arXiv subjects

Bin Luo

Publications and source records attributed to Bin Luo.

At least 181 records · Page 10Linked to original sources

arXiv2022

Beyond Greedy Search: Tracking by Multi-Agent Reinforcement Learning-based Beam Search

To track the target in a video, current visual trackers usually adopt greedy search for target object localization in each frame, that is, the candidate region with the maximum response score will be selected as the tracking result of each frame. However, we found that this may be not an optimal choice, especially when encountering challenging tracking scenarios such as heavy occlusion and fast motion. To address this issue, we propose to maintain multiple tracking trajectories and apply beam search strategy for visual tracking, so that the trajectory with fewer accumulated errors can be identified. Accordingly, this paper introduces a novel multi-agent reinforcement learning based beam search tracking strategy, termed BeamTracking. It is mainly inspired by the image captioning task, which takes an image as input and generates diverse descriptions using beam search algorithm. Accordingly, we formulate the tracking as a sample selection problem fulfilled by multiple parallel decision-making processes, each of which aims at picking out one sample as their tracking result in each frame. Each maintained trajectory is associated with an agent to perform the decision-making and determine what actions should be taken to update related information. When all the frames are processed, we select the trajectory with the maximum accumulated score as the tracking result. Extensive experiments on seven popular tracking benchmark datasets validated the effectiveness of the proposed algorithm.

cs.CV↗

arXiv2022

Few-Shot Learning Meets Transformer: Unified Query-Support Transformers for Few-Shot Classification

Few-shot classification which aims to recognize unseen classes using very limited samples has attracted more and more attention. Usually, it is formulated as a metric learning problem. The core issue of few-shot classification is how to learn (1) consistent representations for images in both support and query sets and (2) effective metric learning for images between support and query sets. In this paper, we show that the two challenges can be well modeled simultaneously via a unified Query-Support TransFormer (QSFormer) model. To be specific,the proposed QSFormer involves global query-support sample Transformer (sampleFormer) branch and local patch Transformer (patchFormer) learning branch. sampleFormer aims to capture the dependence of samples in support and query sets for image representation. It adopts the Encoder, Decoder and Cross-Attention to respectively model the Support, Query (image) representation and Metric learning for few-shot classification task. Also, as a complementary to global learning branch, we adopt a local patch Transformer to extract structural representation for each image sample by capturing the long-range dependence of local image patches. In addition, a novel Cross-scale Interactive Feature Extractor (CIFE) is proposed to extract and fuse multi-scale CNN features as an effective backbone module for the proposed few-shot learning method. All modules are integrated into a unified framework and trained in an end-to-end manner. Extensive experiments on four popular datasets demonstrate the effectiveness and superiority of the proposed QSFormer.

cs.CV↗

arXiv2022

High-dimensional robust approximated M-estimators for mean regression with asymmetric data

Asymmetry along with heteroscedasticity or contamination often occurs with the growth of data dimensionality. In ultra-high dimensional data analysis, such irregular settings are usually overlooked for both theoretical and computational convenience. In this paper, we establish a framework for estimation in high-dimensional regression models using Penalized Robust Approximated quadratic M-estimators (PRAM). This framework allows general settings such as random errors lack of symmetry and homogeneity, or the covariates are not sub-Gaussian. To reduce the possible bias caused by the data's irregularity in mean regression, PRAM adopts a loss function with a flexible robustness parameter growing with the sample size. Theoretically, we first show that, in the ultra-high dimension setting, PRAM estimators have local estimation consistency at the minimax rate enjoyed by the LS-Lasso. Then we show that PRAM with an appropriate non-convex penalty in fact agrees with the local oracle solution, and thus obtain its oracle property. Computationally, we demonstrate the performances of six PRAM estimators using three types of loss functions for approximation (Huber, Tukey's biweight and Cauchy loss) combined with two types of penalty functions (Lasso and MCP). Our simulation studies and real data analysis demonstrate satisfactory finite sample performances of the PRAM estimator under general irregular settings.

math.ST↗

arXiv2022

Definitive upper bound on the negligible contribution of quasars to cosmic reionization

