arXiv subjects

Jia Zhang

Publications and source records attributed to Jia Zhang.

At least 109 records · Page 6Linked to original sources

arXiv2023

Weakly Supervised Semantic Segmentation by Knowledge Graph Inference

Currently, existing efforts in Weakly Supervised Semantic Segmentation (WSSS) based on Convolutional Neural Networks (CNNs) have predominantly focused on enhancing the multi-label classification network stage, with limited attention given to the equally important downstream segmentation network. Furthermore, CNN-based local convolutions lack the ability to model the extensive inter-category dependencies. Therefore, this paper introduces a graph reasoning-based approach to enhance WSSS. The aim is to improve WSSS holistically by simultaneously enhancing both the multi-label classification and segmentation network stages. In the multi-label classification network segment, external knowledge is integrated, coupled with GCNs, to globally reason about inter-class dependencies. This encourages the network to uncover features in non-salient regions of images, thereby refining the completeness of generated pseudo-labels. In the segmentation network segment, the proposed Graph Reasoning Mapping (GRM) module is employed to leverage knowledge obtained from textual databases, facilitating contextual reasoning for class representation within image regions. This GRM module enhances feature representation in high-level semantics of the segmentation network's local convolutions, while dynamically learning semantic coherence for individual samples. Using solely image-level supervision, we have achieved state-of-the-art performance in WSSS on the PASCAL VOC 2012 and MS-COCO datasets. Extensive experimentation on both the multi-label classification and segmentation network stages underscores the effectiveness of the proposed graph reasoning approach for advancing WSSS.

cs.CV↗

arXiv2023

An extremely bad-cavity laser

Lasing in the bad-cavity regime has promising applications in precision measurement and frequency metrology due to the reduced sensitivity of the laser frequency to cavity length fluctuations. Thus far, relevant studies have been mainly focused on conventional cavities whose finesse is high enough that the resonance linewidth is sufficiently narrow compared to the cavity's free spectral range, though still in the bad-cavity regime. However, lasing output from the cavity whose finesse is close to the limit of 2 has never been experimentally accessed. Here, we demonstrate an extremely bad-cavity laser, analyze the physical mechanisms limiting cavity finesse, and report on the worst ever laser cavity with finesse reaching 2.01. The optical cavity has a reflectance close to zero and only provides a weak optical feedback. The laser power can be as high as tens of $μ$W and the spectral linewidth reaches a few kHz, over one thousand times narrower than the gain bandwidth. In addition, the measurement of cavity pulling reveals a pulling coefficient of 0.0148, the lowest value ever achieved for a continuous wave laser. Our findings open up an unprecedentedly innovative perspective for future new ultra-stable lasers, which could possibly trigger the future discoveries in optical clocks, cavity QED, continuous wave superradiant laser, and explorations of quantum manybody physics.

physics.optics↗

arXiv2023

Giant 2D Skyrmion Topological Hall Effect with Ultrawide Temperature Window and Low-Current Manipulation in 2D Room-Temperature Ferromagnetic Crystals

The discovery and manipulation of topological Hall effect (THE), an abnormal magnetoelectric response mostly related to the Dzyaloshinskii-Moriya interaction (DMI), are promising for next-generation spintronic devices based on topological spin textures such as magnetic skyrmions. However, most skyrmions and THE are stabilized in a narrow temperature window either below or over room temperature with high critical current manipulation. It is still elusive and challenging to achieve large THE with both wide temperature window till room temperature and low critical current manipulation. Here, by using controllable, naturally-oxidized, sub-20 and sub-10 nm 2D van der Waals room-temperature ferromagnetic Fe3GaTe2-x crystals, robust 2D THE with ultrawide temperature window ranging in three orders of magnitude from 2 to 300 K is reported, combining with giant THE of ~5.4 micro-ohm cm at 10 K and ~0.15 micro-ohm cm at 300 K which is 1-3 orders of magnitude larger than that of all known room-temperature 2D skyrmion systems. Moreover, room-temperature current-controlled THE is also realized with a low critical current density of ~6.2*10^5 A cm^-2. First-principles calculations unveil natural oxidation-induced highly-enhanced 2D interfacial DMI reasonable for robust giant THE. This work paves the way to room-temperature, electrically-controlled 2D THE-based practical spintronic devices.

