arXiv subjects

Ruihan Zhang

Publications and source records attributed to Ruihan Zhang.

34 records · Page 2Linked to original sources

arXiv2024

How Does Bayes Error Limit Probabilistic Robust Accuracy

Adversarial examples pose a security threat to many critical systems built on neural networks. Given that deterministic robustness often comes with significantly reduced accuracy, probabilistic robustness (i.e., the probability of having the same label with a vicinity is $\ge 1-κ$) has been proposed as a promising way of achieving robustness whilst maintaining accuracy. However, existing training methods for probabilistic robustness still experience non-trivial accuracy loss. It is unclear whether there is an upper bound on the accuracy when optimising towards probabilistic robustness, and whether there is a certain relationship between $κ$ and this bound. This work studies these problems from a Bayes error perspective. We find that while Bayes uncertainty does affect probabilistic robustness, its impact is smaller than that on deterministic robustness. This reduced Bayes uncertainty allows a higher upper bound on probabilistic robust accuracy than that on deterministic robust accuracy. Further, we prove that with optimal probabilistic robustness, each probabilistically robust input is also deterministically robust in a smaller vicinity. We also show that voting within the vicinity always improves probabilistic robust accuracy and the upper bound of probabilistic robust accuracy monotonically increases as $κ$ grows. Our empirical findings also align with our results.

cs.LG↗

arXiv2024

Unconventional phonon spectra and obstructed edge phonon modes

Based on the elementary band representations (EBR), some topologically trivial materials are classified as unconventional ones (obstructed atomic limit), where the EBR decomposition of electronic states is not consistent with the atomic valence-electron band representations. In the work, we identify that the unconventional nature can also exist in phonon spectra, where the EBR decomposition of the phonon modes is not consistent with atomic vibration band representations (ABR). The unconventionality has two types: type I is on an empty site; type II is on an atom site with non-atomic vibration orbitals. Our detailed calculations show that black phosphorus (BP) and 1H-MoSe2 have unconventional both phonon spectra and electronic band structures. The BP has the type-I unconventional phonon spectrum, while 1H-MoSe2 has the type-II one. The obstructed phonon modes are obtained for two types of unconventional phonon spectra.

cond-mat.mtrl-sci↗

arXiv2024

Superconductivity in unconventional metals

Based on first-principles calculations, we demonstrate that 1H/2H-phase transition metal dichalcogenides MX2 (M=Nb,Ta; X=S,Se,Te) are unconventional metals, which have an empty-site band of $A_1'@1e$ elementary band representation at the Fermi level. The computed phonon dispersions indicate the stability of the system at high temperatures, while the presence of the soft phonon mode suggests a phase transition to the charge density wave state at low temperatures. Based on the Bardeen-Cooper-Schrieffer theory and computed electron-phonon coupling, our calculations show that the superconductivity (SC) in NbSe2 is mainly attributed to the soft phonon mode due to the half filling of the empty-site band. Accordingly, the SC has been predicted in unconventional metals TaNS monolayer and 2H-TaN2 bulk with computed $T_C=$ 10 K and 26 K respectively. These results demonstrate that the unconventional metals with partial filling of the empty-site band offer an attractive platform to search for superconductors.

cond-mat.mtrl-sci↗

arXiv2023

A Possible Correlation between Metallicity and Near-IR Color for Late-M and L Dwarfs

