arXiv subjects

Ke Xu

Publications and source records attributed to Ke Xu.

At least 325 records · Page 18Linked to original sources

arXiv2022

GPUMD: A package for constructing accurate machine-learned potentials and performing highly efficient atomistic simulations

We present our latest advancements of machine-learned potentials (MLPs) based on the neuroevolution potential (NEP) framework introduced in [Fan et al., Phys. Rev. B 104, 104309 (2021)] and their implementation in the open-source package GPUMD. We increase the accuracy of NEP models both by improving the radial functions in the atomic-environment descriptor using a linear combination of Chebyshev basis functions and by extending the angular descriptor with some four-body and five-body contributions as in the atomic cluster expansion approach. We also detail our efficient implementation of the NEP approach in graphics processing units as well as our workflow for the construction of NEP models, and we demonstrate their application in large-scale atomistic simulations. By comparing to state-of-the-art MLPs, we show that the NEP approach not only achieves above-average accuracy but also is far more computationally efficient. These results demonstrate that the GPUMD package is a promising tool for solving challenging problems requiring highly accurate, large-scale atomistic simulations. To enable the construction of MLPs using a minimal training set, we propose an active-learning scheme based on the latent space of a pre-trained NEP model. Finally, we introduce three separate Python packages, GPYUMD, CALORINE, and PYNEP, which enable the integration of GPUMD into Python workflows.

physics.comp-ph↗

arXiv2022

Structural Entropy Guided Graph Hierarchical Pooling

Following the success of convolution on non-Euclidean space, the corresponding pooling approaches have also been validated on various tasks regarding graphs. However, because of the fixed compression quota and stepwise pooling design, these hierarchical pooling methods still suffer from local structure damage and suboptimal problem. In this work, inspired by structural entropy, we propose a hierarchical pooling approach, SEP, to tackle the two issues. Specifically, without assigning the layer-specific compression quota, a global optimization algorithm is designed to generate the cluster assignment matrices for pooling at once. Then, we present an illustration of the local structure damage from previous methods in the reconstruction of ring and grid synthetic graphs. In addition to SEP, we further design two classification models, SEP-G and SEP-N for graph classification and node classification, respectively. The results show that SEP outperforms state-of-the-art graph pooling methods on graph classification benchmarks and obtains superior performance on node classifications.

cs.LG↗

arXiv2022

Variable thermal transport in black, blue, and violet phosphorene from extensive atomistic simulations with a neuroevolution potential

Phosphorus has diverse chemical bonds and even in its two-dimensional form there are three stable allotropes: black phosphorene (Black-P), blue phosphorene (Blue-P), and violet phosphorene (Violet-P). Due to the complexity of these structures, no efficient and accurate classical interatomic potential has been developed for them. In this paper, we develop an efficient machine-learned neuroevolution potential model for these allotropes and apply it to study thermal transport in them via extensive molecular dynamics (MD) simulations. Based on the homogeneous nonequilibrium MD method, the thermal conductivities are predicted to be $12.5 \pm 0.2$ (Black-P in armchair direction), $78.4 \pm 0.4$ (Black-P in zigzag direction), $128 \pm 3$ (Blue-P), and $2.36 \pm 0.05$ (Violet-P) $\mathrm{Wm^{-1}K^{-1}}$. The underlying reasons for the significantly different thermal conductivity values in these allotropes are unraveled through spectral decomposition, phonon eigenmodes, and phonon participation ratio. Under external tensile strain, the thermal conductivity in black-P and violet-P are finite, while that in blue-P appears unbounded due to the linearization of the flexural phonon dispersion that increases the phonon mean free paths in the zero-frequency limit.

physics.comp-ph↗

arXiv2022

A Simple yet Effective Method for Graph Classification

In deep neural networks, better results can often be obtained by increasing the complexity of previously developed basic models. However, it is unclear whether there is a way to boost performance by decreasing the complexity of such models. Intuitively, given a problem, a simpler data structure comes with a simpler algorithm. Here, we investigate the feasibility of improving graph classification performance while simplifying the learning process. Inspired by structural entropy on graphs, we transform the data sample from graphs to coding trees, which is a simpler but essential structure for graph data. Furthermore, we propose a novel message passing scheme, termed hierarchical reporting, in which features are transferred from leaf nodes to root nodes by following the hierarchical structure of coding trees. We then present a tree kernel and a convolutional network to implement our scheme for graph classification. With the designed message passing scheme, the tree kernel and convolutional network have a lower runtime complexity of $O(n)$ than Weisfeiler-Lehman subtree kernel and other graph neural networks of at least $O(hm)$. We empirically validate our methods with several graph classification benchmarks and demonstrate that they achieve better performance and lower computational consumption than competing approaches.

