Search arXiv⌕ Search

arXiv subjects

Xin Wang

Publications and source records attributed to Xin Wang.

At least 1,585 records · Page 88Linked to original sources
arXiv2019

Elliptic Blowup Equations for 6d SCFTs. II: Exceptional Cases

The building blocks of 6d $(1,0)$ SCFTs include certain rank one theories with gauge group $G=SU(3),SO(8),F_4,E_{6,7,8}$. In this paper, we propose a universal recursion formula for the elliptic genera of all such theories. This formula is solved from the elliptic blowup equations introduced in our previous paper. We explicitly compute the elliptic genera and refined BPS invariants, which recover all previous results from topological string theory, modular bootstrap, Hilbert series, 2d quiver gauge theories and 4d $\mathcal{N}=2$ superconformal $H_{G}$ theories. We also observe an intriguing relation between the $k$-string elliptic genus and the Schur indices of rank $k$ $H_{G}$ SCFTs, as a generalization of Lockhart-Zotto's conjecture at the rank one cases. In a subsequent paper, we deal with all other non-Higgsable clusters with matters.

hep-th↗
arXiv2019

Homophily on social networks changes evolutionary advantage in competitive information diffusion

Competitive information diffusion on large-scale social networks reveals fundamental characteristics of rumor contagions and has profound influence on public opinion formation. There has been growing interest in exploring dynamical mechanisms of the competing evolutions recently. Nevertheless, the impacts of population homophily, which determines powerful collective human behaviors, remains unclear. In this paper, we incorporate homophily effects into a modified competitive ignorant-spreader-ignorant (SIS) rumor diffusion model with generalized population preference. Using microscopic Markov chain approach, we first derive the phase diagram of competing diffusion results and examine how competitive information spreads and evolves on social networks. We then explore the detailed effects of homophily, which is modeled by a rewiring mechanism. Results show that homophily promotes the formation of divided "echo chambers" and protects the disadvantaged information from extinction, which further changes or even reverses the evolutionary advantage, i.e., the difference of final proportions of the competitive information. We highlight the conclusion that the reversals may happen only when the initially disadvantaged information has stronger transmission ability, owning diffusion advantage over the other one. Our framework provides profound insight into competing dynamics with population homophily, which may pave ways for further controlling misinformation and guiding public belief systems. Moreover, the reversing condition sheds light on designing effective competing strategies in many real scenarios.

physics.soc-ph↗
arXiv2019

On SDN-Enabled Online and Dynamic Bandwidth Allocation for Stream Analytics

Data communication in cloud-based distributed stream data analytics often involves a collection of parallel and pipelined TCP flows. As the standard TCP congestion control mechanism is designed for achieving "fairness" among competing flows and is agnostic to the application layer contexts, the bandwidth allocation among a set of TCP flows traversing bottleneck links often leads to sub-optimal application-layer performance measures, e.g., stream processing throughput or average tuple complete latency. Motivated by this and enabled by the rapid development of the Software-Defined Networking (SDN) techniques, in this paper, we re-investigate the design space of the bandwidth allocation problem and propose a cross-layer framework which utilizes the additional information obtained from the application layer and provides on-the-fly and dynamic bandwidth adjustment algorithms for helping the stream analytics applications achieving better performance during the runtime. We implement a prototype cross-layer bandwidth allocation framework based on a popular open-source distributed stream processing platform, Apache Storm, together with the OpenDaylight controller, and carry out extensive experiments with real-world analytical workloads on top of a local cluster consisting of 10 workstations interconnected by a SDN-enabled switch. The experiment results clearly validate the effectiveness and efficiency of our proposed framework and algorithms.

cs.NI↗
arXiv2019

Local Supports Global: Deep Camera Relocalization with Sequence Enhancement

We propose to leverage the local information in image sequences to support global camera relocalization. In contrast to previous methods that regress global poses from single images, we exploit the spatial-temporal consistency in sequential images to alleviate uncertainty due to visual ambiguities by incorporating a visual odometry (VO) component. Specifically, we introduce two effective steps called content-augmented pose estimation and motion-based refinement. The content-augmentation step focuses on alleviating the uncertainty of pose estimation by augmenting the observation based on the co-visibility in local maps built by the VO stream. Besides, the motion-based refinement is formulated as a pose graph, where the camera poses are further optimized by adopting relative poses provided by the VO component as additional motion constraints. Thus, the global consistency can be guaranteed. Experiments on the public indoor 7-Scenes and outdoor Oxford RobotCar benchmark datasets demonstrate that benefited from local information inherent in the sequence, our approach outperforms state-of-the-art methods, especially in some challenging cases, e.g., insufficient texture, highly repetitive textures, similar appearances, and over-exposure.