Cosmic (hydrogen) reionization marks one of the major phase transitions of the universe at redshift z >= 6. During this epoch, hydrogen atoms in the intergalactic medium (IGM) were ionized by Lyman continuum (LyC) photons. However, it remains challenging to identify the major sources of the LyC photons responsible for reionization. In particular, individual contributions of quasars (or active galactic nuclei, AGNs) and galaxies are still under debate. Here we construct the far-ultraviolet (far-UV) luminosity function for type 1 quasars at z >= 6 that spans 10 magnitudes (-19 < M_UV < -29), conclusively showing that quasars made a negligible contribution to reionization. We mainly search for quasars in the low-luminosity range of M_UV > -23 mag that is critical to determine quasars' total LyC photon production but has been barely explored previously. We find that the quasar population can only provide less than 7% (95% confidence level) of the total photons needed to keep the universe ionized at z = 6.0 - 6.6. Our result suggests that galaxies, presumably low-luminosity star-forming systems, are the major sources of hydrogen reionization.

astro-ph.GA↗

arXiv2022

Spectral Energy Distributions in Three Deep-Drilling Fields of the Vera C. Rubin Observatory Legacy Survey of Space and Time: Source Classification and Galaxy Properties

W-CDF-S, ELAIS-S1, and XMM-LSS will be three Deep-Drilling Fields (DDFs) of the Vera C. Rubin Observatory Legacy Survey of Space and Time (LSST), but their extensive multi-wavelength data have not been fully utilized as done in the COSMOS field, another LSST DDF. To prepare for future science, we fit source spectral energy distributions (SEDs) from X-ray to far-infrared in these three fields mainly to derive galaxy stellar masses and star-formation rates. We use CIGALE v2022.0, a code that has been regularly developed and evaluated, for the SED fitting. Our catalog includes 0.8 million sources covering $4.9~\mathrm{deg^2}$ in W-CDF-S, 0.8 million sources covering $3.4~\mathrm{deg^2}$ in ELAIS-S1, and 1.2 million sources covering $4.9~\mathrm{deg^2}$ in XMM-LSS. Besides fitting normal galaxies, we also select candidates that may host active galactic nuclei (AGNs) or are experiencing recent star-formation variations and use models specifically designed for these sources to fit their SEDs; this increases the utility of our catalog for various projects in the future. We calibrate our measurements by comparison with those in well-studied smaller regions and briefly discuss the implications of our results. We also perform detailed tests of the completeness and purity of SED-selected AGNs. Our data can be retrieved from a public website.

astro-ph.GA↗

arXiv2022

An X-ray fading, UV brightening QSO at $z\approx6$

Explaining the existence of $\gtrsim10^8\,\mathrm{M_\odot}$ SMBHs at $z>6$ is a persistent challenge to modern astrophysics. Multi-wavelength observations of $z\gtrsim6$ QSOs reveal that, on average, their accretion physics is similar to that of their counterparts at lower redshift. However, QSOs showing properties that deviate from the general behavior can provide useful insights into the physical processes responsible for the rapid growth of SMBHs in the early universe. We present X-ray (XMM-Newton, 100 ks) follow-up observations of a $z\approx6$ QSO, J1641+3755, which was found to be remarkably X-ray bright in a 2018 Chandra dataset. J1641+3755 is not detected in the 2021 XMM-Newton observation, implying that its X-ray flux decreased by a factor $\gtrsim7$ on a notably short timescale (i.e., $\approx115$ rest-frame days), making it the $z>4$ QSO with the largest variability amplitude. We also obtained rest-frame UV spectroscopic and photometric data with textit{LBT}, and compared them with archival datasets. Surprisingly, we found that J1641+3755 became brighter in the rest-frame UV band from 2003 to 2016, while no strong variation occurred from 2016 to 2021. Multiple narrow absorption features are detected in its rest-frame UV spectrum, and several of them can be associated with an intervening system at $z=5.67$. The variability properties of J1641+3755 can be due to intrinsic variations of the accretion rate, a small-scale obscuration event, gravitational lensing due to an intervening object, or an unrelated X-ray transient in a foreground galaxy in 2018. Accounting for all of the $z>6$ QSOs with multiple X-ray observations separated by $>10$ rest-frame days, we found an enhancement of strongly (i.e., by a factor $>3$) X-ray variable objects compared to QSOs at later cosmic times. This finding may be related to the physics of fast accretion in high-redshift QSOs.

astro-ph.HE↗

arXiv2022

SPT-Code: Sequence-to-Sequence Pre-Training for Learning Source Code Representations

