arXiv subjects

Qi Li

Publications and source records attributed to Qi Li.

At least 433 records · Page 24Linked to original sources

arXiv2021

Optical properties of elongated conducting grains

Extremely elongated, conducting dust particles (also known as metallic "needles" or "whiskers") are seen in carbonaceous chondrites and in samples brought back from the Itokawa asteroid. Their formation in protostellar nebulae and subsequent injection into the interstellar medium have been demonstrated, both experimentally and theoretically. Metallic needles have been suggested to explain a wide variety of astrophysical phenomena, ranging from the mid-infrared interstellar extinction at ~3--8 micron to the thermalization of starlight to generate the cosmic microwave background. To validate (or invalidate) these suggestions, an accurate knowledge of the optics (e.g., the amplitude and the wavelength dependence of the absorption cross sections) of metallic needles is crucial. Here we calculate the absorption cross sections of iron needles of various aspect ratios over a wide wavelength range, by exploiting the discrete dipole approximation, the most powerful technique for rigorously calculating the optics of irregular or nonspherical grains. Our calculations support the earlier findings that the antenna theory and the Rayleigh approximation, which are often taken to approximate the optical properties of metallic needles are indeed inapplicable.

astro-ph.GA↗

arXiv2021

Low-light Image Restoration with Short- and Long-exposure Raw Pairs

Low-light imaging with handheld mobile devices is a challenging issue. Limited by the existing models and training data, most existing methods cannot be effectively applied in real scenarios. In this paper, we propose a new low-light image restoration method by using the complementary information of short- and long-exposure images. We first propose a novel data generation method to synthesize realistic short- and longexposure raw images by simulating the imaging pipeline in lowlight environment. Then, we design a new long-short-exposure fusion network (LSFNet) to deal with the problems of low-light image fusion, including high noise, motion blur, color distortion and misalignment. The proposed LSFNet takes pairs of shortand long-exposure raw images as input, and outputs a clear RGB image. Using our data generation method and the proposed LSFNet, we can recover the details and color of the original scene, and improve the low-light image quality effectively. Experiments demonstrate that our method can outperform the state-of-the art methods.

eess.IV↗

arXiv2021

Data Poisoning Attacks and Defenses to Crowdsourcing Systems

A key challenge of big data analytics is how to collect a large volume of (labeled) data. Crowdsourcing aims to address this challenge via aggregating and estimating high-quality data (e.g., sentiment label for text) from pervasive clients/users. Existing studies on crowdsourcing focus on designing new methods to improve the aggregated data quality from unreliable/noisy clients. However, the security aspects of such crowdsourcing systems remain under-explored to date. We aim to bridge this gap in this work. Specifically, we show that crowdsourcing is vulnerable to data poisoning attacks, in which malicious clients provide carefully crafted data to corrupt the aggregated data. We formulate our proposed data poisoning attacks as an optimization problem that maximizes the error of the aggregated data. Our evaluation results on one synthetic and two real-world benchmark datasets demonstrate that the proposed attacks can substantially increase the estimation errors of the aggregated data. We also propose two defenses to reduce the impact of malicious clients. Our empirical results show that the proposed defenses can substantially reduce the estimation errors of the data poisoning attacks.

cs.CR↗

arXiv2021

Truncation-Free Matching System for Display Advertising at Alibaba

Matching module plays a critical role in display advertising systems. Without query from user, it is challenging for system to match user traffic and ads suitably. System packs up a group of users with common properties such as the same gender or similar shopping interests into a crowd. Here term crowd can be viewed as a tag over users. Then advertisers bid for different crowds and deliver their ads to those targeted users. Matching module in most industrial display advertising systems follows a two-stage paradigm. When receiving a user request, matching system (i) finds the crowds that the user belongs to; (ii) retrieves all ads that have targeted those crowds. However, in applications such as display advertising at Alibaba, with very large volumes of crowds and ads, both stages of matching have to truncate the long-tailed parts for online serving, under limited latency. That's to say, not all ads have the chance to participate in online matching. This results in sub-optimal result for both advertising performance and platform revenue. In this paper, we study the truncation problem and propose a Truncation Free Matching System (TFMS). The basic idea is to decouple the matching computation from the online pipeline. Instead of executing the two-stage matching when user visits, TFMS utilizes a near-line truncation-free matching to pre-calculate and store those top valuable ads for each user. Then the online pipeline just needs to fetch the pre-stored ads as matching results. In this way, we can jump out of online system's latency and computation cost limitations, and leverage flexible computation resource to finish the user-ad matching. TFMS has been deployed in our productive system since 2019, bringing (i) more than 50% improvement of impressions for advertisers who encountered truncation before, (ii) 9.4% Revenue Per Mile gain, which is significant enough for the business.