We examine the relationship between metallicity and $J-K$ color for 64 benchmark late-M and L dwarfs, all of which are wide companions to higher mass stars, and 6 of which are new discoveries. We assess the correlation between the $Δ(J-K)$ color anomaly (the difference of an object's $J-K$ color with the median color for field objects of the same spectral type) and the host star metallicity to investigate how metallicity affects ultracool photospheres. Using Spearman's rank correlation test and Student's t test, the late-M dwarf (L dwarf) sample's $Δ(J-K)$ and metallicity show a positive correlation with 95\% (90\%) confidence level. A linear fit to color anomaly as a function of metallicity finds a slope of $0.17\pm0.07$ for the late-M dwarfs and a slope of $0.20^{+0.07}_{-0.08}$ for the L dwarfs. We also computed the $Δ(J-K)$ versus metallicity relationship predicted by multi-metallicity model spectra generated using Drift-Phoenix. The modeled late-M dwarfs show a slope of 0.202$\pm$0.03, which is close to our observational results, but the modeled L dwarfs show a slope of 0.493$\pm$0.02, steeper than our observational results. Both our empirical results and the models indicate that more metal-rich objects should appear redder photometrically. We speculate that higher metallicity drives more condensate formation in these atmospheres, thus making these ultracool dwarfs appear redder.

astro-ph.SR↗

arXiv2023

Two elementary band representation model, Fermi surface nesting, and surface topological superconductivity in $A$V$_{3}$Sb$_ {5}$ ($A = \text{K, Rb, Cs}$)

The recently discovered vanadium-based Kagome metals $A$V$_{3}$Sb$_{5}$ ($A = \text{K, Rb, Cs}$) are of great interest with the interplay of charge density wave (CDW) order, band topology and superconductivity. In this paper, by identifying elementary band representations (EBRs), we construct a two-EBR graphene-Kagome model to capture the two low-energy van-Hove-singularity dispersions and, more importantly, the nontrivial band topology in these Kagome metals. This model consists of $A_g@3g$ (V-$d_{x^2-y^2/z^2}$, Kagome sites) and $A_2''@2d$ EBRs (Sb1-$p_z$, honeycomb sites). We have investigated the Fermi surface instability by calculating the electronic susceptibility $χ(\mathbf{q})$. Prominent Fermi-surface nesting peaks are obtained at three L points, where the $z$ component of the nesting vector shows intimate relationship with the anticrossing point along M--L. The nesting peaks at L are consistent with the $2\times 2\times 2$ CDW reconstruction in these compounds. In addition, the sublattice-resolved bare susceptibility is calculated and similar sharp peaks are observed at the L points, indicating a strong antiferromagnetic fluctuation. Assuming a bulk $s$-wave superconducting pairing, helical surface states and nontrivial superconducting gap are obtained on the (001) surface. In analogous to FeTe$_{1-x}$Se$_{x}$ superconductor, our results establish another material realization of a stoichiometric superconductor with nontrivial band topology, providing a promising platform for studying exotic Majorana physics in condensed matter

cond-mat.mtrl-sci↗

arXiv2023

Towards Certified Probabilistic Robustness with High Accuracy

Adversarial examples pose a security threat to many critical systems built on neural networks (such as face recognition systems, and self-driving cars). While many methods have been proposed to build robust models, how to build certifiably robust yet accurate neural network models remains an open problem. For example, adversarial training improves empirical robustness, but they do not provide certification of the model's robustness. On the other hand, certified training provides certified robustness but at the cost of a significant accuracy drop. In this work, we propose a novel approach that aims to achieve both high accuracy and certified probabilistic robustness. Our method has two parts, i.e., a probabilistic robust training method with an additional goal of minimizing variance in terms of divergence and a runtime inference method for certified probabilistic robustness of the prediction. The latter enables efficient certification of the model's probabilistic robustness at runtime with statistical guarantees. This is supported by our training objective, which minimizes the variance of the model's predictions in a given vicinity, derived from a general definition of model robustness. Our approach works for a variety of perturbations and is reasonably efficient. Our experiments on multiple models trained on different datasets demonstrate that our approach significantly outperforms existing approaches in terms of both certification rate and accuracy.

