined with code encoding, and fed into Transformer to generate high-quality code summaries. Comprehensive experiments on benchmarks have demonstrated that BASTS significantly outperforms state-of-the-art approaches in terms of various evaluation metrics. To facilitate reproducibility, our implementation is available at https://github.com/XMUDM/BASTS.

cs.SE↗

arXiv2021

Code Completion by Modeling Flattened Abstract Syntax Trees as Graphs

Code completion has become an essential component of integrated development environments. Contemporary code completion methods rely on the abstract syntax tree (AST) to generate syntactically correct code. However, they cannot fully capture the sequential and repetitive patterns of writing code and the structural information of the AST. To alleviate these problems, we propose a new code completion approach named CCAG, which models the flattened sequence of a partial AST as an AST graph. CCAG uses our proposed AST Graph Attention Block to capture different dependencies in the AST graph for representation learning in code completion. The sub-tasks of code completion are optimized via multi-task learning in CCAG, and the task balance is automatically achieved using uncertainty without the need to tune task weights. The experimental results show that CCAG has superior performance than state-of-the-art approaches and it is able to provide intelligent code completion.

cs.SE↗

arXiv2021

RFN-Nest: An end-to-end residual fusion network for infrared and visible images

In the image fusion field, the design of deep learning-based fusion methods is far from routine. It is invariably fusion-task specific and requires a careful consideration. The most difficult part of the design is to choose an appropriate strategy to generate the fused image for a specific task in hand. Thus, devising learnable fusion strategy is a very challenging problem in the community of image fusion. To address this problem, a novel end-to-end fusion network architecture (RFN-Nest) is developed for infrared and visible image fusion. We propose a residual fusion network (RFN) which is based on a residual architecture to replace the traditional fusion approach. A novel detail-preserving loss function, and a feature enhancing loss function are proposed to train RFN. The fusion model learning is accomplished by a novel two-stage training strategy. In the first stage, we train an auto-encoder based on an innovative nest connection (Nest) concept. Next, the RFN is trained using the proposed loss functions. The experimental results on public domain data sets show that, compared with the existing methods, our end-to-end fusion network delivers a better performance than the state-of-the-art methods in both subjective and objective evaluation. The code of our fusion method is available at https://github.com/hli1221/imagefusion-rfn-nest

cs.CV↗

arXiv2021

Orbital evolution of binary black holes in active galactic nucleus disks: a disk channel for binary black hole mergers?

We perform a series of high-resolution 2D hydrodynamical simulations of equal-mass binary black holes (BBHs) embedded in active galactic nucleus (AGN) accretion disks to study whether these binaries can be driven to merger by the surrounding gas. We find that the gravitational softening adopted for the BBH has a profound impact on this result. When the softening is less than ten percent of the binary separation, we show that, in agreement with recent simulations of isolated equal-mass binaries, prograde BBHs expand in time rather than contract. Eventually, however, the binary separation becomes large enough that the tidal force of the central AGN disrupts them. Only when the softening is relatively large do we find that prograde BBHs harden. We determine through detailed analysis of the binary torque, that this dichotomy is due to a loss of spiral structure in the circum-single disks orbiting each BH when the softening is a significant fraction of the binary separation. Properly resolving these spirals -- both with high resolution and small softening -- results in a significant source of binary angular momentum. Only for retrograde BBHs do we find consistent hardening, regardless of softening, as these BBHs lack the important spiral structure in their circum-single disks. This suggests that the gas-driven inspiral of retrograde binaries can produce a population of compact BBHs in the gravitational-wave-emitting regime in AGN disks, which may contribute a large fraction to the observed BBH mergers.

astro-ph.HE↗

arXiv2021

Towards Enhancing Database Education: Natural Language Generation Meets Query Execution Plans