cs.IR↗

arXiv2021

A Directed Spanning Tree Adaptive Control Framework for Time-Varying Formations

In this paper, the time-varying formation and time-varying formation tracking problems are solved for linear multi-agent systems over digraphs without the knowledge of the eigenvalues of the Laplacian matrix associated to the digraph. The solution to these problems relies on a framework that generalizes the directed spanning tree adaptive method, which was originally limited to consensus problems. Necessary and sufficient conditions for the existence of solutions to the formation problems are derived. Asymptotic convergence of the formation errors is proved via graph theory and Lyapunov analysis.

eess.SY↗

arXiv2021

Detecting Localized Adversarial Examples: A Generic Approach using Critical Region Analysis

Deep neural networks (DNNs) have been applied in a wide range of applications,e.g.,face recognition and image classification; however,they are vulnerable to adversarial examples. By adding a small amount of imperceptible perturbations,an attacker can easily manipulate the outputs of a DNN. Particularly,the localized adversarial examples only perturb a small and contiguous region of the target object,so that they are robust and effective in both digital and physical worlds. Although the localized adversarial examples have more severe real-world impacts than traditional pixel attacks,they have not been well addressed in the literature. In this paper,we propose a generic defense system called TaintRadar to accurately detect localized adversarial examples via analyzing critical regions that have been manipulated by attackers. The main idea is that when removing critical regions from input images,the ranking changes of adversarial labels will be larger than those of benign labels. Compared with existing defense solutions,TaintRadar can effectively capture sophisticated localized partial attacks, e.g.,the eye-glasses attack,while not requiring additional training or fine-tuning of the original model's structure. Comprehensive experiments have been conducted in both digital and physical worlds to verify the effectiveness and robustness of our defense.

cs.CV↗

arXiv2021

Reproducing sub-millimetre galaxy number counts with cosmological hydrodynamic simulations

Matching the number counts of high-$z$ sub-millimetre-selected galaxies (SMGs) has been a long standing problem for galaxy formation models. In this paper, we use 3D dust radiative transfer to model the sub-mm emission from galaxies in the SIMBA cosmological hydrodynamic simulations, and compare predictions to the latest single-dish observational constraints on the abundance of 850$\mathrm{μm}$-selected sources. We find good agreement with the shape of the integrated 850$\mathrm{μm}$ luminosity function, and the normalisation is within 0.25 dex at $> 3 \; \mathrm{mJy}$, unprecedented for a fully cosmological hydrodynamic simulation, along with good agreement in the redshift distribution of bright SMGs. The agreement is driven primarily by SIMBA's good match to infrared measures of the star formation rate (SFR) function between $z = 2-4$ at high SFRs. Also important is the self-consistent on-the-fly dust model in SIMBA, which predicts, on average, higher dust masses (by up to a factor of 2.5) compared to using a fixed dust-to-metals ratio of 0.3. We construct a lightcone to investigate the effect of far-field blending, and find that 52% of sources are blends of multiple components, which makes a small contribution to the normalisation of the bright-end of the number counts. We provide new fits to the 850$\mathrm{μm}$ luminosity as a function of SFR and dust mass. Our results demonstrate that exotic solutions to the discrepancy between sub-mm counts in simulations and observations, such as a top-heavy IMF, are unnecessary, and that sub-millimetre-bright phases are a natural consequence of massive galaxy evolution.