Recent years have seen the successful application of large pre-trained models to code representation learning, resulting in substantial improvements on many code-related downstream tasks. But there are issues surrounding their application to SE tasks. First, the majority of the pre-trained models focus on pre-training only the encoder of the Transformer. For generation tasks that are addressed using models with the encoder-decoder architecture, however, there is no reason why the decoder should be left out during pre-training. Second, many existing pre-trained models, including state-of-the-art models such as T5-learning, simply reuse the pre-training tasks designed for natural languages. Moreover, to learn the natural language description of source code needed eventually for code-related tasks such as code summarization, existing pre-training tasks require a bilingual corpus composed of source code and the associated natural language description, which severely limits the amount of data for pre-training. To this end, we propose SPT-Code, a sequence-to-sequence pre-trained model for source code. In order to pre-train SPT-Code in a sequence-to-sequence manner and address the aforementioned weaknesses associated with existing pre-training tasks, we introduce three pre-training tasks that are specifically designed to enable SPT-Code to learn knowledge of source code, the corresponding code structure, as well as a natural language description of the code without relying on any bilingual corpus, and eventually exploit these three sources of information when it is applied to downstream tasks. Experimental results demonstrate that SPT-Code achieves state-of-the-art performance on five code-related downstream tasks after fine-tuning.

cs.SE↗

arXiv2022

Deep Learning Meets Software Engineering: A Survey on Pre-Trained Models of Source Code

Recent years have seen the successful application of deep learning to software engineering (SE). In particular, the development and use of pre-trained models of source code has enabled state-of-the-art results to be achieved on a wide variety of SE tasks. This paper provides an overview of this rapidly advancing field of research and reflects on future research directions.

cs.SE↗

arXiv2022

Vibration-Based Bridge Health Monitoring using Telecommunication Cables

Bridge Health Monitoring (BHM) enables early damage detection of bridges and is thus critical for avoiding more severe damages that might result in major financial and human losses. However, conventional BHM systems require dedicated sensors on bridges, which is costly to install and maintain and hard to scale up. To overcome this challenge, we introduce a new system that uses existing telecommunication cables for Distributed Acoustic Sensing (DAS) to collect bridge dynamic strain responses. In addition, we develop a two-module physics-guided system identification method to extract bridge damage-sensitive information (e.g., natural frequencies and mode shapes) from noisy DAS data by constraining strain and displacement mode shapes by bridge dynamics. This approach does not require installation and maintenance of dedicated sensors on bridges. We evaluate our system with field experiments on a concrete bridge with fiber cable running in a conduit under the deck. Our system successfully identified modal frequencies and reconstructed meter-scale mode shapes.

eess.SY↗

arXiv2022

Unified GCNs: Towards Connecting GCNs with CNNs

Graph Convolutional Networks (GCNs) have been widely demonstrated their powerful ability in graph data representation and learning. Existing graph convolution layers are mainly designed based on graph signal processing and transform aspect which usually suffer from some limitations, such as over-smoothing, over-squashing and non-robustness, etc. As we all know that Convolution Neural Networks (CNNs) have received great success in many computer vision and machine learning. One main aspect is that CNNs leverage many learnable convolution filters (kernels) to obtain rich feature descriptors and thus can have high capacity to encode complex patterns in visual data analysis. Also, CNNs are flexible in designing their network architecture, such as MobileNet, ResNet, Xception, etc. Therefore, it is natural to arise a question: can we design graph convolutional layer as flexibly as that in CNNs? Innovatively, in this paper, we consider connecting GCNs with CNNs deeply from a general perspective of depthwise separable convolution operation. Specifically, we show that GCN and GAT indeed perform some specific depthwise separable convolution operations. This novel interpretation enables us to better understand the connections between GCNs (GCN, GAT) and CNNs and further inspires us to design more Unified GCNs (UGCNs). As two showcases, we implement two UGCNs, i.e., Separable UGCN (S-UGCN) and General UGCN (G-UGCN) for graph data representation and learning. Promising experiments on several graph representation benchmarks demonstrate the effectiveness and advantages of the proposed UGCNs.

cs.CV↗

arXiv2022

A quasar shedding its dust cocoon at redshift 2

We present the first near-IR spectroscopy and joint analyses of multi-wavelength observations for SDSS J082747.14+425241.1, a dust-reddened, weak broad emission-line quasar (WLQ) undergoing a remarkable broad absorption line (BAL) transformation. The systemic redshift is more precisely measured to be $z=2.070\pm0.001$ using H$β$ compared to $z=2.040\pm0.003$ using \mgii\ from the literature, signifying an extreme \mgii\ blueshift of $2140\pm530$ \kms\ relative to H$β$. Using the H$β$-based single-epoch scaling relation with a systematic uncertainty of 0.3 dex, its black hole (BH) mass and Eddington ratio are estimated to be $M_{\rm BH}\sim6.1\times10^8M_\odot$ and $λ_{\rm Edd}\sim0.71$, indicative of being in a rapidly accreting phase. Our investigations confirm the WLQ nature and the LoBAL$\rightarrow$HiBAL transformation, along with a factor of 2 increase in the \mgii+\feii\ emission strength and a decrease of 0.1 in $E(B-V)$ over two decades. The kinetic power of this LoBAL wind at $R\sim$15 pc from its BH is estimated to be $\sim$43\% of the Eddington luminosity, sufficient for quasar feedback upon its host galaxy albeit with an order-of-magnitude uncertainty. This quasar provides a clear example of the long-sought scenario where LoBAL quasars are surrounded by dust cocoons, and wide-angle nuclear winds play a key role in the transition for red quasars evolving into the commonly seen blue quasars.