cs.LG↗

arXiv2023

Large shift current, $π$ Zak phase and unconventional nature in Se and Te

Recently, unconventional materials (or obstructed atomic insulators) have attracted much attention owing to the unconventional feature of mismatch between Wannier centers and atomic positions. In this paper, we demonstrate that the trigonal selenium and tellurium host an unconventional nature in both electronic and phonon spectra. In electronic band structures, the band representation (BR) decomposition for occupied bands has to contain the essential BR of $A@3b$, and the real-space invariant is $δ_1@3b=-1$. The $π$ Zak phase suggests that the one-dimensional Se/Te chain is a chiral Su-Schrieffer-Heeger chain. The effective magnetism can be induced by $p$ states at ends. More importantly, a large shift current is obtained in Se quantum well. In addtion, in phonon spectra, three sets of phonon bands are well separated and assigned to $B@3b$, $B@3a$, and $A@3b$ BRs, respectively. Thus, the obstructed phonon states are predicted on the (0001) surface. As the prototypes of unconventional materials in both electronic and phonon spectra, our findings could create much interest in the study of obstructed surface electronic and phonon states in these novel materials.

cond-mat.mtrl-sci↗

arXiv2023

Magnetic field filtering of the hinge supercurrent in unconventional metal NiTe$_2$-based Josephson junctions

Topological materials with boundary (surface/edge/hinge) states have attracted tremendous research interest. Besides, unconventional (obstructed atomic) materials have recently drawn lots of attention owing to their obstructed boundary states. Experimentally, Josephson junctions (JJs) constructed on materials with boundary states produce the peculiar boundary supercurrent, which was utilized as a powerful diagnostic approach. Here, we report the observations of conspicuous hinge supercurrent in NiTe$_2$-based JJs. Particularly, applying an in-plane magnetic field along the Josephson current could rapidly suppress the bulk supercurrent and retain the nearly pure hinge supercurrent, namely the magnetic field filtering of supercurrent. Further systematic comparative analysis and theoretical calculations demonstrate the existence of unconventional nature and obstructed hinge states in NiTe$_2$. Our results revealed the unique hinge states in unconventional metal NiTe$_2$, and demonstrated in-plane magnetic field as an efficient method to filter out the futile bulk contributions and thereby to highlight the hinge states hidden in topological/unconventional materials.

cond-mat.mes-hall↗

arXiv2022

HCL-TAT: A Hybrid Contrastive Learning Method for Few-shot Event Detection with Task-Adaptive Threshold

Conventional event detection models under supervised learning settings suffer from the inability of transfer to newly-emerged event types owing to lack of sufficient annotations. A commonly-adapted solution is to follow a identify-then-classify manner, which first identifies the triggers and then converts the classification task via a few-shot learning paradigm. However, these methods still fall far short of expectations due to: (i) insufficient learning of discriminative representations in low-resource scenarios, and (ii) trigger misidentification caused by the overlap of the learned representations of triggers and non-triggers. To address the problems, in this paper, we propose a novel Hybrid Contrastive Learning method with a Task-Adaptive Threshold (abbreviated as HCLTAT), which enables discriminative representation learning with a two-view contrastive loss (support-support and prototype-query), and devises a easily-adapted threshold to alleviate misidentification of triggers. Extensive experiments on the benchmark dataset FewEvent demonstrate the superiority of our method to achieve better results compared to the state-of-the-arts. All the code and data of this paper will be available for online public access.

cs.CL↗

arXiv2022

Representing Affect Information in Word Embeddings

A growing body of research in natural language processing (NLP) and natural language understanding (NLU) is investigating human-like knowledge learned or encoded in the word embeddings from large language models. This is a step towards understanding what knowledge language models capture that resembles human understanding of language and communication. Here, we investigated whether and how the affect meaning of a word (i.e., valence, arousal, dominance) is encoded in word embeddings pre-trained in large neural networks. We used the human-labeled dataset as the ground truth and performed various correlational and classification tests on four types of word embeddings. The embeddings varied in being static or contextualized, and how much affect specific information was prioritized during the pre-training and fine-tuning phase. Our analyses show that word embedding from the vanilla BERT model did not saliently encode the affect information of English words. Only when the BERT model was fine-tuned on emotion-related tasks or contained extra contextualized information from emotion-rich contexts could the corresponding embedding encode more relevant affect information.