cs.LG↗

arXiv2022

Dual-Neighborhood Deep Fusion Network for Point Cloud Analysis

Recently, deep neural networks have made remarkable achievements in 3D point cloud classification. However, existing classification methods are mainly implemented on idealized point clouds and suffer heavy degradation of per-formance on non-idealized scenarios. To handle this prob-lem, a feature representation learning method, named Dual-Neighborhood Deep Fusion Network (DNDFN), is proposed to serve as an improved point cloud encoder for the task of non-idealized point cloud classification. DNDFN utilizes a trainable neighborhood learning method called TN-Learning to capture the global key neighborhood. Then, the global neighborhood is fused with the local neighbor-hood to help the network achieve more powerful reasoning ability. Besides, an Information Transfer Convolution (IT-Conv) is proposed for DNDFN to learn the edge infor-mation between point-pairs and benefits the feature transfer procedure. The transmission of information in IT-Conv is similar to the propagation of information in the graph which makes DNDFN closer to the human reasoning mode. Extensive experiments on existing benchmarks especially non-idealized datasets verify the effectiveness of DNDFN and DNDFN achieves the state of the arts.

cs.CV↗

arXiv2022

Adaptive Task Planning for Large-Scale Robotized Warehouses

Robotized warehouses are deployed to automatically distribute millions of items brought by the massive logistic orders from e-commerce. A key to automated item distribution is to plan paths for robots, also known as task planning, where each task is to deliver racks with items to pickers for processing and then return the rack back. Prior solutions are unfit for large-scale robotized warehouses due to the inflexibility to time-varying item arrivals and the low efficiency for high throughput. In this paper, we propose a new task planning problem called TPRW, which aims to minimize the end-to-end makespan that incorporates the entire item distribution pipeline, known as a fulfilment cycle. Direct extensions from state-of-the-art path finding methods are ineffective to solve the TPRW problem because they fail to adapt to the bottleneck variations of fulfillment cycles. In response, we propose Efficient Adaptive Task Planning, a framework for large-scale robotized warehouses with time-varying item arrivals. It adaptively selects racks to fulfill at each timestamp via reinforcement learning, accounting for the time-varying bottleneck of the fulfillment cycles. Then it finds paths for robots to transport the selected racks. The framework adopts a series of efficient optimizations on both time and memory to handle large-scale item throughput. Evaluations on both synthesized and real data show an improvement of $37.1\%$ in effectiveness and $75.5\%$ in efficiency over the state-of-the-arts.

cs.RO↗

arXiv2022

Night-time Scene Parsing with a Large Real Dataset

Although huge progress has been made on scene analysis in recent years, most existing works assume the input images to be in day-time with good lighting conditions. In this work, we aim to address the night-time scene parsing (NTSP) problem, which has two main challenges: 1) labeled night-time data are scarce, and 2) over- and under-exposures may co-occur in the input night-time images and are not explicitly modeled in existing pipelines. To tackle the scarcity of night-time data, we collect a novel labeled dataset, named {\it NightCity}, of 4,297 real night-time images with ground truth pixel-level semantic annotations. To our knowledge, NightCity is the largest dataset for NTSP. In addition, we also propose an exposure-aware framework to address the NTSP problem through augmenting the segmentation process with explicitly learned exposure features. Extensive experiments show that training on NightCity can significantly improve NTSP performances and that our exposure-aware model outperforms the state-of-the-art methods, yielding top performances on our dataset as well as existing datasets.

cs.CV↗

arXiv2022

A Phase-field model for simulating hydrogen-induced pitting corrosion with solid-solid phase transformation in the metal

