Searcharxiv⌕ Search

arXiv subjects

Guang Yang

Publications and source records attributed to Guang Yang.

At least 343 records · Page 19Linked to original sources
arXiv2023

ChatAgri: Exploring Potentials of ChatGPT on Cross-linguistic Agricultural Text Classification

In the era of sustainable smart agriculture, a massive amount of agricultural news text is being posted on the Internet, in which massive agricultural knowledge has been accumulated. In this context, it is urgent to explore effective text classification techniques for users to access the required agricultural knowledge with high efficiency. Mainstream deep learning approaches employing fine-tuning strategies on pre-trained language models (PLMs), have demonstrated remarkable performance gains over the past few years. Nonetheless, these methods still face many drawbacks that are complex to solve, including: 1. Limited agricultural training data due to the expensive-cost and labour-intensive annotation; 2. Poor domain transferability, especially of cross-linguistic ability; 3. Complex and expensive large models deployment.Inspired by the extraordinary success brought by the recent ChatGPT (e.g. GPT-3.5, GPT-4), in this work, we systematically investigate and explore the capability and utilization of ChatGPT applying to the agricultural informatization field. ....(shown in article).... Code has been released on Github https://github.com/albert-jin/agricultural_textual_classification_ChatGPT.

cs.CL↗
arXiv2023

Extremely red galaxies at $z=5-9$ with MIRI and NIRSpec: dusty galaxies or obscured AGNs?

We study a new population of extremely red objects (EROs) recently discovered by JWST based on their NIRCam colors F277W$-$F444W $>1.5$ mag. We find 37 EROs in the CEERS field with F444W $<28$ mag and photometric redshifts between $5 7$ by up to a factor $\sim$60. Similarly, if they are OSOs with luminosities in the L$_{\rm bol}>10^{46-47}$ erg s$^{-1}$ range, their number would exceed that of bright blue QSOs by more than two orders of magnitude. Additional photometry at mid-IR wavelengths will reveal the true nature of the red continuum emission in these EROs and will place this puzzling population in the right context of galaxy evolution.

astro-ph.GA↗
arXiv2023

Quantum oscillations in field-induced correlated insulators of a moiré superlattice

We report an observation of quantum oscillations (QOs) in the correlated insulators with valley anisotropy of twisted double bilayer graphene (TDBG). The anomalous QOs are best captured in the magneto resistivity oscillations of the insulators at v = -2, with a period of 1/B and an oscillation amplitude as high as 150 kΩ. The QOs can survive up to ~10 K, and above 12 K, the insulating behaviors are dominant. The QOs of the insulator are strongly D dependent: the carrier density extracted from the 1/B periodicity decreases almost linearly with D from -0.7 to -1.1 V/nm, suggesting a reduced Fermi surface; the effective mass from Lifshitz-Kosevich analysis depends nonlinearly on D, reaching a minimal value of 0.1 me at D = ~ -1.0 V/nm. Similar observations of QOs are also found at v = 2, as well as in other devices without graphite gate. We interpret the D sensitive QOs of the correlated insulators in the picture of band inversion. By reconstructing an inverted band model with the measured effective mass and Fermi surface, the density of state at the gap, calculated from thermal broadened Landau levels, agrees qualitatively with the observed QOs in the insulators. While more theoretical understandings are needed in the future to fully account for the anomalous QOs in this moiré system, our study suggests that TDBG is an excellent platform to discover exotic phases where correlation and topology are at play.

cond-mat.mes-hall↗
arXiv2023

Two massive, compact, and dust-obscured candidate $z\sim 8$ galaxies discovered by JWST