astro-ph.GA↗

arXiv2021

Data Poisoning Attacks to Deep Learning Based Recommender Systems

Recommender systems play a crucial role in helping users to find their interested information in various web services such as Amazon, YouTube, and Google News. Various recommender systems, ranging from neighborhood-based, association-rule-based, matrix-factorization-based, to deep learning based, have been developed and deployed in industry. Among them, deep learning based recommender systems become increasingly popular due to their superior performance. In this work, we conduct the first systematic study on data poisoning attacks to deep learning based recommender systems. An attacker's goal is to manipulate a recommender system such that the attacker-chosen target items are recommended to many users. To achieve this goal, our attack injects fake users with carefully crafted ratings to a recommender system. Specifically, we formulate our attack as an optimization problem, such that the injected ratings would maximize the number of normal users to whom the target items are recommended. However, it is challenging to solve the optimization problem because it is a non-convex integer programming problem. To address the challenge, we develop multiple techniques to approximately solve the optimization problem. Our experimental results on three real-world datasets, including small and large datasets, show that our attack is effective and outperforms existing attacks. Moreover, we attempt to detect fake users via statistical analysis of the rating patterns of normal and fake users. Our results show that our attack is still effective and outperforms existing attacks even if such a detector is deployed.

cs.CR↗

arXiv2020

Automatic Historical Feature Generation through Tree-based Method in Ads Prediction

Historical features are important in ads click-through rate (CTR) prediction, because they account for past engagements between users and ads. In this paper, we study how to efficiently construct historical features through counting features. The key challenge of such problem lies in how to automatically identify counting keys. We propose a tree-based method for counting key selection. The intuition is that a decision tree naturally provides various combinations of features, which could be used as counting key candidate. In order to select personalized counting features, we train one decision tree model per user, and the counting keys are selected across different users with a frequency-based importance measure. To validate the effectiveness of proposed solution, we conduct large scale experiments on Twitter video advertising data. In both online learning and offline training settings, the automatically identified counting features outperform the manually curated counting features.

cs.LG↗

arXiv2020

The Origin of the Dust Extinction Curve in Milky Way-like Galaxies

We develop a cosmological model for the evolution of dust grains in galaxies with a distribution of sizes in order to understand the origin of the Milky Way dust extinction curve. Our model considers the formation of active dust in evolved stars, growth by accretion and coagulation, and destruction processes via shattering, sputtering, and astration in the ISM of galaxies over cosmic time. Our main results follow. Galaxies in our cosmological model with masses comparable to the Milky Way's at z~0 exhibit a diverse range of extinction laws, though with slopes and bump strengths comparable to the range observed in the Galaxy. The progenitors of the Milky Way have steeper slopes, and only flatten to slopes comparable to the Galaxy at $z \approx 1$. This owes to increased grain growth rates at late times/in high-metallicity environments driving up the ratio of large to small grains, with a secondary dependence on the graphite to silicate ratio evolution. The UV bump strengths depend primarily on the graphite to silicate ratio, and remain broadly constant in MW-like galaxies between z=3 and z=0, though show slight variability. Our models span comparable regions of bump-slope space as sightlines in the Galaxy do, though there is a lack of clear relationship between the model slopes and bump strengths owing to small scale fluctuations in the bump strength. Our models naturally produce slopes for some non-Milky Way analogs as steep as those of the LMC and SMC in metal poor galaxies, though notably the bump strengths are, on average, too large when comparing to the Magellanic clouds. This owes to the fact that we evolve the grain size distributions of graphites and silicates simultaneously, which is an oversimplification. Our model provides a novel framework to study the origins and variations of dust extinction curves in galaxies over cosmic time.

astro-ph.GA↗

arXiv2020

Robust Attacks on Deep Learning Face Recognition in the Physical World

Deep neural networks (DNNs) have been increasingly used in face recognition (FR) systems. Recent studies, however, show that DNNs are vulnerable to adversarial examples, which can potentially mislead the FR systems using DNNs in the physical world. Existing attacks on these systems either generate perturbations working merely in the digital world, or rely on customized equipments to generate perturbations and are not robust in varying physical environments. In this paper, we propose FaceAdv, a physical-world attack that crafts adversarial stickers to deceive FR systems. It mainly consists of a sticker generator and a transformer, where the former can craft several stickers with different shapes and the latter transformer aims to digitally attach stickers to human faces and provide feedbacks to the generator to improve the effectiveness of stickers. We conduct extensive experiments to evaluate the effectiveness of FaceAdv on attacking 3 typical FR systems (i.e., ArcFace, CosFace and FaceNet). The results show that compared with a state-of-the-art attack, FaceAdv can significantly improve success rate of both dodging and impersonating attacks. We also conduct comprehensive evaluations to demonstrate the robustness of FaceAdv.

cs.CV↗

arXiv2020

Style Intervention: How to Achieve Spatial Disentanglement with Style-based Generators?

