arXiv subjects

Hui Zhang

Publications and source records attributed to Hui Zhang.

At least 325 records · Page 18Linked to original sources

arXiv2023

Machine-learning-informed parameter estimation improves the reliability of spinal cord diffusion MRI

Purpose: We address the challenge of inaccurate parameter estimation in diffusion MRI when the signal-to-noise ratio (SNR) is very low, as in the spinal cord. The accuracy of conventional maximum-likelihood estimation (MLE) depends highly on initialisation. Unfavourable choices could result in suboptimal parameter estimates. Current methods to address this issue, such as grid search (GS) can increase computation time substantially. Methods: We propose a machine learning (ML) informed MLE approach that combines conventional MLE with ML approaches synergistically. ML-based methods have been developed recently to improve the speed and precision of parameter estimation. However, they can generate high systematic bias in estimated parameters when SNR is low. In the proposed ML-MLE approach, an artificial neural network model is trained to provide sensible initialisation for MLE efficiently, with the final solution determined by MLE, avoiding biases typically affecting pure ML estimations. Results: Using parameter estimation of neurite orientation dispersion and density imaging as an example, simulation and in vivo experiments suggest that the ML-MLE method can reduce outlier estimates from conventional MLE in white matter voxels affected by CSF contamination. It also accelerates computation compared to GS-MLE. Conclusion: The ML-MLE method can improve the reliability of parameter estimation with reduced computation time compared to GS-MLE, making it a practical tool for diffusion dataset with low SNR.

physics.med-ph↗

arXiv2022

Mirror frameworks for relatively Lipschitz and monotone-like variational inequalities

Nonconvex-nonconcave saddle-point optimization in machine learning has triggered lots of research for studying non-monotone variational inequalities (VI). In this work, we introduce two mirror frameworks, called mirror extragradient method and mirror extrapolation method, for approximating solutions to relatively Lipschitz and monotone-like VIs. The former covers the well-known Nemirovski's mirror prox method and Nesterov's dual extrapolation method, and the recently proposed Bregman extragradient method; all of them can be reformulated into a scheme that is very similar to the original form of extragradient method. The latter includes the operator extrapolation method and the Bregman extrapolation method as its special cases. The proposed mirror frameworks allow us to present a unified and improved convergence analysis for all these existing methods under relative Lipschitzness and monotone-like conditions that may be the currently weakest assumptions.

math.OC↗

arXiv2022

Personalized local heating neutralizing individual, spatial and temporal thermo-physiological variances in extreme cold environments

In this paper, we investigate the feasibility, robustness and optimization of introducing personal comfort systems (PCS), apparatuses that promises in energy saving and comfort improvement, into a broader range of environments. We report a series of laboratory experiments systematically examining the effect of personalized heating in neutralizing individual, spatial and temporal variations of thermal demands. The experiments were conducted in an artificial climate chamber at -15 degC in order to simulate extreme cold environments. We developed a heating garment with 20 pieces of 20 * 20 cm2 heating cloth (grouped into 9 regions) comprehensively covering human body. Surface temperatures of the garment can be controlled independently, quickly (within 20 seconds), precisely (within 1 degC) and easily (through a tablet) up to 45 degC. Participants were instructed to adjust surface temperatures of each segment to their preferences, with their physiological, psychological and adjustment data collected. We found that active heating could significantly and stably improve thermal satisfaction. The overall TSV and TCV were improved 1.50 and 1.53 during the self-adjustment phase. Preferred heating surface temperatures for different segments varied widely. Further, even for the same segment, individual differences among participants were considerable. Such variances were observed through local heating powers, while unnoticeable among thermal perception votes. In other words, all these various differences could be neutralized given the flexibility in personalized adjustments. Our research reaffirms the paradigm of "adaptive thermal comfort" and will promote innovations on human-centric design for more efficient PCSs.

cs.HC↗

arXiv2022

A photonic chip-based machine learning approach for the prediction of molecular properties