We present a search for extremely red, dust-obscured, $z>7$ galaxies with $\textit{JWST}$/NIRCam+MIRI imaging over the first 20 arcmin$^2$ of publicly-available Cycle 1 data from the COSMOS-Web, CEERS, and PRIMER surveys. Based on their red color in F277W$-$F444W ($\sim 2.5$ mag) and detection in MIRI/F770W ($\sim 25$ mag), we identify two galaxies$\unicode{x2014}$COS-z8M1 and CEERS-z7M1$\unicode{x2014}$which have best-fit photometric redshifts of $z=8.5^{+0.3}_{-0.4}$ and $z=7.6^{+0.1}_{-0.1}$, respectively. We perform SED fitting with a variety of codes (including BAGPIPES, PROSPECTOR, BEAGLE, and CIGALE), and find a $>95\%$ probability that these indeed lie at $z>7$. Both sources are compact ($R_{\rm eff} \lesssim 200$ pc), highly obscured ($A_V \sim 1.5$$\unicode{x2013}$$2.5$), and, at our best-fit redshift estimates, likely have strong [OIII]+H$β$ emission contributing to their $4.4\,μ$m photometry. We estimate stellar masses of $\sim 10^{10}~M_\odot$ for both sources; by virtue of detection in MIRI at $7.7\,μ$m, these measurements are robust to the inclusion of bright emission lines, for example, from an AGN. We identify a marginal (2.9$σ$) ALMA detection at 2 mm within $0.5''$ of COS-z8M1, which if real, would suggest a remarkably high IR luminosity of $\sim 10^{12} L_\odot$. These two galaxies, if confirmed at $z\sim 8$, would be extreme in their stellar and dust masses, and may be representative of a substantial population of modestly dust-obscured galaxies at cosmic dawn.

astro-ph.GA↗
arXiv2023

CLEAR: High-Ionization [Ne V] $λ$3426 Emission-line Galaxies at $1.4 <z< 2.3$

We analyze a sample of 25 [Ne V] $λ$3426 emission-line galaxies at $1.4<z<2.3$ using Hubble Space Telescope/Wide Field Camera 3 G102 and G141 grism observations from the CANDELS Lyman-$α$ Emission at Reionization (CLEAR) survey. [Ne V] emission probes extremely energetic photoionization (97.11-126.21 eV), and is often attributed to energetic radiation from active galactic nuclei (AGN), shocks from supernova, or an otherwise very hard ionizing spectrum from the stellar continuum. In this work, we use [Ne V] in conjunction with other rest-frame UV/optical emission lines ([O II] $λλ$3726,3729, [Ne III] $λ$3869, H$β$, [O III] $λλ$4959,5007, H$α$+[N II] $λλ$6548,6583, [S II] $λλ$6716,6731), deep (2--7 Ms) X-ray observations (from Chandra), and mid-infrared imaging (from Spitzer) to study the origin of this emission and to place constraints on the nature of the ionizing engine. The majority of the [Ne V]-detected galaxies have properties consistent with ionization from AGN. However, for our [Ne V]-selected sample, the X-ray luminosities are consistent with local ($z\lesssim 0.1$) X-ray-selected Seyferts, but the [Ne V] luminosities are more consistent with those from $z\sim 1$ X-ray-selected QSOs. The excess [Ne V] emission requires either reduced hard X-rays, or a $\sim$0.1 keV excess. We discuss possible origins of the apparent [Ne V] excess, which could be related to the ``soft (X-ray) excess'' observed in some QSOs and Seyferts, and/or be a consequence of a complex/anisotropic geometry for the narrow line region, combined with absorption from a warm, relativistic wind ejected from the accretion disk. We also consider implications for future studies of extreme high-ionization systems in the epoch of reionization ($z \gtrsim 6$) with JWST.

astro-ph.GA↗
arXiv2023

Deep Learning-based Diffusion Tensor Cardiac Magnetic Resonance Reconstruction: A Comparison Study

In vivo cardiac diffusion tensor imaging (cDTI) is a promising Magnetic Resonance Imaging (MRI) technique for evaluating the micro-structure of myocardial tissue in the living heart, providing insights into cardiac function and enabling the development of innovative therapeutic strategies. However, the integration of cDTI into routine clinical practice is challenging due to the technical obstacles involved in the acquisition, such as low signal-to-noise ratio and long scanning times. In this paper, we investigate and implement three different types of deep learning-based MRI reconstruction models for cDTI reconstruction. We evaluate the performance of these models based on reconstruction quality assessment and diffusion tensor parameter assessment. Our results indicate that the models we discussed in this study can be applied for clinical use at an acceleration factor (AF) of $\times 2$ and $\times 4$, with the D5C5 model showing superior fidelity for reconstruction and the SwinMR model providing higher perceptual scores. There is no statistical difference with the reference for all diffusion tensor parameters at AF $\times 2$ or most DT parameters at AF $\times 4$, and the quality of most diffusion tensor parameter maps are visually acceptable. SwinMR is recommended as the optimal approach for reconstruction at AF $\times 2$ and AF $\times 4$. However, we believed the models discussed in this studies are not prepared for clinical use at a higher AF. At AF $\times 8$, the performance of all models discussed remains limited, with only half of the diffusion tensor parameters being recovered to a level with no statistical difference from the reference. Some diffusion tensor parameter maps even provide wrong and misleading information.

