Searcharxiv⌕ Search

arXiv subjects

Qi Li

Publications and source records attributed to Qi Li.

At least 361 records · Page 20Linked to original sources
arXiv2023

Trackerless freehand ultrasound with sequence modelling and auxiliary transformation over past and future frames

Three-dimensional (3D) freehand ultrasound (US) reconstruction without a tracker can be advantageous over its two-dimensional or tracked counterparts in many clinical applications. In this paper, we propose to estimate 3D spatial transformation between US frames from both past and future 2D images, using feed-forward and recurrent neural networks (RNNs). With the temporally available frames, a further multi-task learning algorithm is proposed to utilise a large number of auxiliary transformation-predicting tasks between them. Using more than 40,000 US frames acquired from 228 scans on 38 forearms of 19 volunteers in a volunteer study, the hold-out test performance is quantified by frame prediction accuracy, volume reconstruction overlap, accumulated tracking error and final drift, based on ground-truth from an optical tracker. The results show the importance of modelling the temporal-spatially correlated input frames as well as output transformations, with further improvement owing to additional past and/or future frames. The best performing model was associated with predicting transformation between moderately-spaced frames, with an interval of less than ten frames at 20 frames per second (fps). Little benefit was observed by adding frames more than one second away from the predicted transformation, with or without LSTM-based RNNs. Interestingly, with the proposed approach, explicit within-sequence loss that encourages consistency in composing transformations or minimises accumulated error may no longer be required. The implementation code and volunteer data will be made publicly available ensuring reproducibility and further research.

eess.IV↗
arXiv2023

Detecting Unknown Encrypted Malicious Traffic in Real Time via Flow Interaction Graph Analysis

In this paper, we propose HyperVision, a realtime unsupervised machine learning (ML) based malicious traffic detection system. Particularly, HyperVision is able to detect unknown patterns of encrypted malicious traffic by utilizing a compact inmemory graph built upon the traffic patterns. The graph captures flow interaction patterns represented by the graph structural features, instead of the features of specific known attacks. We develop an unsupervised graph learning method to detect abnormal interaction patterns by analyzing the connectivity, sparsity, and statistical features of the graph, which allows HyperVision to detect various encrypted attack traffic without requiring any labeled datasets of known attacks. Moreover, we establish an information theory model to demonstrate that the information preserved by the graph approaches the ideal theoretical bound. We show the performance of HyperVision by real-world experiments with 92 datasets including 48 attacks with encrypted malicious traffic. The experimental results illustrate that HyperVision achieves at least 0.92 AUC and 0.86 F1, which significantly outperform the state-of-the-art methods. In particular, more than 50% attacks in our experiments can evade all these methods. Moreover, HyperVision achieves at least 80.6 Gb/s detection throughput with the average detection latency of 0.83s.

cs.CR↗
arXiv2023

A Framework for Modeling Polycyclic Aromatic Hydrocarbon Emission in Galaxy Evolution Simulations

We present a new methodology for simulating mid-infrared emission from polycyclic aromatic hydrocarbons (PAHs) in galaxy evolution simulations. To do this, we combine theoretical models of PAH emission features as they respond to varying interstellar radiation fields, grain size distributions, and ionization states with a new on-the-fly model for dust evolution in hydrodynamic galaxy simulations. We apply these models to 3 idealized arepo galaxy evolution simulations within the smuggle physics framework. We use these simulations to develop numerical experiments investigating the buildup of PAH masses and luminosities in galaxies in idealized analogs of the Milky Way, a dwarf galaxy, and starburst disk. Our main results follow. Galaxies with high specific star formation rates have increased feedback energy per unit mass, and are able to efficiently shatter dust grains, driving up the fraction of ultra small grains. At the same time, in our model large radiation fields per unit gas density convert aliphatic grains into aromatics. The fraction of dust grains in the form of PAHs (q_PAH) can be understood as a consequence of these processes, and in our model PAHs form primarily from interstellar processing (shattering) of larger grains rather than from the growth of smaller grains. We find that the hardness of the radiation field plays a larger role than variations in the grain size distribution in setting the total integrated PAH luminosities, though cosmological simulations are necessary to fully investigate the complex interplay of processes that drive PAH band luminosities in galaxies. Finally, we highlight feature PAH strength variations, cautioning against the usage of emission templates with constant feature strength ratios.