cs.CV↗
arXiv2019

On converse bounds for classical communication over quantum channels

We explore several new converse bounds for classical communication over quantum channels in both the one-shot and asymptotic regimes. First, we show that the Matthews-Wehner meta-converse bound for entanglement-assisted classical communication can be achieved by activated, no-signalling assisted codes, suitably generalizing a result for classical channels. Second, we derive a new efficiently computable meta-converse on the amount of classical information unassisted codes can transmit over a single use of a quantum channel. As applications, we provide a finite resource analysis of classical communication over quantum erasure channels, including the second-order and moderate deviation asymptotics. Third, we explore the asymptotic analogue of our new meta-converse, the $Υ$-information of the channel. We show that its regularization is an upper bound on the classical capacity, which is generally tighter than the entanglement-assisted capacity and other known efficiently computable strong converse bounds. For covariant channels we show that the $Υ$-information is a strong converse bound.

quant-ph↗
arXiv2019

Discovery of Strongly Inverted Metallicity Gradients in Dwarf Galaxies at $z$$\sim$2

We report the first sub-kiloparsec spatial resolution measurements of strongly inverted gas-phase metallicity gradients in two dwarf galaxies at $z$$\sim$2. The galaxies have stellar masses $\sim$$10^9M_\odot$, specific star-formation rate $\sim$20 Gyr$^{-1}$, and global metallicity $12+\log({\rm O/H})\sim8.1$ (1/4 solar), assuming the Maiolino et al. (2008) strong line calibrations of OIII/Hb and OII/Hb. Their metallicity radial gradients are measured to be highly inverted, i.e., 0.122$\pm$0.008 and 0.111$\pm$0.017 dex/kpc, which is hitherto unseen at such small masses in similar redshift ranges. From the Hubble Space Telescope observations of the source nebular emission and stellar continuum, we present the 2-dimensional spatial maps of star-formation rate surface density, stellar population age, and gas fraction, which show that our galaxies are currently undergoing rapid mass assembly via disk inside-out growth. More importantly, using a simple chemical evolution model, we find that the gas fractions for different metallicity regions cannot be explained by pure gas accretion. Our spatially resolved analysis based on a more advanced gas regulator model results in a spatial map of net gaseous outflows, triggered by active central starbursts, that potentially play a significant role in shaping the spatial distribution of metallicity by effectively transporting stellar nucleosynthesis yields outwards. The relation between wind mass loading factors and stellar surface densities measured in different regions of our galaxies shows that a single type of wind mechanism, driven by either energy or momentum conservation, cannot explain the entire galaxy. These sources present a unique constraint on the effects of gas flows on the early phase of disk growth from the perspective of spatially resolved chemical evolution within individual systems.

astro-ph.GA↗
arXiv2019

Hubble Frontier Field Photometric Catalogues of Abell 370 and RXC J2248.7-4431: Multiwavelength photometry, photometric redshifts, and stellar properties

This paper presents multiwavelength photometric catalogues of the last two Hubble Frontier Fields (HFF), the massive galaxy clusters Abell 370 and RXC J2248.7-4431. The photometry ranges from imaging performed on the Hubble Space Telescope (HST) to ground based Very Large Telescope (VLT) and Spitzer/IRAC, in collaboration with the ASTRODEEP team, and using the ASTRODEEP pipeline. While the main purpose of this paper is to release the catalogues, we also perform, as a proof of concept, a brief analysis of z > 6 objects selected using drop-out method, as well as spectroscopically confirmed sources and multiple images in both clusters. While dropout methods yield a sample of high-z galaxies, the addition of longer wavelength data reveals that as expected the samples have substantial contamination at the ~30-45% level by dusty galaxies at lower redshifts. Furthermore, we show that spectroscopic redshifts are still required to unambiguously determine redshifts of multiply imaged systems. Finally, the now publicly available ASTRODEEP catalogues were combined for all HFFs and used to explore stellar properties of a large sample of 20,000 galaxies across a large photometric redshift range. The powerful magnification provided by the HFF clusters allows us an exploration of the properties of galaxies with intrinsic stellar masses as low as $M_* \gtrsim 10^7M_{\odot}$ and intrinsic star formation rates $\mbox{SFRs}\sim 0.1\mbox{-}1M_\odot/\mbox yr$ at z > 6.