The database systems course is offered as part of an undergraduate computer science degree program in many major universities. A key learning goal of learners taking such a course is to understand how SQL queries are processed in a RDBMS in practice. Since a query execution plan (QEP) describes the execution steps of a query, learners can acquire the understanding by perusing the QEPs generated by a RDBMS. Unfortunately, in practice, it is often daunting for a learner to comprehend these QEPs containing vendor-specific implementation details, hindering her learning process. In this paper, we present a novel, end-to-end, generic system called lantern that generates a natural language description of a qep to facilitate understanding of the query execution steps. It takes as input an SQL query and its QEP, and generates a natural language description of the execution strategy deployed by the underlying RDBMS. Specifically, it deploys a declarative framework called pool that enables subject matter experts to efficiently create and maintain natural language descriptions of physical operators used in QEPs. A rule-based framework called RULE-LANTERN is proposed that exploits pool to generate natural language descriptions of QEPs. Despite the high accuracy of RULE-LANTERN, our engagement with learners reveal that, consistent with existing psychology theories, perusing such rule-based descriptions lead to boredom due to repetitive statements across different QEPs. To address this issue, we present a novel deep learning-based language generation framework called NEURAL-LANTERN that infuses language variability in the generated description by exploiting a set of paraphrasing tools and word embedding. Our experimental study with real learners shows the effectiveness of lantern in facilitating comprehension of QEPs.

cs.DB↗

arXiv2021

A Bioinspired Retinal Neural Network for Accurately Extracting Small-Target Motion Information in Cluttered Backgrounds

Robust and accurate detection of small moving targets in cluttered moving backgrounds is a significant and challenging problem for robotic visual systems to perform search and tracking tasks. Inspired by the neural circuitry of elementary motion vision in the mammalian retina, this paper proposes a bioinspired retinal neural network based on a new neurodynamics-based temporal filtering and multiform 2-D spatial Gabor filtering. This model can estimate motion direction accurately via only two perpendicular spatiotemporal filtering signals, and respond to small targets of different sizes and velocities by adjusting the dendrite field size of the spatial filter. Meanwhile, an algorithm of directionally selective inhibition is proposed to suppress the target-like features in the moving background, which can reduce the influence of background motion effectively. Extensive synthetic and real-data experiments show that the proposed model works stably for small targets of a wider size and velocity range, and has better detection performance than other bioinspired models. Additionally, it can also extract the information of motion direction and motion energy accurately and rapidly.

cs.AI↗

arXiv2021

A Bioinspired Approach-Sensitive Neural Network for Collision Detection in Cluttered and Dynamic Backgrounds

Rapid, accurate and robust detection of looming objects in cluttered moving backgrounds is a significant and challenging problem for robotic visual systems to perform collision detection and avoidance tasks. Inspired by the neural circuit of elementary motion vision in the mammalian retina, this paper proposes a bioinspired approach-sensitive neural network (ASNN) that contains three main contributions. Firstly, a direction-selective visual processing module is built based on the spatiotemporal energy framework, which can estimate motion direction accurately via only two mutually perpendicular spatiotemporal filtering channels. Secondly, a novel approach-sensitive neural network is modeled as a push-pull structure formed by ON and OFF pathways, which responds strongly to approaching motion while insensitivity to lateral motion. Finally, a method of directionally selective inhibition is introduced, which is able to suppress the translational backgrounds effectively. Extensive synthetic and real robotic experiments show that the proposed model is able to not only detect collision accurately and robustly in cluttered and dynamic backgrounds but also extract more collision information like position and direction, for guiding rapid decision making.

cs.RO↗

arXiv2021

CelebA-Spoof Challenge 2020 on Face Anti-Spoofing: Methods and Results

As facial interaction systems are prevalently deployed, security and reliability of these systems become a critical issue, with substantial research efforts devoted. Among them, face anti-spoofing emerges as an important area, whose objective is to identify whether a presented face is live or spoof. Recently, a large-scale face anti-spoofing dataset, CelebA-Spoof which comprised of 625,537 pictures of 10,177 subjects has been released. It is the largest face anti-spoofing dataset in terms of the numbers of the data and the subjects. This paper reports methods and results in the CelebA-Spoof Challenge 2020 on Face AntiSpoofing which employs the CelebA-Spoof dataset. The model evaluation is conducted online on the hidden test set. A total of 134 participants registered for the competition, and 19 teams made valid submissions. We will analyze the top ranked solutions and present some discussion on future work directions.

cs.CV↗

arXiv2021

Measuring the Mass of Missing Baryons in the Halo of Andromeda Galaxy with Gamma-Ray Observations