astro-ph.GA↗
arXiv2023

The time-evolving impact of tree size on nighttime street canyon microclimate: Wind tunnel modeling of aerodynamic effects and heat removal

Urban trees play a crucial role in urban climate in many aspects. However, existing research has not adequately explored the impact from a time-evolving perspective, that is, tree growth over time. To bridge this research gap, this study investigates in a wind tunnel the effects of tree-to-canyon foliage cover and relative height (0.32-1.1 times canyon height), mimicking growth of trees, on conditions in street canyons during moderate and extreme heat. The results reveal that trees may affect canyon-wide ventilation and heat removal in two different scenarios. First, when canyons are in isothermal conditions, medium and large trees, that fill half the canyon height or reach slightly above the canyon, decelerate the shear layer and weaken the vortical flow, as a result reducing the canyon-wide ventilation. Second, in extreme heat conditions, medium and large trees trap heat at the pedestrian level due to the blockage of air entrainment and the suppression of upward buoyancy-driven flow from the ground surface. An air temperature rise that corresponds to 1.5 degree Celsius in a full-scale urban setting is observed in measurements. These observations suggest that urban trees' foliage cover must be managed for a canyon's optimal ventilation and heat removal during nighttime.

physics.flu-dyn↗
arXiv2023

Heterogeneous Graph Contrastive Multi-view Learning

Inspired by the success of contrastive learning (CL) in computer vision and natural language processing, graph contrastive learning (GCL) has been developed to learn discriminative node representations on graph datasets. However, the development of GCL on Heterogeneous Information Networks (HINs) is still in the infant stage. For example, it is unclear how to augment the HINs without substantially altering the underlying semantics, and how to design the contrastive objective to fully capture the rich semantics. Moreover, early investigations demonstrate that CL suffers from sampling bias, whereas conventional debiasing techniques are empirically shown to be inadequate for GCL. How to mitigate the sampling bias for heterogeneous GCL is another important problem. To address the aforementioned challenges, we propose a novel Heterogeneous Graph Contrastive Multi-view Learning (HGCML) model. In particular, we use metapaths as the augmentation to generate multiple subgraphs as multi-views, and propose a contrastive objective to maximize the mutual information between any pairs of metapath-induced views. To alleviate the sampling bias, we further propose a positive sampling strategy to explicitly select positives for each node via jointly considering semantic and structural information preserved on each metapath view. Extensive experiments demonstrate HGCML consistently outperforms state-of-the-art baselines on five real-world benchmark datasets.

cs.LG↗
arXiv2022

The entanglement entropy of the quantum Hall edge and its geometric contribution

Generally speaking, the entanglement entropy (EE) between two subregions of a gapped quantum many-body state is proportional to the area/length of their interface due to the short range quantum correlation. However, the so-called area law is violated logarithmically in a quantum critical phase. Moreover, the subleading correction exists in a long range entangled topological phases. It is referred to as topological EE which is related to the quantum dimension of the collective excitation in the bulk. Further more, if a non-smooth sharp angle is in the presence of the subsystem boundary, a universal angle dependent geometric contribution is expected to appear in the subleading correction. In this work, we simultaneously explore the geometric and edge contribution in the integer quantum Hall (IQH) state and its edge reconstruction in a unified bipartite method. Their scaling is found to be consistent with the conformal field theory (CFT) predictions and recent results of the particle number fluctuation calculations.

cond-mat.str-el↗
arXiv2022

Ultra-high Resolution Image Segmentation via Locality-aware Context Fusion and Alternating Local Enhancement

Ultra-high resolution image segmentation has raised increasing interests in recent years due to its realistic applications. In this paper, we innovate the widely used high-resolution image segmentation pipeline, in which an ultra-high resolution image is partitioned into regular patches for local segmentation and then the local results are merged into a high-resolution semantic mask. In particular, we introduce a novel locality-aware context fusion based segmentation model to process local patches, where the relevance between local patch and its various contexts are jointly and complementarily utilized to handle the semantic regions with large variations. Additionally, we present the alternating local enhancement module that restricts the negative impact of redundant information introduced from the contexts, and thus is endowed with the ability of fixing the locality-aware features to produce refined results. Furthermore, in comprehensive experiments, we demonstrate that our model outperforms other state-of-the-art methods in public benchmarks. Our released codes are available at: https://github.com/liqiokkk/FCtL.