astro-ph.GA↗
arXiv2019

Parity-odd Neutrino Torque Detection

Cosmological observations are promising ways to improve our understanding of neutrino mass properties. The upper bound on the sum of masses is given by the cosmic microwave background and large scale structure. These measurements are all parity-even, and potentially contaminated by unmodeled baryonic effects. In this paper we propose a novel parity-odd gravitational effect of neutrinos: A unique contribution to the directions of the angular momentum field of galaxies and halos. This observable is free of contamination in linear perturbation theory, and thus likely more cleanly separated from other nongravitational effects. A deep 21-cm survey to redshift 1 can potentially yield a $5σ$ significance on neutrino mass detection for a fiducial sum of neutrino masses of 0.05 eV.

astro-ph.CO↗
arXiv2019

Modeling the Uncertainty in Electronic Health Records: a Bayesian Deep Learning Approach

Deep learning models have exhibited superior performance in predictive tasks with the explosively increasing Electronic Health Records (EHR). However, due to the lack of transparency, behaviors of deep learning models are difficult to interpret. Without trustworthiness, deep learning models will not be able to assist in the real-world decision-making process of healthcare issues. We propose a deep learning model based on Bayesian Neural Networks (BNN) to predict uncertainty induced by data noise. The uncertainty is introduced to provide model predictions with an extra level of confidence. Our experiments verify that instances with high uncertainty are harmful to model performance. Moreover, by investigating the distributions of model prediction and uncertainty, we show that it is possible to identify a group of patients for timely intervention, such that decreasing data noise will benefit more on the prediction accuracy for these patients.

cs.LG↗
arXiv2019

Resource theory of entanglement for bipartite quantum channels

The traditional perspective in quantum resource theories concerns how to use free operations to convert one resourceful quantum state to another one. For example, a fundamental and well known question in entanglement theory is to determine the distillable entanglement of a bipartite state, which is equal to the maximum rate at which fresh Bell states can be distilled from many copies of a given bipartite state by employing local operations and classical communication for free. It is the aim of this paper to take this kind of question to the next level, with the main question being: What is the best way of using free channels to convert one resourceful quantum channel to another? Here we focus on the the resource theory of entanglement for bipartite channels and establish several fundamental tasks and results regarding it. In particular, we establish bounds on several pertinent information processing tasks in channel entanglement theory, and we define several entanglement measures for bipartite channels, including the logarithmic negativity and the $κ$-entanglement. We also show that the max-Rains information of [Bäuml et al., Physical Review Letters, 121, 250504 (2018)] has a divergence interpretation, which is helpful for simplifying the results of this earlier work.

quant-ph↗
arXiv2019

Accel: A Corrective Fusion Network for Efficient Semantic Segmentation on Video

We present Accel, a novel semantic video segmentation system that achieves high accuracy at low inference cost by combining the predictions of two network branches: (1) a reference branch that extracts high-detail features on a reference keyframe, and warps these features forward using frame-to-frame optical flow estimates, and (2) an update branch that computes features of adjustable quality on the current frame, performing a temporal update at each video frame. The modularity of the update branch, where feature subnetworks of varying layer depth can be inserted (e.g. ResNet-18 to ResNet-101), enables operation over a new, state-of-the-art accuracy-throughput trade-off spectrum. Over this curve, Accel models achieve both higher accuracy and faster inference times than the closest comparable single-frame segmentation networks. In general, Accel significantly outperforms previous work on efficient semantic video segmentation, correcting warping-related error that compounds on datasets with complex dynamics. Accel is end-to-end trainable and highly modular: the reference network, the optical flow network, and the update network can each be selected independently, depending on application requirements, and then jointly fine-tuned. The result is a robust, general system for fast, high-accuracy semantic segmentation on video.