cond-mat.mtrl-sci↗

arXiv2023

Deep3DSketch+: Rapid 3D Modeling from Single Free-hand Sketches

The rapid development of AR/VR brings tremendous demands for 3D content. While the widely-used Computer-Aided Design (CAD) method requires a time-consuming and labor-intensive modeling process, sketch-based 3D modeling offers a potential solution as a natural form of computer-human interaction. However, the sparsity and ambiguity of sketches make it challenging to generate high-fidelity content reflecting creators' ideas. Precise drawing from multiple views or strategic step-by-step drawings is often required to tackle the challenge but is not friendly to novice users. In this work, we introduce a novel end-to-end approach, Deep3DSketch+, which performs 3D modeling using only a single free-hand sketch without inputting multiple sketches or view information. Specifically, we introduce a lightweight generation network for efficient inference in real-time and a structural-aware adversarial training approach with a Stroke Enhancement Module (SEM) to capture the structural information to facilitate learning of the realistic and fine-detailed shape structures for high-fidelity performance. Extensive experiments demonstrated the effectiveness of our approach with the state-of-the-art (SOTA) performance on both synthetic and real datasets.

cs.CV↗

arXiv2023

Low-Quality Training Data Only? A Robust Framework for Detecting Encrypted Malicious Network Traffic

Machine learning (ML) is promising in accurately detecting malicious flows in encrypted network traffic; however, it is challenging to collect a training dataset that contains a sufficient amount of encrypted malicious data with correct labels. When ML models are trained with low-quality training data, they suffer degraded performance. In this paper, we aim at addressing a real-world low-quality training dataset problem, namely, detecting encrypted malicious traffic generated by continuously evolving malware. We develop RAPIER that fully utilizes different distributions of normal and malicious traffic data in the feature space, where normal data is tightly distributed in a certain area and the malicious data is scattered over the entire feature space to augment training data for model training. RAPIER includes two pre-processing modules to convert traffic into feature vectors and correct label noises. We evaluate our system on two public datasets and one combined dataset. With 1000 samples and 45% noises from each dataset, our system achieves the F1 scores of 0.770, 0.776, and 0.855, respectively, achieving average improvements of 352.6%, 284.3%, and 214.9% over the existing methods, respectively. Furthermore, We evaluate RAPIER with a real-world dataset obtained from a security enterprise. RAPIER effectively achieves encrypted malicious traffic detection with the best F1 score of 0.773 and improves the F1 score of existing methods by an average of 272.5%.

cs.CR↗

arXiv2023

Probing Dark QCD Sector through the Higgs Portal with Machine Learning at the LHC

The QCD-like dark sector with GeV-scale dark hadrons has the potential to generate new signatures at the Large Hadron Collider (LHC). In this paper, we consider a singlet scalar mediator in the tens of GeV-scale that connects the dark sector and the Standard Model (SM) sector via the Higgs portal. We focus on the Higgs-strahlung process, $q\overline{q}'\rightarrow W^{\ast}\rightarrow WH $, to produce a highly boosted Higgs boson. Our scenario predicts two different processes that can generate dark mesons: (1) the cascade decay from the Higgs boson to two light scalar mediators and then to four dark mesons; (2) the Higgs boson decaying to two dark quarks, which then undergo a QCD-like shower and hadronization to produce dark mesons. We apply machine learning techniques, such as Convolutional Neural Network (CNN) and Energy Flow Network (EFN), to the fat jet structure to distinguish these signal processes from large SM backgrounds. We find that the branching ratio of the Higgs boson to two light scalar mediators can be constrained to be less than $10\%$ at 14 TeV LHC with $\mathcal{L} = 3000 fb^{-1}$.