Generative Adversarial Networks (GANs) with style-based generators (e.g. StyleGAN) successfully enable semantic control over image synthesis, and recent studies have also revealed that interpretable image translations could be obtained by modifying the latent code. However, in terms of the low-level image content, traveling in the latent space would lead to `spatially entangled changes' in corresponding images, which is undesirable in many real-world applications where local editing is required. To solve this problem, we analyze properties of the 'style space' and explore the possibility of controlling the local translation with pre-trained style-based generators. Concretely, we propose 'Style Intervention', a lightweight optimization-based algorithm which could adapt to arbitrary input images and render natural translation effects under flexible objectives. We verify the performance of the proposed framework in facial attribute editing on high-resolution images, where both photo-realism and consistency are required. Extensive qualitative results demonstrate the effectiveness of our method, and quantitative measurements also show that the proposed algorithm outperforms state-of-the-art benchmarks in various aspects.

cs.CV↗

arXiv2020

Dependence of great geomagnetic storm intensity ($Δ$SYM-H$\le$-200 nT) on associated solar wind parameters

We use $Δ$SYM-H to capture the variation in the SYM-H index during the main phase of a geomagnetic storm. We define great geomagnetic storms as those with $Δ$SYM-H $\le$ -200 nT. After analyzing the data that were not obscured by solar winds, we determined that 11 such storms occurred during solar cycle 23. We calculated time integrals for the southward interplanetary magnetic field component I(B$_s$), the solar wind electric field I(E$_y$), and a combination of E$_y$ and the solar wind dynamic pressure I(Q) during the main phase of a great geomagnetic storm. The strength of the correlation coefficient (CC) between $Δ$SYM-H and each of the three integrals I(B$_s$) (CC = 0.74), I(E$_y$) (CC = 0.85), and I(Q) (CC = 0.94) suggests that Q, which encompasses both the solar wind electric field and the solar wind dynamic pressure, is the main driving factor that determines the intensity of a great geomagnetic storm. The results also suggest that the impact of B$_s$ on the great geomagnetic storm intensity is much more significant than that of the solar wind speed and the dynamic pressure during the main phase of associated great geomagnetic storm. How to estimate the intensity of an extreme geomagnetic storm based on solar wind parameters is also discussed.

physics.space-ph↗

arXiv2020

Causality Extraction based on Self-Attentive BiLSTM-CRF with Transferred Embeddings

Causality extraction from natural language texts is a challenging open problem in artificial intelligence. Existing methods utilize patterns, constraints, and machine learning techniques to extract causality, heavily depending on domain knowledge and requiring considerable human effort and time for feature engineering. In this paper, we formulate causality extraction as a sequence labeling problem based on a novel causality tagging scheme. On this basis, we propose a neural causality extractor with the BiLSTM-CRF model as the backbone, named SCITE (Self-attentive BiLSTM-CRF wIth Transferred Embeddings), which can directly extract cause and effect without extracting candidate causal pairs and identifying their relations separately. To address the problem of data insufficiency, we transfer contextual string embeddings, also known as Flair embeddings, which are trained on a large corpus in our task. In addition, to improve the performance of causality extraction, we introduce a multihead self-attention mechanism into SCITE to learn the dependencies between causal words. We evaluate our method on a public dataset, and experimental results demonstrate that our method achieves significant and consistent improvement compared to baselines.