Machine learning methods have revolutionized the discovery process of new molecules and materials. However, the intensive training process of neural networks for molecules with ever-increasing complexity has resulted in exponential growth in computation cost, leading to long simulation time and high energy consumption. Photonic chip technology offers an alternative platform for implementing neural networks with faster data processing and lower energy usage compared to digital computers. Photonics technology is naturally capable of implementing complex-valued neural networks at no additional hardware cost. Here, we demonstrate the capability of photonic neural networks for predicting the quantum mechanical properties of molecules. To the best of our knowledge, this work is the first to harness photonic technology for machine learning applications in computational chemistry and molecular sciences, such as drug discovery and materials design. We further show that multiple properties can be learned simultaneously in a photonic chip via a multi-task regression learning algorithm, which is also the first of its kind as well, as most previous works focus on implementing a network in the classification task.

cs.ET↗

arXiv2022

On proximal gradient mapping and its minimization in norm via potential function-based acceleration

The proximal gradient descent method, well-known for composite optimization, can be completely described by the concept of proximal gradient mapping. In this paper, we highlight our previous two discoveries of proximal gradient mapping--norm monotonicity and refined descent, with which we are able to extend the recently proposed potential function-based framework from gradient descent to proximal gradient descent.

math.OC↗

arXiv2022

Recycling of Perovskite Substrate

The use of water-soluble sacrificial layer of Sr$_3$Al$_2$O$_6$ has tremendously boosted the research on freestanding functional oxide thin films, especially thanks to its ultimate capability to produce high-quality epitaxial perovskite thin films. However, the costly single-crystalline substrates, e.g. SrTiO$_3$, were generally discarded after obtaining the freestanding thin films. Here, we demonstrate that the SrTiO$_3$ substrates can be recycled to fabricate La$_{0.7}$Sr$_{0.3}$MnO$_3$ films with nearly identical structural and electrical properties. After attaining freestanding thin films, the residues on SrTiO$_3$ can be removed by 80 \degree C hot water soaking and rinsing treatments. Consequently, the surface of SrTiO$_3$ reverted to its original step-and-terrace structure.

physics.app-ph↗

arXiv2022

Fast method and convergence analysis of fractional magnetohydrodynamic coupled flow and heat transfer model for generalized second-grade fluid

In this paper, we first establish a new fractional magnetohydrodynamic (MHD) coupled flow and heat transfer model for a generalized second-grade fluid. This coupled model consists of a fractional momentum equation and a heat conduction equation with a generalized form of Fourier law. The second-order fractional backward difference formula is applied to the temporal discretization and the Legendre spectral method is used for the spatial discretization. The fully discrete scheme is proved to be stable and convergent with an accuracy of $O(τ^2+N^{-r})$, where $τ$ is the time step size and $N$ is the polynomial degree. To reduce the memory requirements and computational cost, a fast method is developed, which is based on a globally uniform approximation of the trapezoidal rule for integrals on the real line. And the strict convergence of the numerical scheme with this fast method is proved. We present the results of several numerical experiments to verify the effectiveness of the proposed method. Finally, we simulate the unsteady fractional MHD flow and heat transfer of the generalized second-grade fluid through a porous medium. The effects of the relevant parameters on the velocity and temperature are presented and analyzed in detail.

math.NA↗

arXiv2022

Electronic properties of monolayer copper selenide with one-dimensional moiré patterns