hep-ph↗

arXiv2023

Achieving Covert Communication With A Probabilistic Jamming Strategy

In this work, we consider a covert communication scenario, where a transmitter Alice communicates to a receiver Bob with the aid of a probabilistic and uninformed jammer against an adversary warden's detection. The transmission status and power of the jammer are random and follow some priori probabilities. We first analyze the warden's detection performance as a function of the jammer's transmission probability, transmit power distribution, and Alice's transmit power. We then maximize the covert throughput from Alice to Bob subject to a covertness constraint, by designing the covert communication strategies from three different perspectives: Alice's perspective, the jammer's perspective, and the global perspective. Our analysis reveals that the minimum jamming power should not always be zero in the probabilistic jamming strategy, which is different from that in the continuous jamming strategy presented in the literature. In addition, we prove that the minimum jamming power should be the same as Alice's covert transmit power, depending on the covertness and average jamming power constraints. Furthermore, our results show that the probabilistic jamming can outperform the continuous jamming in terms of achieving a higher covert throughput under the same covertness and average jamming power constraints.

eess.SP↗

arXiv2023

mmHawkeye: Passive UAV Detection with a COTS mmWave Radar

Small Unmanned Aerial Vehicles (UAVs) are becoming potential threats to security-sensitive areas and personal privacy. A UAV can shoot photos at height, but how to detect such an uninvited intruder is an open problem. This paper presents mmHawkeye, a passive approach for UAV detection with a COTS millimeter wave (mmWave) radar. mmHawkeye doesn't require prior knowledge of the type, motions, and flight trajectory of the UAV, while exploiting the signal feature induced by the UAV's periodic micro-motion (PMM) for long-range accurate detection. The design is therefore effective in dealing with low-SNR and uncertain reflected signals from the UAV. mmHawkeye can further track the UAV's position with dynamic programming and particle filtering, and identify it with a Long Short-Term Memory (LSTM) based detector. We implement mmHawkeye on a commercial mmWave radar and evaluate its performance under varied settings. The experimental results show that mmHawkeye has a detection accuracy of 95.8% and can realize detection at a range up to 80m.

cs.NI↗

arXiv2023

Anatomy of spin Hall effect in ferromagnetic metals

The spin Hall effect in nonmagnetic materials has been intensively studied and became one of the most crucial spin-charge conversion mechanism in spintronics. However, the spin Hall effect in ferromagnetic metals has been less investigated and remains unclear. In this work, we investigate the spin Hall effect in representative ferromagnetic alloy by using first-principles calculations. We first clarify the spin Hall effect into three different types including conventional (CSHE), spin anomalous (SAHE) and magnetic spin Hall effect (MSHE) and then calculate the corresponding spin Hall conductivity and spin Hall angle for (Fe, Co, Ni)Pt, NiFe and CoFe alloy. We find the above three spin Hall mechanisms do coexist in ferromagnetic metals. Particularly, for Pt-based ferromagnetic alloy, a sizable conventional and magnetic spin Hall angles comparable to that of Pt have been predicted. The remarkable unconventional spin Hall effect in ferromagnetic metal may enrich the spin-charge conversion phenomena. For instance, the spin current generated by remarkable MSHE with out-of-plane spin-polarization should be helpful for field-free switching of perpendicular magnetization through spin-orbit torque effect. This work may stimulate future studies on the spin Hall effect in ferromagnetic metals and pave their promising applications for spin-charge conversion devices in spintronics.

cond-mat.mtrl-sci↗

arXiv2023

A Survey of mmWave-based Human Sensing: Technology, Platform and Applications

With the rapid development of the Internet of Things (IoT) and the rise of 5G communication networks and automatic driving, millimeter wave (mmWave) sensing is emerging and starts impacting our life and workspace. mmWave sensing can sense humans and objects in a contactless way, providing fine-grained sensing ability. In the past few years, many mmWave sensing techniques have been proposed and applied in various human sensing applications (e.g., human localization, gesture recognition, and vital monitoring). We discover the need of a comprehensive survey to summarize the technology, platforms and applications of mmWave-based human sensing. In this survey, we first present the mmWave hardware platforms and some key techniques of mmWave sensing. We then provide a comprehensive review of existing mmWave-based human sensing works. Specifically, we divide existing works into four categories according to the sensing granularity: human tracking and localization, motion recognition, biometric measurement and human imaging. Finally, we discuss the potential research challenges and present future directions in this area.

