arXiv subjects

Chang Liu

Publications and source records attributed to Chang Liu.

At least 793 records · Page 44Linked to original sources

arXiv2022

THMA: Tencent HD Map AI System for Creating HD Map Annotations

Nowadays, autonomous vehicle technology is becoming more and more mature. Critical to progress and safety, high-definition (HD) maps, a type of centimeter-level map collected using a laser sensor, provide accurate descriptions of the surrounding environment. The key challenge of HD map production is efficient, high-quality collection and annotation of large-volume datasets. Due to the demand for high quality, HD map production requires significant manual human effort to create annotations, a very time-consuming and costly process for the map industry. In order to reduce manual annotation burdens, many artificial intelligence (AI) algorithms have been developed to pre-label the HD maps. However, there still exists a large gap between AI algorithms and the traditional manual HD map production pipelines in accuracy and robustness. Furthermore, it is also very resource-costly to build large-scale annotated datasets and advanced machine learning algorithms for AI-based HD map automatic labeling systems. In this paper, we introduce the Tencent HD Map AI (THMA) system, an innovative end-to-end, AI-based, active learning HD map labeling system capable of producing and labeling HD maps with a scale of hundreds of thousands of kilometers. In THMA, we train AI models directly from massive HD map datasets via supervised, self-supervised, and weakly supervised learning to achieve high accuracy and efficiency required by downstream users. THMA has been deployed by the Tencent Map team to provide services to downstream companies and users, serving over 1,000 labeling workers and producing more than 30,000 kilometers of HD map data per day at most. More than 90 percent of the HD map data in Tencent Map is labeled automatically by THMA, accelerating the traditional HD map labeling process by more than ten times.

cs.CV↗

arXiv2022

An Efficient Dynamic Multi-Sources To Single-Destination (DMS-SD) Algorithm In Smart City Navigation Using Adjacent Matrix

Dijkstra's algorithm is one of the most popular classic path planning algorithms, achieving optimal solutions across a wide range of challenging tasks. However, it only calculates the shortest distance from one vertex to another, which is hard to directly apply to the Dynamic Multi-Sources to Single-Destination (DMS-SD) problem. This paper proposes a modified Dijkstra algorithm to address the DMS-SD problem, where the destination can be dynamically changed. Our method deploys the concept of Adjacent Matrix from Floyd's algorithm and achieves the goal with mathematical calculations. We formally show that all-pairs shortest distance information in Floyd's algorithm is not required in our algorithm. Extensive experiments verify the scalability and optimality of the proposed method.

cs.DS↗

arXiv2022

Position-Aware Subgraph Neural Networks with Data-Efficient Learning

Data-efficient learning on graphs (GEL) is essential in real-world applications. Existing GEL methods focus on learning useful representations for nodes, edges, or entire graphs with ``small'' labeled data. But the problem of data-efficient learning for subgraph prediction has not been explored. The challenges of this problem lie in the following aspects: 1) It is crucial for subgraphs to learn positional features to acquire structural information in the base graph in which they exist. Although the existing subgraph neural network method is capable of learning disentangled position encodings, the overall computational complexity is very high. 2) Prevailing graph augmentation methods for GEL, including rule-based, sample-based, adaptive, and automated methods, are not suitable for augmenting subgraphs because a subgraph contains fewer nodes but richer information such as position, neighbor, and structure. Subgraph augmentation is more susceptible to undesirable perturbations. 3) Only a small number of nodes in the base graph are contained in subgraphs, which leads to a potential ``bias'' problem that the subgraph representation learning is dominated by these ``hot'' nodes. By contrast, the remaining nodes fail to be fully learned, which reduces the generalization ability of subgraph representation learning. In this paper, we aim to address the challenges above and propose a Position-Aware Data-Efficient Learning framework for subgraph neural networks called PADEL. Specifically, we propose a novel node position encoding method that is anchor-free, and design a new generative subgraph augmentation method based on a diffused variational subgraph autoencoder, and we propose exploratory and exploitable views for subgraph contrastive learning. Extensive experiment results on three real-world datasets show the superiority of our proposed method over state-of-the-art baselines.