Strain engineering is a vital way to manipulate the electronic properties of two-dimensional (2D) materials. As a typical representative of transition metal mono-chalcogenides (TMMs), a honeycomb CuSe monolayer features with one-dimensional (1D) moiré patterns owing to the uniaxial strain along one of three equivalent orientations of Cu(111) substrates. Here, by combining low-temperature scanning tunneling microscopy/spectroscopy (STM/S) experiments and density functional theory (DFT) calculations, we systematically investigate the electronic properties of the strained CuSe monolayer on the Cu(111) substrate. Our results show the semiconducting feature of CuSe monolayer with a band gap of 1.28 eV and the 1D periodical modulation of electronic properties by the 1D moiré patterns. Except for the uniaxially strained CuSe monolayer, we observed domain boundary and line defects in the CuSe monolayer, where the biaxial-strain and strain-free conditions can be investigated respectively. STS measurements for the three different strain regions show that the first peak in conduction band will move downward with the increasing strain. DFT calculations based on the three CuSe atomic models with different strain inside reproduced the peak movement. The present findings not only enrich the fundamental comprehension toward the influence of strain on electronic properties at 2D limit, but also offer the benchmark for the development of 2D semiconductor materials.

cond-mat.mtrl-sci↗

arXiv2022

Self-Regularized Prototypical Network for Few-Shot Semantic Segmentation

The deep CNNs in image semantic segmentation typically require a large number of densely-annotated images for training and have difficulties in generalizing to unseen object categories. Therefore, few-shot segmentation has been developed to perform segmentation with just a few annotated examples. In this work, we tackle the few-shot segmentation using a self-regularized prototypical network (SRPNet) based on prototype extraction for better utilization of the support information. The proposed SRPNet extracts class-specific prototype representations from support images and generates segmentation masks for query images by a distance metric - the fidelity. A direct yet effective prototype regularization on support set is proposed in SRPNet, in which the generated prototypes are evaluated and regularized on the support set itself. The extent to which the generated prototypes restore the support mask imposes an upper limit on performance. The performance on the query set should never exceed the upper limit no matter how complete the knowledge is generalized from support set to query set. With the specific prototype regularization, SRPNet fully exploits knowledge from the support and offers high-quality prototypes that are representative for each semantic class and meanwhile discriminative for different classes. The query performance is further improved by an iterative query inference (IQI) module that combines a set of regularized prototypes. Our proposed SRPNet achieves new state-of-art performance on 1-shot and 5-shot segmentation benchmarks.

cs.CV↗

arXiv2022

A sub-Neptune transiting the young field star HD 18599 at 40 pc

Transiting exoplanets orbiting young nearby stars are ideal laboratories for testing theories of planet formation and evolution. However, to date only a handful of stars with age <1 Gyr have been found to host transiting exoplanets. Here we present the discovery and validation of a sub-Neptune around HD 18599, a young (300 Myr), nearby (d=40 pc) K star. We validate the transiting planet candidate as a bona fide planet using data from the TESS, Spitzer, and Gaia missions, ground-based photometry from IRSF, LCO, PEST, and NGTS, speckle imaging from Gemini, and spectroscopy from CHIRON, NRES, FEROS, and Minerva-Australis. The planet has an orbital period of 4.13 d, and a radius of 2.7Rearth. The RV data yields a 3-sigma mass upper limit of 30.5Mearth which is explained by either a massive companion or the large observed jitter typical for a young star. The brightness of the host star (V~9 mag) makes it conducive to detailed characterization via Doppler mass measurement which will provide a rare view into the interior structure of young planets.

astro-ph.EP↗

arXiv2022

TOI 560 : Two Transiting Planets Orbiting a K Dwarf Validated with iSHELL, PFS and HIRES RVs