cs.NI↗

arXiv2023

Optimal Sparse Sliced Inverse Regression via Random Projection

We propose a novel sparse sliced inverse regression method based on random projections in a large $p$ small $n$ setting. Embedded in a generalized eigenvalue framework, the proposed approach finally reduces to parallel execution of low-dimensional (generalized) eigenvalue decompositions, which facilitates high computational efficiency. Theoretically, we prove that this method achieves the minimax optimal rate of convergence under suitable assumptions. Furthermore, our algorithm involves a delicate reweighting scheme, which can significantly enhance the identifiability of the active set of covariates. Extensive numerical studies demonstrate high superiority of the proposed algorithm in comparison to competing methods.

stat.ME↗

arXiv2023

Interfacial Resonance States-Induced Negative Tunneling Magneto-resistance in Orthogonally-Magnetized CoFeB/MgO/CoFeB

Magnetic tunneling junctions (MTJs) are essential for non-volatile magneto-resistive random access memory (MRAM) applications. Here, we report the observation of a large negative tunneling magneto-resistance (TMR) in the CoFeB/MgO/CoFeB system with an orthogonally-magnetized configuration. Through the thickness modulation of the MgO barrier, the negative TMR component can be enhanced up to 20% under a negative voltage bias. Moreover, the tunnel anisotropic magneto-resistance measurements unveil that the negative TMR component likely arises from the interfacial resonance states (IRS) in the minority band of the bottom ferromagnetic layer. Complementary first principle calculations further quantify the IRS location and strength with respect to the Fermi level position. Our work not only confirm the vital role of IRS in the electrical transport of MTJ, but also provide valuable insights for the design of new-generation voltage-controlled MRAM and related spintronic applications.

physics.app-ph↗

arXiv2023

Properties of the optical component stars in X-ray binary systems

We collected a total of 4,058 X-ray binary stars, out of which 339 stars had three atmospheric parameters for optical companions from Gaia and LAMOST, while 264 stars had masses and radii of optical companions determined using stellar evolution models. We conducted a thorough discussion on the reliability of each parameter. The statistical analysis revealed a noticeable bimodal distribution in the mass, radius, and age of the optical components. Our findings led to the proposal of a new quantitative classification criterion for X-ray binary stars. In this classification, one type is categorized as high-mass, high-temperature, and young, while the other type is classified as low-mass, low-temperature, and old, corresponding to High-Mass X-Ray Binaries (HMXBs) and Low-Mass X-Ray Binaries (LMXBs), respectively. The dividing lines were established at 11,500 K, 1.7 $M_{\odot}$, and 0.14 Gyr. The classification results of the three parameters showed consistency with one another in 90\% of the cases and were in agreement with previous classifications in 80\% of the cases. We found that the parameters of the two types had well-defined boundaries and distinct patterns. Based on our findings, we suggest that temperature is the best parameter for classification. Therefore, we propose that X-ray binary stars should be classified into high-temperature X-ray binaries (HTXBs) and low-temperature X-ray binaries (LTXBs). We believe that this classification is more convenient in practice and aligns well with physics.

astro-ph.SR↗

arXiv2023

NeuralStagger: Accelerating Physics-constrained Neural PDE Solver with Spatial-temporal Decomposition

Neural networks have shown great potential in accelerating the solution of partial differential equations (PDEs). Recently, there has been a growing interest in introducing physics constraints into training neural PDE solvers to reduce the use of costly data and improve the generalization ability. However, these physics constraints, based on certain finite dimensional approximations over the function space, must resolve the smallest scaled physics to ensure the accuracy and stability of the simulation, resulting in high computational costs from large input, output, and neural networks. This paper proposes a general acceleration methodology called NeuralStagger by spatially and temporally decomposing the original learning tasks into several coarser-resolution subtasks. We define a coarse-resolution neural solver for each subtask, which requires fewer computational resources, and jointly train them with the vanilla physics-constrained loss by simply arranging their outputs to reconstruct the original solution. Due to the perfect parallelism between them, the solution is achieved as fast as a coarse-resolution neural solver. In addition, the trained solvers bring the flexibility of simulating with multiple levels of resolution. We demonstrate the successful application of NeuralStagger on 2D and 3D fluid dynamics simulations, which leads to an additional $10\sim100\times$ speed-up. Moreover, the experiment also shows that the learned model could be well used for optimal control.