Hydrogen-induced pitting corrosion of metallic is a common phenomenon that damages the integrity and durability of the materials. Its numerical simulation is still a challenge due to many complex mechanisms, especially solid-solid phase transformation and mechanical interaction, leading to the anisotropic growth of hydride and inducing some bulges on the metal surface. In our work, we propose a phase-field model and numerical technique for simulation of hydrogen-induced pitting corrosion, and apply it to the system of $α$-Uranium. In our model, the elastic strain energy is introduced to approximate the anisotropic pit morphology induced by the mechanical interaction between metal and hydride. For the numerical technique, the free boundary condition based on the finite element method is adopted to introduce the bulges of the metal surface. By the application of our model and numerical technique, the anisotropic pit morphology with a bulge on the metal surface in agreement with experiments of $α$-Uranium is obtained. Moreover, the compression of $α$-Uranium and the dilation of its hydride are discovered, which develops the deep understanding of hydrogen-induced pitting corrosion. This model is expected to be applied to the health detection of hydrogen-induced pitting corrosion of metal in the industry.

cond-mat.mtrl-sci↗

arXiv2022

Bi-directional Object-context Prioritization Learning for Saliency Ranking

The saliency ranking task is recently proposed to study the visual behavior that humans would typically shift their attention over different objects of a scene based on their degrees of saliency. Existing approaches focus on learning either object-object or object-scene relations. Such a strategy follows the idea of object-based attention in Psychology, but it tends to favor those objects with strong semantics (e.g., humans), resulting in unrealistic saliency ranking. We observe that spatial attention works concurrently with object-based attention in the human visual recognition system. During the recognition process, the human spatial attention mechanism would move, engage, and disengage from region to region (i.e., context to context). This inspires us to model the region-level interactions, in addition to the object-level reasoning, for saliency ranking. To this end, we propose a novel bi-directional method to unify spatial attention and object-based attention for saliency ranking. Our model includes two novel modules: (1) a selective object saliency (SOS) module that models objectbased attention via inferring the semantic representation of the salient object, and (2) an object-context-object relation (OCOR) module that allocates saliency ranks to objects by jointly modeling the object-context and context-object interactions of the salient objects. Extensive experiments show that our approach outperforms existing state-of-theart methods. Our code and pretrained model are available at https://github.com/GrassBro/OCOR.

cs.CV↗

arXiv2022

Exploring Memorization in Adversarial Training

Deep learning models have a propensity for fitting the entire training set even with random labels, which requires memorization of every training sample. In this paper, we explore the memorization effect in adversarial training (AT) for promoting a deeper understanding of model capacity, convergence, generalization, and especially robust overfitting of the adversarially trained models. We first demonstrate that deep networks have sufficient capacity to memorize adversarial examples of training data with completely random labels, but not all AT algorithms can converge under the extreme circumstance. Our study of AT with random labels motivates further analyses on the convergence and generalization of AT. We find that some AT approaches suffer from a gradient instability issue and most recently suggested complexity measures cannot explain robust generalization by considering models trained on random labels. Furthermore, we identify a significant drawback of memorization in AT that it could result in robust overfitting. We then propose a new mitigation algorithm motivated by detailed memorization analyses. Extensive experiments on various datasets validate the effectiveness of the proposed method.

cs.LG↗

arXiv2022

The molecular gas resolved by ALMA in the low-metallicity dwarf merging galaxy Haro 11

The physical mechanisms for starburst or quenching in less massive ($M_* < 10^{10} M_{\odot}$) galaxies are unclear. The merger is one of the inescapable processes referred to as both starburst and quenching in massive galaxies. However, the effects of the merger on star formation in dwarf galaxies and their evolution results are still uncertain. We aim to explore how to trigger and quench star formation in dwarf galaxies by studying the metal-poor gas-rich dwarf mergers based on the multi-band observations at a spatial resolution of $\sim$ 460 pc. We use the archival data of ALMA (band 3, 8) and VLT/MUSE to map CO($J=$1-0), [CI]($^3$P$_1 - ^3$P$_0$), and H$α$ emission in one of the most extreme starburst merging dwarf galaxies, Haro 11. We find the molecular gas is assembled around the central two star-forming regions. The molecular/ionized gas and stellar components show complex kinematics, indicating that the gas is probably at a combined stage of collision of clouds and feedback from star formation. The peak location and distribution of [CI](1-0) strongly resemble the CO(1-0) emission, meaning that it might trace the same molecular gas as CO in such a dwarf merger starburst galaxy. The enhancement of line ratios ($\sim 0.5$) of [CI]/CO around knot C is probably generated by the dissociation of CO molecules by cosmic rays and far-ultraviolet photons. Globally, Haro 11 and its star-forming regions share similar SFEs as the high-$z$ starburst galaxies or the clumps in nearby (U)LIRGs. Given the high SFE, sSFR, small stellar mass, low metallicity, and deficient HI gas, Haro 11 could be an analog of high-$z$ dwarf starburst and the potential progenitor of the nearby less massive elliptical galaxies. The significantly smaller turbulent pressure and viral parameter will probably trigger the intense starbursts. We also predict that it will quench at $M_* < 8.5 \times 10^9 M_{\odot}$.