We validate the presence of a two-planet system orbiting the 0.15--1.4 Gyr K4 dwarf TOI 560 (HD 73583). The system consists of an inner moderately eccentric transiting mini-Neptune (TOI 560 b, $P = 6.3980661^{+0.0000095}_{-0.0000097}$ days, $e=0.294^{+0.13}_{-0.062}$, $M= 0.94^{+0.31}_{-0.23}M_{Nep}$) initially discovered in the Sector 8 \tess\ mission observations, and a transiting mini-Neptune (TOI 560 c, $P = 18.8805^{+0.0024}_{-0.0011}$ days, $M= 1.32^{+0.29}_{-0.32}M_{Nep}$) discovered in the Sector 34 observations, in a rare near-1:3 orbital resonance. We utilize photometric data from \tess\, \textit{Spitzer}, and ground-based follow-up observations to confirm the ephemerides and period of the transiting planets, vet false positive scenarios, and detect the photo-eccentric effect for TOI 560 b. We obtain follow-up spectroscopy and corresponding precise radial velocities (RVs) with the iSHELL spectrograph at the NASA Infrared Telescope Facility and the HIRES Spectrograph at Keck Observatory to validate the planetary nature of these signals, which we combine with published PFS RVs from Magellan Observatory. We detect the masses of both planets at $> 3-σ$ significance. We apply a Gaussian process (GP) model to the \tess\ light curves to place priors on a chromatic radial velocity GP model to constrain the stellar activity of the TOI 560 host star, and confirm a strong wavelength dependence for the stellar activity demonstrating the ability of NIR RVs in mitigating stellar activity for young K dwarfs. TOI 560 is a nearby moderately young multi-planet system with two planets suitable for atmospheric characterization with James Webb Space Telescope (JWST) and other upcoming missions. In particular, it will undergo six transit pairs separated by $<$6 hours before June 2027.

astro-ph.EP↗

arXiv2022

Searching a High-Performance Feature Extractor for Text Recognition Network

Feature extractor plays a critical role in text recognition (TR), but customizing its architecture is relatively less explored due to expensive manual tweaking. In this work, inspired by the success of neural architecture search (NAS), we propose to search for suitable feature extractors. We design a domain-specific search space by exploring principles for having good feature extractors. The space includes a 3D-structured space for the spatial model and a transformed-based space for the sequential model. As the space is huge and complexly structured, no existing NAS algorithms can be applied. We propose a two-stage algorithm to effectively search in the space. In the first stage, we cut the space into several blocks and progressively train each block with the help of an auxiliary head. We introduce the latency constraint into the second stage and search sub-network from the trained supernet via natural gradient descent. In experiments, a series of ablation studies are performed to better understand the designed space, search algorithm, and searched architectures. We also compare the proposed method with various state-of-the-art ones on both hand-written and scene TR tasks. Extensive results show that our approach can achieve better recognition performance with less latency.

cs.CV↗

arXiv2022

Optimality Analysis and Block Sparse Algorithm for Complex Compressed Sensing

Recently, many new challenges in Compressed Sensing (CS), such as block sparsity, arose. In this paper, we present a new algorithm for solving CS with block sparse constraints (BSC) in complex fields. Firstly, based on block sparsity characteristics, we propose a new model to deal with CS with BSC and analyze the properties of the functions involved in this model. We then present a new $τ$-stationary point and analyze corresponding first-order sufficient and necessary conditions. That ensures we to further develop a block Newton hard-thresholding pursuit (BNHTP) algorithm for efficiently solving CS with BSC. Finally, preliminary numerical experiments demonstrate that the BNHTP algorithm has superior performance in terms of recovery accuracy and calculation time when compared with the classical AMP algorithm.

cs.IT↗

arXiv2022

CSL: A Large-scale Chinese Scientific Literature Dataset

Scientific literature serves as a high-quality corpus, supporting a lot of Natural Language Processing (NLP) research. However, existing datasets are centered around the English language, which restricts the development of Chinese scientific NLP. In this work, we present CSL, a large-scale Chinese Scientific Literature dataset, which contains the titles, abstracts, keywords and academic fields of 396k papers. To our knowledge, CSL is the first scientific document dataset in Chinese. The CSL can serve as a Chinese corpus. Also, this semi-structured data is a natural annotation that can constitute many supervised NLP tasks. Based on CSL, we present a benchmark to evaluate the performance of models across scientific domain tasks, i.e., summarization, keyword generation and text classification. We analyze the behavior of existing text-to-text models on the evaluation tasks and reveal the challenges for Chinese scientific NLP tasks, which provides a valuable reference for future research. Data and code are available at https://github.com/ydli-ai/CSL