cs.LG↗

arXiv2022

The minimum spectral radius of graphs with a given domination number

Let $\mathbb{G}_{n,γ}$ be the set of simple and connected graphs on $n$ vertices and with domination number $γ$. The graph with minimum spectral radius among $\mathbb{G}_{n,γ}$ is called the minimizer graph. In this paper, we first prove that the minimizer graph of $\mathbb{G}_{n,γ}$ must be a tree. Moreover, for $γ\in\{1,2,3,\lceil\frac{n}{3}\rceil,\lfloor\frac{n}{2}\rfloor\}$, we characterize all minimizer graphs in $\mathbb{G}_{n,γ}$.

math.CO↗

arXiv2022

SsciBERT: A Pre-trained Language Model for Social Science Texts

The academic literature of social sciences records human civilization and studies human social problems. With its large-scale growth, the ways to quickly find existing research on relevant issues have become an urgent demand for researchers. Previous studies, such as SciBERT, have shown that pre-training using domain-specific texts can improve the performance of natural language processing tasks. However, the pre-trained language model for social sciences is not available so far. In light of this, the present research proposes a pre-trained model based on the abstracts published in the Social Science Citation Index (SSCI) journals. The models, which are available on GitHub (https://github.com/S-T-Full-Text-Knowledge-Mining/SSCI-BERT), show excellent performance on discipline classification, abstract structure-function recognition, and named entity recognition tasks with the social sciences literature.

cs.CL↗

arXiv2022

Scalable Predictive Beamforming for IRS-Assisted Multi-User Communications: A Deep Learning Approach

Beamforming design for intelligent reflecting surface (IRS)-assisted multi-user communication (IRS-MUC) systems critically depends on the acquisition of accurate channel state information (CSI). However, channel estimation (CE) in IRS-MUC systems causes a large signaling overhead for training due to the large number of IRS elements. In this paper, taking into account user mobility, we adopt a deep learning (DL) approach to implicitly learn the historical line-of-sight (LoS) channel features and predict the IRS phase shifts to be adopted for the next time slot for maximization of the weighted sum-rate (WSR) of the IRS-MUC system. With the proposed predictive approach, we can avoid full-scale CSI estimation and facilitate low-dimensional CE for transmit beamforming design such that the signaling overhead is reduced by a scale of $\frac{1}{N}$, where $N$ is the number of IRS elements. To this end, we first develop a universal DL-based predictive beamforming (DLPB) framework featuring a two-stage predictive-instantaneous beamforming mechanism. As a realization of the developed framework, a location-aware convolutional long short-term memory (CLSTM) graph neural network (GNN) is developed to facilitate effective predictive beamforming at the IRS, where a CLSTM module is first adopted to exploit the spatial and temporal features of the considered channels and a GNN is then applied to empower the designed neural network with high scalability and generalizability. Furthermore, in the second stage, based on the predicted IRS phase shifts, an instantaneous CSI-aware fully-connected neural network is designed to optimize the transmit beamforming at the access point. Simulation results demonstrate that the proposed framework not only achieves a better WSR performance and requires a lower CE overhead compared with state-of-the-art benchmarks, but also is highly scalable in the numbers of users.

eess.SP↗

arXiv2022

Kilonova and Optical Afterglow from Binary Neutron Star Mergers. II. Optimal Search Strategy for Serendipitous Observations and Target-of-opportunity Observations of Gravitational-wave Triggers