astro-ph.GA↗

arXiv2022

Asymmetric star formation triggered by gas inflow in a barred lenticular galaxy PGC 34107

Comparing to the inactive and gas-poor normal lenticular galaxies (S0s) in the local universe, we study a barred star-forming S0 galaxy, PGC 34107, which has been observed by the Centro Astronómico Hispano Alemán (CAHA) 3.5-m telescope and the Northern Extended Millimeter Array (NOEMA). The spatially resolved ionized gas and molecular gas traced by $^{12}$CO(1-0), hereafter CO(1-0), show the similar distribution and kinematics to the stellar component with an off-center star-forming region, $\sim$380 pc away from the center. The resolved kinematics of molecular CO(1-0) emission reveals that there is a blueshifted (redshifted) velocity component on the receding (approaching) side of the galaxy along the stellar bar. This might provide a plausible evidence of non-circular motion, such as the bar-induced molecular gas inflow. The velocity of molecular gas inflow decreases with approaching towards the peak of the off-center star formation in the north, which might be associated with the inner Lindblad resonance (ILR). In addition to CO(1-0), we also detect the isotopic line of $^{13}$CO(1-0). Most $\rm Hα$, CO(1-0) and $^{13}$CO(1-0) emissions are concentrated on this northern star-forming region. We find that PGC 34107 follows the local stellar mass-metallicity relation, star-forming main sequence, and the Kennicutt-Schmidt law. The resolved and integrated molecular gas main sequence suggest that there is a higher gas fraction in the galaxy central region, which supports a scenario that the bar-induced gas reservoir provides the raw material, and subsequently triggers the central star formation.

astro-ph.GA↗

arXiv2022

Abnormally High Thermal Conductivity in Fivefold Twinned Diamond Nanowires

Fivefold twins (5FTs), discovered nearly 200 years ago, are a common multiply twinned structure that usually dramatically deteriorate the thermal transport properties of nanomaterials. Here, we report the anomalous thermal conductivity ($κ$) in a novel fivefold twinned diamond nanowires (5FT-DNWs). The $κ$ of 5FT-DNWs is effectively enhanced by the defects of 5FT boundaries, and non-monotonically changes with the cross-sectional area ($\textit{S}$). Above the critical $\textit{S}$ = 7.1 nm$^{2}$, 5FT-DNWs show a constant value of $κ$, whereas below it, there appears a sharp increase in $κ$ with decreasing $\textit{S}$. More importantly, 5FT-DNWs with minimal $\textit{S}$ show a superior $κ$ over the bulk diamond. By confirming the Normal-process-dominated scattering event, it is demonstrated that the phonon hydrodynamic behavior plays a determinative role in abnormally high $κ$ of 5FT-DNWs with small $\textit{S}$. The super-transported phonon hydrodynamic phenomenon unveiled in the twinned diamond nanowires may provide a new route for pursuing highly thermally conductive nanomaterials.

physics.comp-ph↗

arXiv2021

Adaptive Channel Encoding for Point Cloud Analysis

Attention mechanism plays a more and more important role in point cloud analysis and channel attention is one of the hotspots. With so much channel information, it is difficult for neural networks to screen useful channel information. Thus, an adaptive channel encoding mechanism is proposed to capture channel relationships in this paper. It improves the quality of the representation generated by the network by explicitly encoding the interdependence between the channels of its features. Specifically, a channel-wise convolution (Channel-Conv) is proposed to adaptively learn the relationship between coordinates and features, so as to encode the channel. Different from the popular attention weight schemes, the Channel-Conv proposed in this paper realizes adaptability in convolution operation, rather than simply assigning different weights for channels. Extensive experiments on existing benchmarks verify our method achieves the state of the arts.