cs.CV↗
arXiv2019

Self-Supervised Dialogue Learning

The sequential order of utterances is often meaningful in coherent dialogues, and the order changes of utterances could lead to low-quality and incoherent conversations. We consider the order information as a crucial supervised signal for dialogue learning, which, however, has been neglected by many previous dialogue systems. Therefore, in this paper, we introduce a self-supervised learning task, inconsistent order detection, to explicitly capture the flow of conversation in dialogues. Given a sampled utterance pair triple, the task is to predict whether it is ordered or misordered. Then we propose a sampling-based self-supervised network SSN to perform the prediction with sampled triple references from previous dialogue history. Furthermore, we design a joint learning framework where SSN can guide the dialogue systems towards more coherent and relevant dialogue learning through adversarial training. We demonstrate that the proposed methods can be applied to both open-domain and task-oriented dialogue scenarios, and achieve the new state-of-the-art performance on the OpenSubtitiles and Movie-Ticket Booking datasets.

cs.CL↗
arXiv2019

Towards Generating Long and Coherent Text with Multi-Level Latent Variable Models

Variational autoencoders (VAEs) have received much attention recently as an end-to-end architecture for text generation with latent variables. In this paper, we investigate several multi-level structures to learn a VAE model to generate long, and coherent text. In particular, we use a hierarchy of stochastic layers between the encoder and decoder networks to generate more informative latent codes. We also investigate a multi-level decoder structure to learn a coherent long-term structure by generating intermediate sentence representations as high-level plan vectors. Empirical results demonstrate that a multi-level VAE model produces more coherent and less repetitive long text compared to the standard VAE models and can further mitigate the posterior-collapse issue.

cs.CL↗
arXiv2019

Steering Eco-Evolutionary Games Dynamics with Manifold Control

Feedback loops between population dynamics of individuals and their ecological environment are ubiquitously found in nature, and have shown profound effects on the resulting eco-evolutionary dynamics. Incorporating linear environmental feedback law into replicator dynamics of two-player games, recent theoretical studies shed light on understanding the oscillating dynamics of social dilemma. However, detailed effects of more general nonlinear feedback loops in multi-player games, which is more common especially in microbial systems, remain unclear. Here, we focus on ecological public goods games with environmental feedbacks driven by nonlinear selection gradient. Unlike previous models, multiple segments of stable and unstable equilibrium manifolds can emerge from the population dynamical systems. We find that a larger relative asymmetrical feedback speed for group interactions centered on cooperators not only accelerates the convergence of stable manifolds, but also increases the attraction basin of these stable manifolds. Furthermore, our work offers an innovative manifold control approach: by designing appropriate switching control laws, we are able to steer the eco-evolutionary dynamics to any desired population states. Our mathematical framework is an important generalization and complement to coevolutionary game dynamics, and also fills the theoretical gap in guiding the widespread problem of population state control in microbial experiments.

physics.soc-ph↗
arXiv2019

Evolutionary dynamics of group cooperation with asymmetrical environmental feedback

In recent years, there has been growing interest in studying evolutionary games with environmental feedback. Previous studies exclusively focus on two-player games. However, extension to multi-player game is needed to study problems such as microbial cooperation and crowdsourcing collaborations. Here, we study coevolutionary public goods games where strategies coevolve with the multiplication factors of group cooperation. Asymmetry can arise in such environmental feedback, where games organized by focal cooperators may have a different efficiency than the ones by defectors. Our analysis shows that co-evolutionary dynamics with asymmetrical environmental feedback can yield oscillatory convergence to persistent cooperation, if the relative changing speed of cooperators' multiplication factor is above a certain threshold. Our work provides useful insights into sustaining group cooperation in a changing world.

physics.soc-ph↗
arXiv2019

Self-Supervised Learning for Contextualized Extractive Summarization

Existing models for extractive summarization are usually trained from scratch with a cross-entropy loss, which does not explicitly capture the global context at the document level. In this paper, we aim to improve this task by introducing three auxiliary pre-training tasks that learn to capture the document-level context in a self-supervised fashion. Experiments on the widely-used CNN/DM dataset validate the effectiveness of the proposed auxiliary tasks. Furthermore, we show that after pre-training, a clean model with simple building blocks is able to outperform previous state-of-the-art that are carefully designed.

cs.CL↗