physics.med-ph↗
arXiv2023

CEERS Key Paper V: A triality on the nature of HST-dark galaxies

The new capabilities that JWST offers in the near- and mid-infrared (IR) are used to investigate in unprecedented detail the nature of optical/near-IR faint, mid-IR bright sources, HST-dark galaxies among them. We gather JWST data from the CEERS survey in the EGS, jointly with HST data, and analyze spatially resolved optical-to-mid-IR spectral energy distributions (SEDs) to estimate both photometric redshifts in 2 dimensions and stellar populations properties in a pixel-by-pixel basis. We select 138 galaxies with F150W-F356W>1.5 mag, F356W<27.5 mag. The nature of these sources is threefold: (1) 71% are dusty star-forming galaxies at 2 100 Gyr^-1); (2) 18% are quiescent/dormant (i.e., subject to reignition and rejuvenation) galaxies at 3<z<5, masses log M/M_sun~10 and post-starburst stellar mass-weighted ages (0.5-1 Gyr); and (3) 11% are strong young starbursts with indications of high-EW emission lines (typically, [OIII]+Hbeta) at 6<z<7 and log M/M_sun~9.5. The sample is dominated by disk-like galaxies with a remarkable compactness for XELG-z6 (effective radii smaller than 0.4 kpc). Large attenuations in SFGs, 2<A(V)<5 mag, are found within 1.5 times the effective radius, approximately 2 kpc, while QGs present A(V)~0.2 mag. Our SED-fitting technique reproduces the expected dust emission luminosities of IR-bright and sub-millimeter galaxies. This study implies high levels of star formation activity between z~20 and z~10, where virtually 100% of our galaxies had already formed 10^8 M_sun of their stellar content, 60% of them had assembled 10^9 M_sun, and 10% up to 10^10 M_sun (in situ or ex situ). (abridged)

astro-ph.GA↗
arXiv2023

Spontaneous rotational symmetry breaking in KTaO$_3$ heterointerface superconductors

Broken symmetries play a fundamental role in superconductivity and influence many of its properties in a profound way. Understanding these symmetry breaking states is essential to elucidate the various exotic quantum behaviors in non-trivial superconductors. Here, we report an experimental observation of spontaneous rotational symmetry breaking of superconductivity at the heterointerface of amorphous (a)-YAlO$_3$/KTaO$_3$(111) with a superconducting transition temperature of 1.86 K. Both the magnetoresistance and superconducting critical field in an in-plane field manifest striking twofold symmetric oscillations deep inside the superconducting state, whereas the anisotropy vanishes in the normal state, demonstrating that it is an intrinsic property of the superconducting phase. We attribute this behavior to the mixed-parity superconducting state, which is an admixture of \emph{s}-wave and \emph{p}-wave pairing components induced by strong spin-orbit coupling inherent to inversion symmetry breaking at the heterointerface of a-YAlO$_3$/KTaO$_3$. Our work suggests an unconventional nature of the underlying pairing interaction in the KTaO$_3$ heterointerface superconductors, and brings a new broad of perspective on understanding non-trivial superconducting properties at the artificial heterointerfaces.

cond-mat.supr-con↗
arXiv2023

CEERS: Spatially Resolved UV and mid-IR Star Formation in Galaxies at 0.2 < z < 2.5: The Picture from the Hubble and James Webb Space Telescopes