cs.CL↗

arXiv2020

Manifold learning-based feature extraction for structural defect reconstruction

Data-driven quantitative defect reconstructions using ultrasonic guided waves has recently demonstrated great potential in the area of non-destructive testing. In this paper, we develop an efficient deep learning-based defect reconstruction framework, called NetInv, which recasts the inverse guided wave scattering problem as a data-driven supervised learning progress that realizes a mapping between reflection coefficients in wavenumber domain and defect profiles in the spatial domain. The superiorities of the proposed NetInv over conventional reconstruction methods for defect reconstruction have been demonstrated by several examples. Results show that NetInv has the ability to achieve the higher quality of defect profiles with remarkable efficiency and provides valuable insight into the development of effective data driven structural health monitoring and defect reconstruction using machine learning.

cs.CE↗

arXiv2020

Capping layer influence and isotropic in-plane upper critical field of the superconductivity at the FeSe/SrTiO3 interface

Understanding the superconductivity at the interface of FeSe/SrTiO3 is a problem of great contemporary interest due to the significant increase in critical temperature (Tc) compared to that of bulk FeSe, as well as the possibility of an unconventional pairing mechanism and topological superconductivity. We report a study of the influence of a capping layer on superconductivity in thin films of FeSe grown on SrTiO3 using molecular beam epitaxy. We used in vacuo four-probe electrical resistance measurements and ex situ magneto-transport measurements to examine the effect of three capping layers that provide distinctly different charge transfer into FeSe: compound FeTe, non-metallic Te, and metallic Zr. Our results show that FeTe provides an optimal cap that barely influences the inherent Tc found in pristine FeSe/SrTiO3, while the transfer of holes from a non-metallic Te cap completely suppresses superconductivity and leads to insulating behavior. Finally, we used ex situ magnetoresistance measurements in FeTe-capped FeSe films to extract the angular dependence of the in-plane upper critical magnetic field. Our observations reveal an almost isotropic in-plane upper critical field, providing insight into the symmetry and pairing mechanism of high temperature superconductivity in FeSe.

cond-mat.supr-con↗

arXiv2020

A cost-effective method for improving and re-purposing large, pre-trained GANs by fine-tuning their class-embeddings

Large, pre-trained generative models have been increasingly popular and useful to both the research and wider communities. Specifically, BigGANs a class-conditional Generative Adversarial Networks trained on ImageNet---achieved excellent, state-of-the-art capability in generating realistic photos. However, fine-tuning or training BigGANs from scratch is practically impossible for most researchers and engineers because (1) GAN training is often unstable and suffering from mode-collapse; and (2) the training requires a significant amount of computation, 256 Google TPUs for 2 days or 8xV100 GPUs for 15 days. Importantly, many pre-trained generative models both in NLP and image domains were found to contain biases that are harmful to society. Thus, we need computationally-feasible methods for modifying and re-purposing these huge, pre-trained models for downstream tasks. In this paper, we propose a cost-effective optimization method for improving and re-purposing BigGANs by fine-tuning only the class-embedding layer. We show the effectiveness of our model-editing approach in three tasks: (1) significantly improving the realism and diversity of samples of complete mode-collapse classes; (2) re-purposing ImageNet BigGANs for generating images for Places365; and (3) de-biasing or improving the sample diversity for selected ImageNet classes.

cs.LG↗

arXiv2020

On the Absorption Properties of Metallic Needles

Needle-like metallic particles have been suggested to explain a wide variety of astrophysical phenomena, ranging from the mid-infrared interstellar extinction to the thermalization of starlight to generate the cosmic microwave background. These suggestions rely on the amplitude and the wavelength dependence of the absorption cross sections of metallic needles. On the absence of an exact solution to the absorption properties of metallic needles, their absorption cross sections are often derived from the antenna approximation. However, it is shown here that the antenna approximation is not an appropriate representation since it violates the Kramers-Kronig relation. Stimulated by the recent discovery of iron whiskers in asteroid Itokawa and graphite whiskers in carbonaceous chondrites, we call for rigorous calculations of the absorption cross sections of metallic needle-like particles, presumably with the discrete dipole approximation. We also call for experimental studies of the formation and growth mechanisms of metallic needle-like particles as well as experimental measurements of the absorption cross sections of metallic needles of various aspect ratios over a wide wavelength range to bound theoretical calculations.

astro-ph.GA↗