In the second work of this series, we explore the optimal search strategy for serendipitous and gravitational-wave-triggered target-of-opportunity (ToO) observations of kilonovae and optical short-duration gamma-ray burst (sGRB) afterglows from binary neutron star (BNS) mergers, assuming that cosmological kilonovae are AT2017gfo-like (but with viewing-angle dependence) and that the properties of afterglows are consistent with those of cosmological sGRB afterglows. A one-day cadence serendipitous search strategy with an exposure time of $\sim30\,$s can always achieve an optimal search strategy of kilonovae and afterglows for various survey projects. We show that the optimal detection rates of the kilonovae (afterglows) are $\sim0.3/0.6/1/20\,$yr$^{-1}$ ($\sim50/60/100/800\,$yr$^{-1}$) for ZTF/Mephisto/WFST/LSST, respectively A better search strategy for SiTian than the current design is to increase the exposure time. In principle, a fully built SiTian can detect $\sim7({2000})\,$yr$^{-1}$ kilonovae (afterglows). Population properties of electromagnetic (EM) signals detected via the serendipitous observations are studied in detail. For ToO observations, we predict that one can detect $\sim11\,{\rm{yr}}^{-1}$ BNS gravitational wave (GW) events during the fourth observing run (O4) by considering an exact duty cycle of the third observing run. The median GW sky localization area is expected to be $\sim10\,{\rm{deg}}^2$ for detectable BNS GW events. In O4, we predict that ZTF/Mephisto/WFST/LSST can detect $\sim5/4/3/3$ kilonovae ($\sim1/1/1/1$ afterglows) per year, respectively. The GW detection rates, GW population properties, GW sky localizations, and optimistic ToO detection rates of detectable EM counterparts for BNS GW events at the Advanced Plus, LIGO Voyager and ET\&CE eras are detailedly simulated in this paper.

astro-ph.HE↗

arXiv2022

Fuzzy Positive Learning for Semi-supervised Semantic Segmentation

Semi-supervised learning (SSL) essentially pursues class boundary exploration with less dependence on human annotations. Although typical attempts focus on ameliorating the inevitable error-prone pseudo-labeling, we think differently and resort to exhausting informative semantics from multiple probably correct candidate labels. In this paper, we introduce Fuzzy Positive Learning (FPL) for accurate SSL semantic segmentation in a plug-and-play fashion, targeting adaptively encouraging fuzzy positive predictions and suppressing highly-probable negatives. Being conceptually simple yet practically effective, FPL can remarkably alleviate interference from wrong pseudo labels and progressively achieve clear pixel-level semantic discrimination. Concretely, our FPL approach consists of two main components, including fuzzy positive assignment (FPA) to provide an adaptive number of labels for each pixel and fuzzy positive regularization (FPR) to restrict the predictions of fuzzy positive categories to be larger than the rest under different perturbations. Theoretical analysis and extensive experiments on Cityscapes and VOC 2012 with consistent performance gain justify the superiority of our approach.

cs.CV↗

arXiv2022

EDEN: A Plug-in Equivariant Distance Encoding to Beyond the 1-WL Test

The message-passing scheme is the core of graph representation learning. While most existing message-passing graph neural networks (MPNNs) are permutation-invariant in graph-level representation learning and permutation-equivariant in node- and edge-level representation learning, their expressive power is commonly limited by the 1-Weisfeiler-Lehman (1-WL) graph isomorphism test. Recently proposed expressive graph neural networks (GNNs) with specially designed complex message-passing mechanisms are not practical. To bridge the gap, we propose a plug-in Equivariant Distance ENcoding (EDEN) for MPNNs. EDEN is derived from a series of interpretable transformations on the graph's distance matrix. We theoretically prove that EDEN is permutation-equivariant for all level graph representation learning, and we empirically illustrate that EDEN's expressive power can reach up to the 3-WL test. Extensive experiments on real-world datasets show that combining EDEN with conventional GNNs surpasses recent advanced GNNs.

cs.LG↗

arXiv2022

Dark-siren Cosmology with Decihertz Gravitational-wave Detectors

Gravitational waves (GWs) originated from mergers of stellar-mass binary black holes (SBBHs) are considered as dark sirens in cosmology since they usually do not have electromagnetic counterparts. In order to study cosmos with these events, we not only need the luminosity distances extracted from GW signals, but also require the redshift information of sources via, say, matching GW sky localization with galaxy catalogs. Based on such a methodology, we explore how well decihertz GW detectors, DO-Optimal and DECIGO, can constrain cosmological parameters. Using Monte-Carlo simulated dark sirens, we find that DO-Optimal can constrain the Hubble parameter to ${σ_{H_0}} / {H_0}\, \lesssim 0.23\%$ when estimating $H_0$ alone, while DECIGO performs better by a factor of 5 with ${σ_{H_0}} / {H_0}\lesssim 0.043\%$. Such a good precision of $H_0$ will shed light on the $H_0$ tension. For multiple-parameter estimation, DECIGO can still reach a level of relative uncertainty smaller than $7\%$. The reason why decihertz detectors perform well is explained by their large numbers of SBBH GW events with good distance and angular resolution.