We present the mid-IR (MIR) morphologies for 64 star-forming galaxies at $0.2 10^{9}~M_\odot}$ using JWST MIRI observations from the Cosmic Evolution Early Release Science survey (CEERS). The MIRI bands span the MIR (7.7--21~$μ$m), enabling us to measure the effective radii ($R_{\rm{eff}}$) and Sérsic indexes of these SFGs at rest-frame 6.2 and 7.7 $μ$m, which contains strong emission from Polycyclic aromatic hydrocarbon (PAH) features, a well-established tracer of star formation in galaxies. We define a ``PAH-band'' as the MIRI bandpass that contains these features at the redshift of the galaxy. We then compare the galaxy morphologies in the PAH-bands to those in rest-frame Near-UV (NUV) using HST ACS/F435W or ACS/F606W and optical/near-IR using HST WFC3/F160W imaging from UVCANDELS and CANDELS, where the NUV-band and F160W trace the profile of (unobscured) massive stars and the stellar continuum, respectively. The $R_{\rm{eff}}$ of galaxies in the PAH-band are slightly smaller ($\sim$10\%) than those in F160W for galaxies with $\rm{M_*\gtrsim10^{9.5}~M_\odot}$ at $z\leq1.2$, but the PAH-band and F160W have a similar fractions of light within 1 kpc. In contrast, the $R_{\rm{eff}}$ of galaxies in the NUV-band are larger, with lower fractions of light within 1 kpc compared to F160W for galaxies at $z\leq1.2$. Using the MIRI data to estimate the $\rm{SFR_{\rm{IR}}}$ surface density, we find the correlation between the $\rm{SFR_{\rm{IR}}}$ surface density and stellar mass has a steeper slope than that of the $\rm{SFR_{\rm{UV}}}$ surface density and stellar mass, suggesting more massive galaxies having increasing amounts of obscured fraction of star formation in their inner regions. This paper demonstrates how the high-angular resolution data from JWST/MIRI can reveal new information about the morphology of obscured-star formation.

astro-ph.GA↗
arXiv2023

Is Autoencoder Truly Applicable for 3D CT Super-Resolution?

Featured by a bottleneck structure, autoencoder (AE) and its variants have been largely applied in various medical image analysis tasks, such as segmentation, reconstruction and de-noising. Despite of their promising performances in aforementioned tasks, in this paper, we claim that AE models are not applicable to single image super-resolution (SISR) for 3D CT data. Our hypothesis is that the bottleneck architecture that resizes feature maps in AE models degrades the details of input images, thus can sabotage the performance of super-resolution. Although U-Net proposed skip connections that merge information from different levels, we claim that the degrading impact of feature resizing operations could hardly be removed by skip connections. By conducting large-scale ablation experiments and comparing the performance between models with and without the bottleneck design on a public CT lung dataset , we have discovered that AE models, including U-Net, have failed to achieve a compatible SISR result ($p<0.05$ by Student's t-test) compared to the baseline model. Our work is the first comparative study investigating the suitability of AE architecture for 3D CT SISR tasks and brings a rationale for researchers to re-think the choice of model architectures especially for 3D CT SISR tasks. The full implementation and trained models can be found at: https://github.com/Roldbach/Autoencoder-3D-CT-SISR

eess.IV↗
arXiv2023

Multi-dimensional Mean-field Type Backward Stochastic Differential Equations with Diagonally Quadratic Generators

In this paper, we study the multi-dimensional backward stochastic differential equations (BSDEs) whose generator depends also on the mean of both variables. When the generator is diagonally quadratic, we prove that the BSDE admits a unique local solution with a fixed point argument. When the generator has a logarithmic growth of the off-diagonal elements (i.e., for each $i$, the $i$-th component of the generator has a logarithmic growth of the $j$-th row $z^j$ of the variable $z$ for each $j \neq i$), we give a new apriori estimate and obtain the existence and uniqueness of the global solution.

math.PR↗
arXiv2023

CEERS Key Paper IV: Galaxies at $4 < z < 9$ are Bluer than They Appear -- Characterizing Galaxy Stellar Populations from Rest-Frame $\sim 1$ micron Imaging