One of the biggest mysteries in the modern cosmology and galaxy formation is the hideout of the "missing baryons". The leading theory of galaxy formation predicts that a huge amount of baryons resides around galaxies extending out to their virial radii in the form of diffuse and hot gas of $10^6-10^7\,$K, which is also known as the major component of the circumgalactic medium (CGM). Studies by various groups via different techniques, however, have not reached a consensus on the role of CGM in accounting for the missing baryons, with the estimated contribution ranging from a minor fraction to enclosing the baryon budget of the galaxy. In this work we attempt to measure the mass of missing baryons in CGM with a novel method based on the gamma-ray observations of the extended halo of the Andromeda Galaxy. Since cosmic-ray particles that are generated inside the galaxy will eventually escape to the CGM, they will produce gamma-ray emission via the proton-proton collision with CGM. Different from some traditional measurements which are sensitive only to gas in certain specific temperature range, the hadronic gamma-ray flux is sensitive to baryonic gases in all phases and does not rely on the metallicity in the halo. Our result suggests that the total baryon mass contained within the virial radius is less than $(1.4-5)\times 10^{10}M_\odot$ according to the gamma-ray intensity obtained with a model-dependent analysis. It implies that the CGM of Andromeda Galaxy may not account for more than $30\%$ of the missing baryons, but the result is subject to uncertainties from the gamma-ray intensity upper limit, diffusion coefficient of the CRs in the halo as well as the stellar mass and dark matter halo mass of the galaxy. This method will become more constraining provided better understandings on these issues and more sensitive gamma-ray telescopes in the future.

astro-ph.HE↗

arXiv2021

Recent progress and challenges in magnetic tunnel junctions with 2D materials for spintronic applications

As Moore's law is gradually losing its effectiveness, developing alternative high-speed and low-energy-consuming information technology with post-silicon advanced materials is urgently needed. The successful application of tunneling magnetoresistance (TMR) in magnetic tunnel junctions (MTJs) has given rise to a tremendous economic impact on magnetic informatics, including MRAM, radio-frequency sensors, microwave generators and neuromorphic computing networks. The emergence of two-dimensional (2D) materials brings opportunities for MTJs based on 2D materials which have many attractive characters and advantages. Especially, the recently discovered intrinsic 2D ferromagnetic materials with high spin-polarization hold the promise for next-generation nanoscale MTJs. With the development of advanced 2D materials, many efforts on MTJs with 2D materials have been made both theoretically and experimentally. Various 2D materials, such as semi-metallic graphene, insulating h-BN, semiconducting MoS2, magnetic semiconducting CrI3, magnetic metallic Fe3GeTe2 and some other recently emerged 2D materials are discussed as the electrodes and/or central scattering materials of MTJs in this review. We discuss the fundamental and main issues facing MTJs, and review the current progress made with 2D MTJs, briefly comment on work with some specific 2D materials, and highlight how they address the current challenges in MTJs, and finally offer an outlook and perspective of 2D MTJs.

cond-mat.mtrl-sci↗

arXiv2021

Ring Formation in Protoplanetary Disks Driven by an Eccentric Instability

We find that, under certain conditions, protoplanetary disks may spontaneously generate multiple, concentric gas rings without an embedded planet through an eccentric cooling instability. Using both linear theory and non-linear hydrodynamics simulations, we show that a variety of background states may trap a slowly precessing, one-armed spiral mode that becomes unstable when a gravitationally-stable disk rapidly cools. The angular momentum required to excite this spiral comes at the expense of non-uniform mass transport that generically results in multiple rings. For example, one long-term hydrodynamics simulation exhibits four long-lived, axisymmetric gas rings. We verify the instability evolution and ring formation mechanism from first principles with our linear theory, which shows remarkable agreement with the simulation results. Dust trapped in these rings may produce observable features consistent with observed disks. Additionally, direct detection of the eccentric gas motions may be possible when the instability saturates, and any residual eccentricity leftover in the rings at later times may also provide direct observational evidence of this mechanism.