astro-ph.CO↗

arXiv2022

Local Magnification for Data and Feature Augmentation

In recent years, many data augmentation techniques have been proposed to increase the diversity of input data and reduce the risk of overfitting on deep neural networks. In this work, we propose an easy-to-implement and model-free data augmentation method called Local Magnification (LOMA). Different from other geometric data augmentation methods that perform global transformations on images, LOMA generates additional training data by randomly magnifying a local area of the image. This local magnification results in geometric changes that significantly broaden the range of augmentations while maintaining the recognizability of objects. Moreover, we extend the idea of LOMA and random cropping to the feature space to augment the feature map, which further boosts the classification accuracy considerably. Experiments show that our proposed LOMA, though straightforward, can be combined with standard data augmentation to significantly improve the performance on image classification and object detection. And further combination with our feature augmentation techniques, termed LOMA_IF&FO, can continue to strengthen the model and outperform advanced intensity transformation methods for data augmentation.

cs.CV↗

arXiv2022

Direct visualization of edge state in even-layer MnBi2Te4 at zero magnetic field

Being the first intrinsic antiferromagnetic(AFM) topological insulator(TI), MnBi2Te4 is argued to be a topological axion state in its even-layer form due to the antiparallel magnetization between the top and bottom layers. Here we combine both transport and scanning microwave impedance microscopy(sMIM) to investigate such axion state in atomically thin MnBi2Te4 with even-layer thickness at zero magnetic field. While transport measurements show a zero Hall plateau signaturing the axion state, sMIM uncovers an unexpected edge state raising questions regarding the nature of the 'axion state'. Based on our model calculation, we propose that the edge state of even-layer MnBi2Te4 at zero field is derived from gapped helical edge states of the quantum spin Hall effect with time-reversal-symmetry breaking, when a crossover from a three-dimensional TI MnBi2Te4 to a two-dimensional TI occurs. Our finding thus signifies the richness of topological phases in MnB2Te4 that has yet to be fully explored.

cond-mat.mes-hall↗

arXiv2022

A sequential linear programming (SLP) approach for uncertainty analysis-based data-driven computational mechanics

In this article, an efficient sequential linear programming algorithm (SLP) for uncertainty analysis-based data-driven computational mechanics (UA-DDCM) is presented. By assuming that the uncertain constitutive relationship embedded behind the prescribed data set can be characterized through a convex combination of the local data points, the upper and lower bounds of structural responses pertaining to the given data set, which are more valuable for making decisions in engineering design, can be found by solving a sequential of linear programming problems very efficiently. Numerical examples demonstrate the effectiveness of the proposed approach on sparse data set and its robustness with respect to the existence of noise and outliers in the data set.

math.OC↗

arXiv2022

On the topological surface states of the intrinsic magnetic topological insulator Mn-Bi-Te family

We review recent progress in the electronic structure study of intrinsic magnetic topological insulators (MnBi$_2$Te$_4$)(Bi$_2$Te$_3$)$_n$ ($n=0,1,2,3$) family. Specifically, we focus on the ubiquitously (nearly) gapless behavior of the topological surface state Dirac cone observed by photoemission spectroscopy, even though a large Dirac gap is expected because of surface ferromagnetic order. The dichotomy between experiment and theory concerning this gap behavior is perhaps the most critical and puzzling question in this frontier. We discuss various proposals accounting for the lack of magnetic effect on the topological surface state Dirac cone, which are mainly categorized into two pictures, magnetic reconfiguration, and topological surface state redistribution. Band engineering towards opening a magnetic gap of topological surface states provides great opportunities to realize quantized topological transport and axion electrodynamics at higher temperatures.

cond-mat.mes-hall↗

arXiv2022

Visualizing the ultra-structure of microorganisms using table-top extreme ultraviolet imaging