We present results from the Cosmic Evolution Early Release Survey (CEERS) on the stellar-population parameters for 28 galaxies with redshifts $4<z<9$ using imaging data from the James Webb Space Telescope (JWST) Mid-Infrared Instrument (MIRI) combined with data from the Hubble Space Telescope and the Spitzer Space Telescope. The JWST/MIRI 5.6 and 7.7 $μ$m data extend the coverage of the rest-frame spectral-energy distribution (SED) to nearly 1 micron for galaxies in this redshift range. By modeling the galaxies' SEDs the MIRI data show that the galaxies have, on average, rest-frame UV (1600 Å) $-$ $I$-band colors 0.4 mag bluer than derived when using photometry that lacks MIRI. Therefore, the galaxies have lower (stellar)-mass-to-light ratios. The MIRI data reduce the stellar masses by $\langle Δ\log M_\ast\rangle=0.25$ dex at $4<z<6$ (a factor of 1.8) and 0.37 dex at $6<z<9$ (a factor of 2.3). This also reduces the star-formation rates (SFRs) by $\langle Δ\log\mathrm{SFR} \rangle=0.14$ dex at $4<z<6$ and 0.27 dex at $6<z<9$. The MIRI data also improve constraints on the allowable stellar mass formed in early star-formation. We model this using a star-formation history that includes both a "burst' at $z_f=100$ and a slowly varying ("delayed-$τ$") model. The MIRI data reduce the allowable stellar mass by 0.6 dex at $4<z< 6$ and by $\approx$1 dex at $6<z<9$. Applying these results globally, this reduces the cosmic stellar-mass density by an order of magnitude in the early universe ($z\approx9$). Therefore, observations of rest-frame $\gtrsim$1 $μ$m are paramount for constraining the stellar-mass build-up in galaxies at very high-redshifts.

astro-ph.GA↗
arXiv2023

Less is More: Unsupervised Mask-guided Annotated CT Image Synthesis with Minimum Manual Segmentations

As a pragmatic data augmentation tool, data synthesis has generally returned dividends in performance for deep learning based medical image analysis. However, generating corresponding segmentation masks for synthetic medical images is laborious and subjective. To obtain paired synthetic medical images and segmentations, conditional generative models that use segmentation masks as synthesis conditions were proposed. However, these segmentation mask-conditioned generative models still relied on large, varied, and labeled training datasets, and they could only provide limited constraints on human anatomical structures, leading to unrealistic image features. Moreover, the invariant pixel-level conditions could reduce the variety of synthetic lesions and thus reduce the efficacy of data augmentation. To address these issues, in this work, we propose a novel strategy for medical image synthesis, namely Unsupervised Mask (UM)-guided synthesis, to obtain both synthetic images and segmentations using limited manual segmentation labels. We first develop a superpixel based algorithm to generate unsupervised structural guidance and then design a conditional generative model to synthesize images and annotations simultaneously from those unsupervised masks in a semi-supervised multi-task setting. In addition, we devise a multi-scale multi-task Fréchet Inception Distance (MM-FID) and multi-scale multi-task standard deviation (MM-STD) to harness both fidelity and variety evaluations of synthetic CT images. With multiple analyses on different scales, we could produce stable image quality measurements with high reproducibility. Compared with the segmentation mask guided synthesis, our UM-guided synthesis provided high-quality synthetic images with significantly higher fidelity, variety, and utility ($p<0.05$ by Wilcoxon Signed Ranked test).

eess.IV↗
arXiv2023

Diff-UNet: A Diffusion Embedded Network for Volumetric Segmentation

In recent years, Denoising Diffusion Models have demonstrated remarkable success in generating semantically valuable pixel-wise representations for image generative modeling. In this study, we propose a novel end-to-end framework, called Diff-UNet, for medical volumetric segmentation. Our approach integrates the diffusion model into a standard U-shaped architecture to extract semantic information from the input volume effectively, resulting in excellent pixel-level representations for medical volumetric segmentation. To enhance the robustness of the diffusion model's prediction results, we also introduce a Step-Uncertainty based Fusion (SUF) module during inference to combine the outputs of the diffusion models at each step. We evaluate our method on three datasets, including multimodal brain tumors in MRI, liver tumors, and multi-organ CT volumes, and demonstrate that Diff-UNet outperforms other state-of-the-art methods significantly. Our experimental results also indicate the universality and effectiveness of the proposed model. The proposed framework has the potential to facilitate the accurate diagnosis and treatment of medical conditions by enabling more precise segmentation of anatomical structures. The codes of Diff-UNet are available at https://github.com/ge-xing/Diff-UNet