astro-ph.EP↗

arXiv2021

Hamiltonian circle actions with almost minimal isolated fixed points

Let the circle act in a Hamiltonian fashion on a connected compact symplectic manifold $(M, ω)$ of dimension $2n$. Then the $S^1$-action has at least $n+1$ fixed points. In a previous paper, we study the case when the fixed point set consists of precisely $n+1$ isolated points. In this paper, we study the case when the fixed point set consists of exactly $n+2$ isolated points. We show that in this case $n$ must be even. We find equivalent conditions on the first Chern class of $M$ and a particular weight of the $S^1$-action. We also show that the particular weight can completely determine the integral cohomology ring and the total Chern class of $M$, and the sets of weights of the $S^1$-action at all the fixed points. We will see that all these data are isomorphic to those of known examples, $\widetilde{G}_2(\mathbb{R}^{n+2})$ with $n\geq 2$ even, equipped with standard circle actions.

math.SG↗

arXiv2021

Self Sparse Generative Adversarial Networks

Generative Adversarial Networks (GANs) are an unsupervised generative model that learns data distribution through adversarial training. However, recent experiments indicated that GANs are difficult to train due to the requirement of optimization in the high dimensional parameter space and the zero gradient problem. In this work, we propose a Self Sparse Generative Adversarial Network (Self-Sparse GAN) that reduces the parameter space and alleviates the zero gradient problem. In the Self-Sparse GAN, we design a Self-Adaptive Sparse Transform Module (SASTM) comprising the sparsity decomposition and feature-map recombination, which can be applied on multi-channel feature maps to obtain sparse feature maps. The key idea of Self-Sparse GAN is to add the SASTM following every deconvolution layer in the generator, which can adaptively reduce the parameter space by utilizing the sparsity in multi-channel feature maps. We theoretically prove that the SASTM can not only reduce the search space of the convolution kernel weight of the generator but also alleviate the zero gradient problem by maintaining meaningful features in the Batch Normalization layer and driving the weight of deconvolution layers away from being negative. The experimental results show that our method achieves the best FID scores for image generation compared with WGAN-GP on MNIST, Fashion-MNIST, CIFAR-10, STL-10, mini-ImageNet, CELEBA-HQ, and LSUN bedrooms, and the relative decrease of FID is 4.76% ~ 21.84%.

cs.CV↗

arXiv2021

Homogeneous superconducting gap in DBCO synthesized by oxide molecular beam epitaxy

Much of what is known about high-temperature cuprate superconductors stems from studies based on two surface analytical tools, angle-resolved photoemission spectroscopy (ARPES) and spectroscopic imaging scanning tunneling microscopy (SI-STM). A question of general interest is whether and when the surface properties probed by ARPES and SI-STM are representative of the intrinsic properties of bulk materials. We find this question is prominent in thin films of a rarely studied cuprate DBCO. We synthesize DBCO films by oxide molecular beam epitaxy and study them by in situ ARPES and SI-STM. Both ARPES and SI-STM show that the surface DBCO layer is different from the bulk of the film. It is heavily underdoped, while the doping level in the bulk is close to optimal doping evidenced by bulk-sensitive mutual inductance measurements. ARPES shows the typical electronic structure of a heavily underdoped CuO2 plane and two sets of one-dimensional bands originating from the CuO chains with one of them gapped. SI-STM reveals two different energy scales in the local density of states, with one corresponding to the superconductivity and the other one to the pseudogap. While the pseudogap shows large variations over the length scale of a few nanometers, the superconducting gap is very homogeneous. This indicates that the pseudogap and superconductivity are of different origins.

cond-mat.supr-con↗

arXiv2021

Understand Volatility of Algorithmic Stablecoin: Modeling, Verification and Empirical Analysis

An algorithmic stablecoin is a type of cryptocurrency managed by algorithms (i.e., smart contracts) to dynamically minimize the volatility of its price relative to a specific form of asset, e.g., US dollar. As algorithmic stablecoins have been growing rapidly in recent years, they become much more volatile than expected. In this paper, we took a deep dive into the core of algorithmic stablecoins and shared our answer to two fundamental research questions, i.e., Are algorithmic stablecoins volatile by design? Are they volatile in practice? Specifically, we introduced an in-depth study on three popular types of algorithmic stablecoins and developed a modeling framework to formalize their key design protocols. Through formal verification, the framework can identify critical conditions under which stablecoins might become volatile. Furthermore, we performed a systematic empirical analysis on real transaction activities of the Basis Cash stablecoin to relate theoretical possibilities to market observations. Lastly, we highlighted key design decisions for future development of algorithmic stablecoins.