astro-ph.GA↗

arXiv2022

Detection and characterization of microseismic events from fiber-optic DAS data using deep learning

Microseismic analysis is a valuable tool for fracture characterization in the earth's subsurface. As distributed acoustic sensing (DAS) fibers are deployed at depth inside wells, they hold vast potential for high-resolution microseismic analysis. However, the accurate detection of microseismic signals in continuous DAS data is challenging and time-consuming. We design, train, and deploy a deep learning model to detect microseismic events in DAS data automatically. We create a curated dataset of nearly 7,000 manually-selected events and an equal number of background noise examples. We optimize the deep learning model's network architecture together with its training hyperparameters by Bayesian optimization. The trained model achieves an accuracy of 98.6% on our benchmark dataset and even detects low-amplitude events missed during manual labeling. Our methodology detects more than 100,000 events allowing the reconstruction of spatio-temporal fracture development far more accurately and efficiently than would have been feasible by traditional methods.

physics.geo-ph↗

arXiv2022

Connecting Low- and High-Redshift Weak Emission-Line Quasars via HST Spectroscopy of Ly$α$ Emission

We present ultraviolet spectroscopy covering the Ly$α$ + N V complex of six candidate low-redshift ($0.9 < z < 1.5$) weak emission-line quasars (WLQs) based on observations with the Hubble Space Telescope. The original systematic searches for these puzzling Type 1 quasars with intrinsically weak broad emission lines revealed an $N \approx 100$ WLQ population from optical spectroscopy of high-redshift ($z > 3$) quasars, defined by a Ly$α$ + N V rest-frame equivalent width (EW) threshold $< 15.4$ Å. Identification of lower-redshift ($z < 3$) WLQ candidates, however, has relied primarily on optical spectroscopy of weak broad emission lines at longer rest-frame wavelengths. With these new observations expanding existing optical coverage into the ultraviolet, we explore unifying the low- and high-$z$ WLQ populations via EW[Ly$α$+NV]. Two objects in the sample unify with high-$z$ WLQs, three others appear consistent with the intermediate portion of the population connecting WLQs and normal quasars, and the final object is consistent with typical quasars. The expanded wavelength coverage improves the number of available line diagnostics for our individual targets, allowing a better understanding of the shapes of their ionizing continua. The ratio of EW[Ly$α$+NV] to EW[MgII] in our sample is generally small but varied, favoring a soft ionizing continuum scenario for WLQs, and we find a lack of correlation between EW[Ly$α$+NV] and the X-ray properties of our targets, consistent with a "slim-disk" shielding gas model. We also find indications that weak absorption may be a more significant contaminant in low-$z$ WLQ populations than previously thought.

astro-ph.GA↗

arXiv2022

Tiny Object Tracking: A Large-scale Dataset and A Baseline

Tiny objects, frequently appearing in practical applications, have weak appearance and features, and receive increasing interests in meany vision tasks, such as object detection and segmentation. To promote the research and development of tiny object tracking, we create a large-scale video dataset, which contains 434 sequences with a total of more than 217K frames. Each frame is carefully annotated with a high-quality bounding box. In data creation, we take 12 challenge attributes into account to cover a broad range of viewpoints and scene complexities, and annotate these attributes for facilitating the attribute-based performance analysis. To provide a strong baseline in tiny object tracking, we propose a novel Multilevel Knowledge Distillation Network (MKDNet), which pursues three-level knowledge distillations in a unified framework to effectively enhance the feature representation, discrimination and localization abilities in tracking tiny objects. Extensive experiments are performed on the proposed dataset, and the results prove the superiority and effectiveness of MKDNet compared with state-of-the-art methods. The dataset, the algorithm code, and the evaluation code are available at https://github.com/mmic-lcl/Datasets-and-benchmark-code.

cs.CV↗

arXiv2022

On the Robustness of "Robust reversible data hiding scheme based on two-layer embedding strategy"