eess.IV↗
arXiv2023

Chaos processes as rough paths

In this article we investigate the rough paths structure of a process $X_t$ living in a fixed Wiener chaos. Specifically, we formulate various types of rough lifts of $X_t$ and study their properties. As application, we study the integrabilities of quantities related to rough differential equations driven by $X_t$.

math.PR↗
arXiv2023

Quantifying quantum coherence of multiple-charge states in tunable Josephson junctions

Coherence and tunneling play central roles in quantum phenomena. In a tunneling event, the time that a particle spends inside the barrier has been fiercely debated. This problem becomes more complex when tunneling repeatedly occurs back and forth, and when involving many particles. Here we report the measurement of the coherence time of various charge states tunneling in a nanowire-based tunable Josephson junction; including single charges, multiple charges, and Cooper pairs. We studied all the charge tunneling processes using Landau-Zener-Stückelberg-Majorana (LZSM) interferometry, and observed high-quality interference patterns. In particular, the coherence time of the charge states was extracted from the interference fringes in Fourier space. In addition, our measurements show the break-up of Cooper pairs, from a macroscopic quantum coherent state to individual particle states. Besides the fundamental research interest, our results also establish LZSM interferometry as a powerful technique to explore the coherence time of charges in hybrid devices.

cond-mat.mes-hall↗
arXiv2023

A residual dense vision transformer for medical image super-resolution with segmentation-based perceptual loss fine-tuning

Super-resolution plays an essential role in medical imaging because it provides an alternative way to achieve high spatial resolutions and image quality with no extra acquisition costs. In the past few decades, the rapid development of deep neural networks has promoted super-resolution performance with novel network architectures, loss functions and evaluation metrics. Specifically, vision transformers dominate a broad range of computer vision tasks, but challenges still exist when applying them to low-level medical image processing tasks. This paper proposes an efficient vision transformer with residual dense connections and local feature fusion to achieve efficient single-image super-resolution (SISR) of medical modalities. Moreover, we implement a general-purpose perceptual loss with manual control for image quality improvements of desired aspects by incorporating prior knowledge of medical image segmentation. Compared with state-of-the-art methods on four public medical image datasets, the proposed method achieves the best PSNR scores of 6 modalities among seven modalities. It leads to an average improvement of $+0.09$ dB PSNR with only 38\% parameters of SwinIR. On the other hand, the segmentation-based perceptual loss increases $+0.14$ dB PSNR on average for SOTA methods, including CNNs and vision transformers. Additionally, we conduct comprehensive ablation studies to discuss potential factors for the superior performance of vision transformers over CNNs and the impacts of network and loss function components. The code will be released on GitHub with the paper published.

eess.IV↗
arXiv2023

Gate-tunable negative differential conductance in hybrid semiconductor-superconductor devices

Negative differential conductance (NDC) manifests as a significant characteristic of various underlying physics and transport processes in hybrid superconducting devices. In this work, we report the observation of gate-tunable NDC outside the superconducting energy gap on two types of hybrid semiconductor-superconductor devices, i.e., normal metal-superconducting nanowire-normal metal and normal metal-superconducting nanowire-superconductor devices. Specifically, we study the dependence of the NDCs on back-gate voltage and magnetic field. When the back-gate voltage decreases, these NDCs weaken and evolve into positive differential conductance dips; and meanwhile they move away from the superconducting gap towards high bias voltage, and disappear eventually. In addition, with the increase of magnetic field, the NDCs/dips follow the evolution of the superconducting gap, and disappear when the gap closes. We interpret these observations and reach a good agreement by combining the Blonder-Tinkham-Klapwijk (BTK) model and the critical supercurrent effect in the nanowire, which we call the BTK-supercurrent model. Our results provide an in-depth understanding of the tunneling transport in hybrid semiconductor-superconductor devices.

cond-mat.mes-hall↗