Table-top extreme ultraviolet (EUV) microscopy offers unique opportunities for label-free investigation of biological samples. Here, we demonstrate ptychographic EUV imaging of two dried, unstained model specimens: germlings of a fungus (Aspergillus nidulans), and bacteria (Escherichia coli) cells at 13.5 nm wavelength. We find that the EUV spectral region, which to date has not received much attention for biological imaging, offers sufficient penetration depths for the identification of intracellular features. By implementing a position-correlated ptychography approach, we demonstrate a millimeter-squared field of view enabled by infrared illumination combined with sub-60 nm spatial resolution achieved with EUV illumination on selected regions of interest. The strong element contrast at 13.5 nm wavelength enables the identification of the nanoscale material composition inside the specimens. Our work will advance and facilitate EUV imaging applications and enable further possibilities in life science.

physics.bio-ph↗

arXiv2022

Distilling Representations from GAN Generator via Squeeze and Span

In recent years, generative adversarial networks (GANs) have been an actively studied topic and shown to successfully produce high-quality realistic images in various domains. The controllable synthesis ability of GAN generators suggests that they maintain informative, disentangled, and explainable image representations, but leveraging and transferring their representations to downstream tasks is largely unexplored. In this paper, we propose to distill knowledge from GAN generators by squeezing and spanning their representations. We squeeze the generator features into representations that are invariant to semantic-preserving transformations through a network before they are distilled into the student network. We span the distilled representation of the synthetic domain to the real domain by also using real training data to remedy the mode collapse of GANs and boost the student network performance in a real domain. Experiments justify the efficacy of our method and reveal its great significance in self-supervised representation learning. Code is available at https://github.com/yangyu12/squeeze-and-span.

cs.CV↗

arXiv2022

Local Manifold Augmentation for Multiview Semantic Consistency

Multiview self-supervised representation learning roots in exploring semantic consistency across data of complex intra-class variation. Such variation is not directly accessible and therefore simulated by data augmentations. However, commonly adopted augmentations are handcrafted and limited to simple geometrical and color changes, which are unable to cover the abundant intra-class variation. In this paper, we propose to extract the underlying data variation from datasets and construct a novel augmentation operator, named local manifold augmentation (LMA). LMA is achieved by training an instance-conditioned generator to fit the distribution on the local manifold of data and sampling multiview data using it. LMA shows the ability to create an infinite number of data views, preserve semantics, and simulate complicated variations in object pose, viewpoint, lighting condition, background etc. Experiments show that with LMA integrated, self-supervised learning methods such as MoCov2 and SimSiam gain consistent improvement on prevalent benchmarks including CIFAR10, CIFAR100, STL10, ImageNet100, and ImageNet. Furthermore, LMA leads to representations that obtain more significant invariance to the viewpoint, object pose, and illumination changes and stronger robustness to various real distribution shifts reflected by ImageNet-V2, ImageNet-R, ImageNet Sketch etc.

cs.CV↗

arXiv2022

Beyond Instance Discrimination: Relation-aware Contrastive Self-supervised Learning

Contrastive self-supervised learning (CSL) based on instance discrimination typically attracts positive samples while repelling negatives to learn representations with pre-defined binary self-supervision. However, vanilla CSL is inadequate in modeling sophisticated instance relations, limiting the learned model to retain fine semantic structure. On the one hand, samples with the same semantic category are inevitably pushed away as negatives. On the other hand, differences among samples cannot be captured. In this paper, we present relation-aware contrastive self-supervised learning (ReCo) to integrate instance relations, i.e., global distribution relation and local interpolation relation, into the CSL framework in a plug-and-play fashion. Specifically, we align similarity distributions calculated between the positive anchor views and the negatives at the global level to exploit diverse similarity relations among instances. Local-level interpolation consistency between the pixel space and the feature space is applied to quantitatively model the feature differences of samples with distinct apparent similarities. Through explicitly instance relation modeling, our ReCo avoids irrationally pushing away semantically identical samples and carves a well-structured feature space. Extensive experiments conducted on commonly used benchmarks justify that our ReCo consistently gains remarkable performance improvements.

cs.CV↗