cs.CL↗

arXiv2022

Auto-Encoding Score Distribution Regression for Action Quality Assessment

The action quality assessment (AQA) of videos is a challenging vision task since the relation between videos and action scores is difficult to model. Thus, AQA has been widely studied in the literature. Traditionally, AQA is treated as a regression problem to learn the underlying mappings between videos and action scores. But previous methods ignored data uncertainty in AQA dataset. To address aleatoric uncertainty, we further develop a plug-and-play module Distribution Auto-Encoder (DAE). Specifically, it encodes videos into distributions and uses the reparameterization trick in variational auto-encoders (VAE) to sample scores, which establishes a more accurate mapping between videos and scores. Meanwhile, a likelihood loss is used to learn the uncertainty parameters. We plug our DAE approach into MUSDL and CoRe. Experimental results on public datasets demonstrate that our method achieves state-of-the-art on AQA-7, MTL-AQA, and JIGSAWS datasets. Our code is available at https://github.com/InfoX-SEU/DAE-AQA.

cs.CV↗

arXiv2022

Multiple semiclassical states for fractional Schrodinger equations with asymptotically linear nonlinearities

In this paper, we consider the singularly perturbed fractional Schrödinger equation \begin{equation*} ε^{2α}(-Δ)^αu+V(x)u=f(u),\quad x\in \mathbb{R}^N, \end{equation*} where $ε>0$ is a small parameter, $α\in(0,1)$, $(-Δ)^α$ is the fractional Laplacian operator of order $α$, $V$ possesses global minimum points, and $f$ is asymptotically linear at infinity. We investigate the relationship between the number of positive solutions and the topology of the set where the potential $V$ attains its global minimum. We also construct multiple concentrating solutions if $V$ has several strict global minimum points. In particular, some new tricks and the method of Nehari manifold dependent on a suitable restricted set are introduced to overcome the difficulty resulting from the appearance of asymptotically linear nonlinearity.

math.AP↗

arXiv2022

How can spherical CNNs benefit ML-based diffusion MRI parameter estimation?

This paper demonstrates spherical convolutional neural networks (S-CNN) offer distinct advantages over conventional fully-connected networks (FCN) at estimating scalar parameters of tissue microstructure from diffusion MRI (dMRI). Such microstructure parameters are valuable for identifying pathology and quantifying its extent. However, current clinical practice commonly acquires dMRI data consisting of only 6 diffusion weighted images (DWIs), limiting the accuracy and precision of estimated microstructure indices. Machine learning (ML) has been proposed to address this challenge. However, existing ML-based methods are not robust to differing dMRI gradient sampling schemes, nor are they rotation equivariant. Lack of robustness to sampling schemes requires a new network to be trained for each scheme, complicating the analysis of data from multiple sources. A possible consequence of the lack of rotational equivariance is that the training dataset must contain a diverse range of microstucture orientations. Here, we show spherical CNNs represent a compelling alternative that is robust to new sampling schemes as well as offering rotational equivariance. We show the latter can be leveraged to decrease the number of training datapoints required.

eess.IV↗

arXiv2022

Volume of hyperintense inflammation (VHI): a deep learning-enabled quantitative imaging biomarker of inflammation load in spondyloarthritis

Short inversion time inversion recovery (STIR) MRI is widely used in clinical practice to identify and quantify inflammation in axial spondyloarthritis. However, assessment of STIR images is limited by the need for qualitative evaluation, which depends on observer experience and expertise, creating substantial variability in inflammation assessments. To address this problem, we developed a deep learning-enabled, semiautomated workflow for segmentation of inflammatory lesions, whereby an initial segmentation is generated automatically and a radiologist then 'cleans' the segmentation by removing extraneous segmented voxels. The final cleaned segmentation defines the volume of hyperintense inflammation (VHI), which we propose as a quantitative imaging biomarker of inflammation load in spondyloarthritis. The data, code and models used in the study are available at https://github.com/c-hepburn/Bone_MRI.

eess.IV↗