cs.CV↗
arXiv2022

Cosmic Sands: The Origin of Dusty, Star-forming Galaxies in the Epoch of Reionization

We present the Cosmic Sands suite of cosmological zoom-in simulations based on the Simba galaxy formation model in order to study the build up of the first massive and dusty galaxies in the early Universe. Residing in the most massive halos, we find that the compact proto-massive galaxies undergo nearly continuous mergers with smaller subhalos, boosting star formation rates (SFRs) and the build up of stellar mass. The galaxies are already appreciably chemically evolved by z=10, with modeled dust masses comparable to those inferred from observations in the same epoch. We track gas accretion onto the galaxies to understand how extreme SFRs can be sustained by these early systems. We find that smooth gas accretion can maintain SFRs above 250 M$_{\odot}$ / yr but to achieve SFRs that boost galaxies well above the main sequence, a larger perturbation like a gas-rich major merger is necessary to trigger a starburst episode. Post-processing the Cosmic Sands simulations with dust radiative transfer, we find that while the infrared luminosities of the most dust rich galaxies are comparable to local ULIRGs, they are substantially dimmer than classical z=2 sub-millimeter galaxies. We end with a discussion on the possible reasons for this discrepancy at the highest masses and the future work we intend to carry out to study the chemical enrichment of the earliest dusty galaxies.

astro-ph.GA↗
arXiv2022

Existence of solutions to a perturbed critical biharmonic equation with Hardy potential

\ In this paper, the following biharmonic elliptic problem \begin{eqnarray*} \begin{cases} Δ^2u-λ\frac{|u|^{q-2}u}{|x|^s}=|u|^{2^{**}-2}u+ f(x,u), &x\inΩ,\\ u=\dfrac{\partial u}{\partial n}=0, &x\in\partialΩ\end{cases} \end{eqnarray*} is considered. The main feature of the equation is that it involves a Hardy term and a nonlinearity with critical Sobolev exponent. By combining a careful analysis of the fibering maps of the energy functional associated with the problem with the Mountain Pass Lemma, it is shown, for some positive parameter $λ$ depending on $s$ and $q$, that the problem admits at least one mountain pass type solution under appropriate growth conditions on the nonlinearity $f(x,u)$.

math.AP↗
arXiv2022

Semantic-aware One-shot Face Re-enactment with Dense Correspondence Estimation

One-shot face re-enactment is a challenging task due to the identity mismatch between source and driving faces. Specifically, the suboptimally disentangled identity information of driving subjects would inevitably interfere with the re-enactment results and lead to face shape distortion. To solve this problem, this paper proposes to use 3D Morphable Model (3DMM) for explicit facial semantic decomposition and identity disentanglement. Instead of using 3D coefficients alone for re-enactment control, we take the advantage of the generative ability of 3DMM to render textured face proxies. These proxies contain abundant yet compact geometric and semantic information of human faces, which enable us to compute the face motion field between source and driving images by estimating the dense correspondence. In this way, we could approximate re-enactment results by warping source images according to the motion field, and a Generative Adversarial Network (GAN) is adopted to further improve the visual quality of warping results. Extensive experiments on various datasets demonstrate the advantages of the proposed method over existing start-of-the-art benchmarks in both identity preservation and re-enactment fulfillment.

cs.CV↗
arXiv2022

Lattice eddy simulation of turbulent flows

Kolmogorov's (1941) theory of self-similarity implies the universality of small-scale eddies, and holds promise for a universal sub-grid scale model for large eddy simulation. The fact is the empirical coefficient of a typical sub-grid scale model varies from 0.1 to 0.2 in free turbulence and damps gradually to zero approaching the walls. This work has developed a Lattice Eddy Simulation method (LAES), in which the sole empirical coefficient is constant (Cs=0.08). LAES assumes the fluid properties are stored in the nodes of a typical CFD mesh, treats the nodes as lattices and makes analysis on one specific lattice, i. To be specific, LAES express the domain derivative on that lattice with the influence of nearby lattices. The lattices right next to i, which is named as i+, "collide" with i, imposing convective effects on i. The lattices right next to i+, which is named as i++, impose convective effects on i+ and indirectly influence i. The influence is actually turbulent diffusion. The derived governing equations of LAES look like the Navier-Stokes equations and reduce to filtered Naiver-Stokes equations with the Smagorinsky sub-grid scale model (Smagorinsky 1963) on meshes with isotropic cells. LAES yields accurate predictions of turbulent channel flows at Re=180, 395, and 590 on very coarse meshes and LAES with a constant Cs perform as well as the dynamic LES model (Germano et al. 1991) does. Thus, this work has provided strong evidence for Kolmogorov's theory of self-similarity.