cs.CR↗

arXiv2021

Measuring Turbulence with Young Stars in the Orion Complex

Stars form in molecular clouds in the interstellar medium (ISM) with a turbulent kinematic state. Newborn stars therefore should retain the turbulent kinematics of their natal clouds. Gaia DR2 and APOGEE-2 surveys in combination provide three-dimensional (3D) positions and 3D velocities of young stars in the Orion Molecular Cloud Complex. Using the full 6D measurements, we compute the velocity structure functions (VSFs) of the stars in six different groups within the Orion Complex. We find that the motions of stars in all diffuse groups exhibit strong characteristics of turbulence. Their first-order VSFs have a power-law exponent ranging from $\sim0.2-0.5$ on scales of a few to a few tens of pc, generally consistent with Larson's relation. On the other hand, dense star clusters, such as the Orion Nebula Cluster (ONC), have experienced rapid dynamical relaxation, and have lost the memory of the initial turbulent kinematics. The VSFs of several individual groups and the whole Complex all show features supporting local energy injection from supernovae. The measured strength of turbulence depends on the location relative to the supernova epicenters and the formation history of the groups. Our detection of turbulence traced by young stars introduces a new method of probing the turbulent kinematics of the ISM. Unlike previous gas-based studies with only projected measurements accessible to observations, we utilize the full 6D information of stars, presenting a more complete picture of the 3D interstellar turbulence.

astro-ph.GA↗

arXiv2020

To Talk or to Work: Flexible Communication Compression for Energy Efficient Federated Learning over Heterogeneous Mobile Edge Devices

Recent advances in machine learning, wireless communication, and mobile hardware technologies promisingly enable federated learning (FL) over massive mobile edge devices, which opens new horizons for numerous intelligent mobile applications. Despite the potential benefits, FL imposes huge communication and computation burdens on participating devices due to periodical global synchronization and continuous local training, raising great challenges to battery constrained mobile devices. In this work, we target at improving the energy efficiency of FL over mobile edge networks to accommodate heterogeneous participating devices without sacrificing the learning performance. To this end, we develop a convergence-guaranteed FL algorithm enabling flexible communication compression. Guided by the derived convergence bound, we design a compression control scheme to balance the energy consumption of local computing (i.e., "working") and wireless communication (i.e., "talking") from the long-term learning perspective. In particular, the compression parameters are elaborately chosen for FL participants adapting to their computing and communication environments. Extensive simulations are conducted using various datasets to validate our theoretical analysis, and the results also demonstrate the efficacy of the proposed scheme in energy saving.

cs.LG↗

arXiv2020

Adapting Active Reflector Technology for greater sensitivity and sky-coverage in FAST-like Telescopes

The Five-hundred-meter Aperture Spherical radio Telescope (FAST), the largest single dish radio telescope in the world, has implemented an innovative technology for its huge reflector, which changes the shape of the primary reflector from spherical to that of a paraboloid of 300 m aperture. Here we explore how the current FAST sensitivity can potentially be further improved by increasing the illuminated area (i.e., the aperture of the paraboloid embedded in the spherical surface). Alternatively, the maximum zenith angle can be increased to give greater sky coverage by decreasing the illuminated aperture.Different parabolic apertures within the FAST capability are analyzed in terms of how far the spherical surface would have to move to approximate a paraboloid. The sensitivity of FAST can be improved by approximately 10 % if the aperture of the paraboloid is increased from 300 m to 315 m. The parabolic aperture lies within the main spherical surface and does not extend beyond its edge. The maximum zenith angle can be increased to approximately 35 degrees from 26.4 degrees, if we decrease the aperture of the paraboloid to 220 m. This would still give a sensitivity similar to the Arecibo 305 m radio telescope. Radial deviations between paraboloids of different apertures and the spherical surfaces of differing radii are also investigated. Maximum zenith angles corresponding to different apertures of the paraboloid are further derived. A spherical surface with a different radius can provide a reference baseline for shape-changing applied through active reflector technology to FAST-like telescopes.

astro-ph.IM↗