In the paper "Robust reversible data hiding scheme based on two-layer embedding strategy" published in INS recently, Kumar et al. proposed a robust reversible data hiding (RRDH) scheme based on two-layer embedding. Secret data was embedded into the most significant bit (MSB) planes to increase robustness, and a sorting strategy based on local complexity was adopted to reduce distortion. However, Kumar et al.'s reversible data hiding (RDH) scheme is not as robust against joint photographic experts group (JPEG) compression as stated and can not be called RRDH. This comment first gives a brief description of their RDH scheme, then analyses their scheme's robustness from the perspective of JPEG compression principles. JPEG compression will change pixel values, thereby destroying auxiliary information and pixel value ordering required to extract secret data correctly, making their scheme not robust. Next, the changes in both bit plane and pixel value ordering after JPEG compression are shown and analysed by different robustness-testing experiments. Finally, some suggestions are given to improve the robustness.

cs.MM↗

arXiv2021

Dependency Learning for Legal Judgment Prediction with a Unified Text-to-Text Transformer

Given the fact of a case, Legal Judgment Prediction (LJP) involves a series of sub-tasks such as predicting violated law articles, charges and term of penalty. We propose leveraging a unified text-to-text Transformer for LJP, where the dependencies among sub-tasks can be naturally established within the auto-regressive decoder. Compared with previous works, it has three advantages: (1) it fits in the pretraining pattern of masked language models, and thereby can benefit from the semantic prompts of each sub-task rather than treating them as atomic labels, (2) it utilizes a single unified architecture, enabling full parameter sharing across all sub-tasks, and (3) it can incorporate both classification and generative sub-tasks. We show that this unified transformer, albeit pretrained on general-domain text, outperforms pretrained models tailored specifically for the legal domain. Through an extensive set of experiments, we find that the best order to capture dependencies is different from human intuitions, and the most reasonable logical order for humans can be sub-optimal for the model. We further include two more auxiliary tasks: court view generation and article content prediction, showing they can not only improve the prediction accuracy, but also provide interpretable explanations for model outputs even when an error is made. With the best configuration, our model outperforms both previous SOTA and a single-tasked version of the unified transformer by a large margin.

cs.CL↗

arXiv2021

The Stellar Age Dependence of X-ray Emission from Normal Star-Forming Galaxies in the GOODS Fields

The Chandra Deep Field-South and North surveys (CDFs) provide unique windows into the cosmic history of X-ray emission from normal (non-active) galaxies. Scaling relations of normal galaxy X-ray luminosity (L_X) with star formation rate (SFR) and stellar mass (M_star) have been used to show that the formation rates of low-mass and high-mass X-ray binaries (LMXBs and HMXBs, respectively) evolve with redshift across z = 0-2 following L_HMXB/SFR ~ 1 + z and L_LMXB/M_star ~ (1 + z)^{2-3}. However, these measurements alone do not directly reveal the physical mechanisms behind the redshift evolution of X-ray binaries (XRBs). We derive star-formation histories for a sample of 344 normal galaxies in the CDFs, using spectral energy distribution (SED) fitting of FUV-to-FIR photometric data, and construct a self-consistent, age-dependent model of the X-ray emission from the galaxies. Our model quantifies how X-ray emission from hot gas and XRB populations vary as functions of host stellar-population age. We find that (1) the ratio L_X/M_star declines by a factor of ~1000 from 0-10 Gyr and (2) the X-ray SED becomes harder with increasing age, consistent with a scenario in which the hot gas contribution to the X-ray SED declines quickly for ages above 10 Myr. When dividing our sample into subsets based on metallicity, we find some indication that L_X/M_star is elevated for low-metallicity galaxies, consistent with recent studies of X-ray scaling relations. However, additional statistical constraints are required to quantify both the age and metallicity dependence of X-ray emission from star-forming galaxies.

astro-ph.HE↗

arXiv2021

AST-Transformer: Encoding Abstract Syntax Trees Efficiently for Code Summarization

Code summarization aims to generate brief natural language descriptions for source code. As source code is highly structured and follows strict programming language grammars, its Abstract Syntax Tree (AST) is often leveraged to inform the encoder about the structural information. However, ASTs are usually much longer than the source code. Current approaches ignore the size limit and simply feed the whole linearized AST into the encoder. To address this problem, we propose AST-Transformer to efficiently encode tree-structured ASTs. Experiments show that AST-Transformer outperforms the state-of-arts by a substantial margin while being able to reduce $90\sim95\%$ of the computational complexity in the encoding process.

cs.CL↗