cs.CV↗

arXiv2021

Gait Identification under Surveillance Environment based on Human Skeleton

As an emerging biological identification technology, vision-based gait identification is an important research content in biometrics. Most existing gait identification methods extract features from gait videos and identify a probe sample by a query in the gallery. However, video data contains redundant information and can be easily influenced by bagging (BG) and clothing (CL). Since human body skeletons convey essential information about human gaits, a skeleton-based gait identification network is proposed in our project. First, extract skeleton sequences from the video and map them into a gait graph. Then a feature extraction network based on Spatio-Temporal Graph Convolutional Network (ST-GCN) is constructed to learn gait representations. Finally, the probe sample is identified by matching with the most similar piece in the gallery. We tested our method on the CASIA-B dataset. The result shows that our approach is highly adaptive and gets the advanced result in BG, CL conditions, and average.

cs.CV↗

arXiv2021

Learning to Detect Instance-level Salient Objects Using Complementary Image Labels

Existing salient instance detection (SID) methods typically learn from pixel-level annotated datasets. In this paper, we present the first weakly-supervised approach to the SID problem. Although weak supervision has been considered in general saliency detection, it is mainly based on using class labels for object localization. However, it is non-trivial to use only class labels to learn instance-aware saliency information, as salient instances with high semantic affinities may not be easily separated by the labels. As the subitizing information provides an instant judgement on the number of salient items, it is naturally related to detecting salient instances and may help separate instances of the same class while grouping different parts of the same instance. Inspired by this observation, we propose to use class and subitizing labels as weak supervision for the SID problem. We propose a novel weakly-supervised network with three branches: a Saliency Detection Branch leveraging class consistency information to locate candidate objects; a Boundary Detection Branch exploiting class discrepancy information to delineate object boundaries; and a Centroid Detection Branch using subitizing information to detect salient instance centroids. This complementary information is then fused to produce a salient instance map. To facilitate the learning process, we further propose a progressive training scheme to reduce label noise and the corresponding noise learned by the model, via reciprocating the model with progressive salient instance prediction and model refreshing. Our extensive evaluations show that the proposed method plays favorably against carefully designed baseline methods adapted from related tasks.

cs.CV↗

arXiv2021

Star-Forming S0 Galaxies in SDSS-IV MaNGA Survey

To investigate star-forming activities in early-type galaxies, we select a sample of 52 star-forming S0 galaxies (SFS0s) from the SDSS-IV MaNGA survey. We find that SFS0s have smaller stellar mass compared to normal S0s in MaNGA. After matching the stellar mass to select the control sample, we find that the mean Sérsic index of SFS0s' bulges (1.76$\pm$0.21) is significantly smaller than that of the control sample (2.57$\pm$0.20), suggesting the existence of a pseudo bulge in SFS0s. After introducing the environmental information, SFS0s show smaller spin parameters in the field than in groups, while the control sample has no obvious difference in different environments, which may suggest different dynamical processes in SFS0s. Furthermore, with derived N/O and O/H abundance ratios, SFS0s in the field show nitrogen enrichment, providing evidence for the accretion of metal-poor gas in the field environment. To study the star formation relation, we show that the slope of the spatially resolved star formation main sequence is nearly 1.0 with MaNGA IFU data, confirming the self-regulation of star formation activities at the kpc scales.

astro-ph.GA↗

arXiv2021

Price graphs: Utilizing the structural information of financial time series for stock prediction

Great research efforts have been devoted to exploiting deep neural networks in stock prediction. While long-range dependencies and chaotic property are still two major issues that lower the performance of state-of-the-art deep learning models in forecasting future price trends. In this study, we propose a novel framework to address both issues. Specifically, in terms of transforming time series into complex networks, we convert market price series into graphs. Then, structural information, referring to associations among temporal points and the node weights, is extracted from the mapped graphs to resolve the problems regarding long-range dependencies and the chaotic property. We take graph embeddings to represent the associations among temporal points as the prediction model inputs. Node weights are used as a priori knowledge to enhance the learning of temporal attention. The effectiveness of our proposed framework is validated using real-world stock data, and our approach obtains the best performance among several state-of-the-art benchmarks. Moreover, in the conducted trading simulations, our framework further obtains the highest cumulative profits. Our results supplement the existing applications of complex network methods in the financial realm and provide insightful implications for investment applications regarding decision support in financial markets.

q-fin.ST↗