physics.flu-dyn↗
arXiv2022

A coupled Hartree system with Hardy-Littlewood-Sobolev critical exponent: existence and multiplicity of high energy positive solutions

This paper deals with a coupled Hartree system with Hardy-Littlewood-Sobolev critical exponent \begin{equation*} \begin{cases} -Δu+(V_1(x)+λ_1)u=μ_1(|x|^{-4}*u^{2})u+β(|x|^{-4}*v^{2})u, \ \ &x\in R^N, -Δv+(V_2(x)+λ_2)v=μ_2(|x|^{-4}*v^{2})v+β(|x|^{-4}*u^{2})v, \ \ &x\in R^N, \end{cases} \end{equation*} where $N\geq 5$, $λ_1$, $λ_2\geq 0$ with $λ_1+λ_2\neq 0$, $V_1(x), V_{2}(x)\in L^{\frac{N}{2}}(R^N)$ are nonnegative functions and $μ_1$, $μ_2$, $β$ are positive constants. Such system arises from mathematical models in Bose-Einstein condensates theory and nonlinear optics. By variational methods combined with degree theory, we prove some results about the existence and multiplicity of high energy positive solutions under the hypothesis $β>\max\{μ_1,μ_2\}$

math.AP↗
arXiv2022

Existence and nonexistence of solutions to a critical biharmonic equation with logarithmic perturbation

In this paper, the following critical biharmonic elliptic problem \begin{eqnarray*} \begin{cases} Δ^2u= λu+μu\ln u^2+|u|^{2^{**}-2}u, &x\inΩ,\\ u=\dfrac{\partial u}{\partial ν}=0, &x\in\partialΩ\end{cases} \end{eqnarray*} is considered, where $Ω\subset \mathbb{R}^{N}$ is a bounded smooth domain with $N\geq5$. Some interesting phenomenon occurs due to the uncertainty of the sign of the logarithmic term. It is shown, mainly by using Mountain Pass Lemma, that the problem admits at lest one nontrivial weak solution under some appropriate assumptions of $λ$ and $μ$. Moreover, a nonexistence result is also obtained. Comparing the results in this paper with the known ones, one sees that some new phenomena occur when the logarithmic perturbation is introduced.

math.AP↗
arXiv2022

Monocular BEV Perception of Road Scenes via Front-to-Top View Projection

HD map reconstruction is crucial for autonomous driving. LiDAR-based methods are limited due to expensive sensors and time-consuming computation. Camera-based methods usually need to perform road segmentation and view transformation separately, which often causes distortion and missing content. To push the limits of the technology, we present a novel framework that reconstructs a local map formed by road layout and vehicle occupancy in the bird's-eye view given a front-view monocular image only. We propose a front-to-top view projection (FTVP) module, which takes the constraint of cycle consistency between views into account and makes full use of their correlation to strengthen the view transformation and scene understanding. In addition, we also apply multi-scale FTVP modules to propagate the rich spatial information of low-level features to mitigate spatial deviation of the predicted object location. Experiments on public benchmarks show that our method achieves the state-of-the-art performance in the tasks of road layout estimation, vehicle occupancy estimation, and multi-class semantic estimation. For multi-class semantic estimation, in particular, our model outperforms all competitors by a large margin. Furthermore, our model runs at 25 FPS on a single GPU, which is efficient and applicable for real-time panorama HD map reconstruction.

cs.CV↗
arXiv2022

Standing waves for two-component elliptic system with critical growth in $\mathbb{R}^{4}$: the attractive case