cs.LG↗

arXiv2023

Learning Physics-Informed Neural Networks without Stacked Back-propagation

Physics-Informed Neural Network (PINN) has become a commonly used machine learning approach to solve partial differential equations (PDE). But, facing high-dimensional secondorder PDE problems, PINN will suffer from severe scalability issues since its loss includes second-order derivatives, the computational cost of which will grow along with the dimension during stacked back-propagation. In this work, we develop a novel approach that can significantly accelerate the training of Physics-Informed Neural Networks. In particular, we parameterize the PDE solution by the Gaussian smoothed model and show that, derived from Stein's Identity, the second-order derivatives can be efficiently calculated without back-propagation. We further discuss the model capacity and provide variance reduction methods to address key limitations in the derivative estimation. Experimental results show that our proposed method can achieve competitive error compared to standard PINN training but is significantly faster. Our code is released at https://github.com/LithiumDA/PINN-without-Stacked-BP.

cs.LG↗

arXiv2023

A Voigt laser operating on $^{87}$Rb 780 nm transition

We report the development of laser systems -- a "Voigt laser" -- using a Voigt anomalous dispersion optical filter as the frequency-selective element, working at the wavelength of 780 nm of $^{87}$Rb-D2 resonance line. Compared with Faraday anomalous dispersion optical filter, the Voigt anomalous dispersion optical filter can generate a stronger and more uniform magnetic field with a compact size of magnet, and obtains a transmission spectrum with narrower linewidth and more stable lineprofile. In this case, the frequency stability of the Voigt laser reaches 5$\times$10$^{-9}$ at the averaging time of 200 s, and the wavelength fluctuation of 8-hours free operation is $\pm$0.1 pm. Besides, the Voigt laser has greater immunity to diode current than the Faraday laser, with a wavelength fluctuation of $\pm$0.5 pm in the current range from 73 mA to 150 mA. Finally, the Voigt laser frequency can be controlled by the cell temperature of the Voigt optical filter, which is expected to achieve a frequency detuning of 20 GHz. Consequently, the Voigt laser, whose frequency could correspond to the atomic transition frequency by tuning the cell temperature, obtains good robustness to the current and temperature fluctuation of laser diode, and could realize a compact optical standard for precise measurement once stabilized by modulation transfer spectroscopy.

physics.optics↗

arXiv2022

Tremendous tunneling magnetoresistance effects based on van der Waals room-temperature ferromagnet Fe$_3$GaTe$_2$ with highly spin-polarized Fermi surfaces

Recently, van der Waals (vdW) magnetic heterostructures have received increasing research attention in spintronics. However, the lack of room-temperature magnetic order of vdW material has largely impedes its development in practical spintronics devices. Inspired by the recently discovered vdW ferromagnet Fe3GaTe2, which has been shown to have magnetic order above room temperature and sizable perpendicular magnetic anisotropy, we investigate the basic electronic structure and magnetic properties of Fe3GaTe2 as well as tunneling magnetoresistance effect in magnetic tunnel junctions (MTJs) with structure of Fe3GaTe2/Insulator/Fe3GaTe2 by using first-principles calculations. It is found that Fe3GaTe2 with highly spin-polarized Fermi surface ensures that such magnetic tunnel junctions may have prominent tunneling magnetoresistance effect at room temperature even comparable to existing conventional AlOx and MgO-based MTJs. Our results suggest that Fe3GaTe2-based MTJs may be the promising candidate for realizing long-waiting full magnetic vdW spintronic devices.

cond-mat.mtrl-sci↗

arXiv2022

Nested sampling statistical errors

Nested sampling (NS) is a popular algorithm for Bayesian computation. We investigate statistical errors in NS both analytically and numerically. We show two analytic results. First, we show that the leading terms in Skilling's expression using information theory match the leading terms in Keeton's expression from an analysis of moments. This approximate agreement was previously only known numerically and was somewhat mysterious. Second, we show that the uncertainty in single NS runs approximately equals the standard deviation in repeated NS runs. Whilst intuitive, this was previously taken for granted. We close by investigating our results and their assumptions in several numerical examples, including cases in which NS uncertainties increase without bound.

astro-ph.IM↗