In this paper, we consider the following two-component elliptic system with critical growth \begin{equation*} \begin{cases} -Δu+(V_1(x)+λ)u=μ_1u^{3}+βuv^{2}, \ \ x\in \mathbb{R}^4, -Δv+(V_2(x)+λ)v=μ_2v^{3}+βvu^{2}, \ \ x\in \mathbb{R}^4 , % u\geq 0, \ \ v\geq 0 \ \text{in} \ \R^4. \end{cases} \end{equation*} where $V_j(x) \in L^{2}(\mathbb{R}^4)$ are nonnegative potentials and the nonlinear coefficients $β,μ_j$, $j=1,2$, are positive. Here we also assume $λ>0$. By variational methods combined with degree theory, we prove some results about the existence and multiplicity of positive solutions under the hypothesis $β>\max\{μ_1,μ_2\}$. These results generalize the results for semilinear Schrödinger equation on half space by Cerami and Passaseo (SIAM J. Math. Anal., 28, 867-885, (1997)) to the above elliptic system, while extending the existence result from Liu and Liu (Calc. Var. Partial Differential Equations, 59:145, (2020)).

math.AP↗
arXiv2022

Supervised Contrastive Learning with Tree-Structured Parzen Estimator Bayesian Optimization for Imbalanced Tabular Data

Class imbalance has a detrimental effect on the predictive performance of most supervised learning algorithms as the imbalanced distribution can lead to a bias preferring the majority class. To solve this problem, we propose a Supervised Contrastive Learning (SCL) method with Tree-structured Parzen Estimator (TPE) technique for imbalanced tabular datasets. Contrastive learning (CL) can extract the information hidden in data even without labels and has shown some potential for imbalanced learning tasks. SCL further considers the label information based on CL, which also addresses the insufficient data augmentation techniques of tabular data. Therefore, in this work, we propose to use SCL to learn a discriminative representation of imbalanced tabular data. Additionally, the hyper-parameter temperature of SCL has a decisive influence on the performance and is difficult to tune. We introduce TPE, a well-known Bayesian optimization technique, to automatically select the best temperature. Experiments are conducted on both binary and multi-class imbalanced tabular datasets. As shown in the results obtained, TPE outperforms three other hyper-parameter optimization (HPO) methods such as grid search, random search, and genetic algorithm. More importantly, the proposed SCL-TPE method achieves much-improved performance compared with the state-of-the-art methods.

cs.LG↗
arXiv2022

GAN-based Facial Attribute Manipulation

Facial Attribute Manipulation (FAM) aims to aesthetically modify a given face image to render desired attributes, which has received significant attention due to its broad practical applications ranging from digital entertainment to biometric forensics. In the last decade, with the remarkable success of Generative Adversarial Networks (GANs) in synthesizing realistic images, numerous GAN-based models have been proposed to solve FAM with various problem formulation approaches and guiding information representations. This paper presents a comprehensive survey of GAN-based FAM methods with a focus on summarizing their principal motivations and technical details. The main contents of this survey include: (i) an introduction to the research background and basic concepts related to FAM, (ii) a systematic review of GAN-based FAM methods in three main categories, and (iii) an in-depth discussion of important properties of FAM methods, open issues, and future research directions. This survey not only builds a good starting point for researchers new to this field but also serves as a reference for the vision community.

cs.CV↗
arXiv2022

Boosting the convergence of low-variance DSMC by GSIS

The low-variance direct simulation Monte Carlo (LVDSMC) is a powerful method to simulate low-speed rarefied gas flows. However, in the near-continuum flow regime, due to limitations on the time step and spatial cell size, it takes plenty of time to find the steady-state solution. Here we remove these deficiencies by coupling the LVDSMC with the general synthetic iterative scheme (GSIS) which permits the simulation at the hydrodynamic scale rather than the much smaller kinetic scale. As a proof of concept, we propose the stochastic-deterministic coupling method based on the Bhatnagar-Gross-Krook kinetic model. First, macroscopic synthetic equations are derived exactly from the kinetic equation, which not only contain the Navier-Stokes-Fourier constitutive relation, but also encompass the higher-order terms describing the rarefaction effects. Then, the high-order terms are extracted from LVDSMC and fed into synthetic equations to predict macroscopic properties which are closer to the steady-state solution than LVDSMC. Finally, the state of simulation particles in LVDSMC is updated to reflect the change of macroscopic properties. As a result, the convergence to steady state is greatly accelerated, and the restriction on cell size and the time step are removed: after simulating several canonical rarefied gas flows, we demonstrate that the LVDSMC-GSIS reduces the computational cost by two orders of magnitude in the near-continuum flow regime.

physics.flu-dyn↗