cs.CL↗

arXiv2022

Large spin Hall conductivity and excellent hydrogen evolution reaction activity in unconventional PtTe1.75 monolayer

Two-dimensional (2D) materials have gained lots of attention due to the potential applications. In this work, we propose that based on first-principles calculations, the (2$\times$2) patterned PtTe$_2$ monolayer with kagome lattice formed by the well-ordered Te vacancy (PtTe$_{1.75}$) hosts large spin Hall conductivity (SHC) and excellent hydrogen evolution reaction (HER) activity. The unconventional nature relies on the $A1@1b$ band representation (BR) of the highest valence band without SOC. The large SHC comes from the Rashba spin-orbit coupling (SOC) in the noncentrosymmetric structure induced by the Te vacancy. Even though it has a metallic SOC band structure, the $\mathbb Z_2$ invariant is well defined due to the existence of the direct band gap and is computed to be nontrivial. The calculated SHC is as large as 1.25$\times 10^3 \frac{\hbar}{e} (Ω~cm)^{-1}$ at the Fermi level ($E_F$). By tuning the chemical potential from $E_F-0.3$ to $E_F+0.3$ eV, it varies rapidly and monotonically from $-1.2\times 10^3$ to 3.1$\times 10^3 \frac{\hbar}{e} (Ω~cm)^{-1}$. In addition, we also find the Te vacancy in the patterned monolayer can induce excellent HER activity. Our results not only offer a new idea to search 2D materials with large SHC, i.e., by introducing inversion-symmetry breaking vacancies in large SOC systems, but also provide a feasible system with tunable SHC (by applying gate voltage) and excellent HER activity.

cond-mat.mtrl-sci↗

arXiv2022

MIPR:Automatic Annotation of Medical Images with Pixel Rearrangement

Most of the state-of-the-art semantic segmentation reported in recent years is based on fully supervised deep learning in the medical domain. How?ever, the high-quality annotated datasets require intense labor and domain knowledge, consuming enormous time and cost. Previous works that adopt semi?supervised and unsupervised learning are proposed to address the lack of anno?tated data through assisted training with unlabeled data and achieve good perfor?mance. Still, these methods can not directly get the image annotation as doctors do. In this paper, inspired by self-training of semi-supervised learning, we pro?pose a novel approach to solve the lack of annotated data from another angle, called medical image pixel rearrangement (short in MIPR). The MIPR combines image-editing and pseudo-label technology to obtain labeled data. As the number of iterations increases, the edited image is similar to the original image, and the labeled result is similar to the doctor annotation. Therefore, the MIPR is to get labeled pairs of data directly from amounts of unlabled data with pixel rearrange?ment, which is implemented with a designed conditional Generative Adversarial Networks and a segmentation network. Experiments on the ISIC18 show that the effect of the data annotated by our method for segmentation task is is equal to or even better than that of doctors annotations

eess.IV↗

arXiv2022

Instance Segmentation of Unlabeled Modalities via Cyclic Segmentation GAN

Instance segmentation for unlabeled imaging modalities is a challenging but essential task as collecting expert annotation can be expensive and time-consuming. Existing works segment a new modality by either deploying a pre-trained model optimized on diverse training data or conducting domain translation and image segmentation as two independent steps. In this work, we propose a novel Cyclic Segmentation Generative Adversarial Network (CySGAN) that conducts image translation and instance segmentation jointly using a unified framework. Besides the CycleGAN losses for image translation and supervised losses for the annotated source domain, we introduce additional self-supervised and segmentation-based adversarial objectives to improve the model performance by leveraging unlabeled target domain images. We benchmark our approach on the task of 3D neuronal nuclei segmentation with annotated electron microscopy (EM) images and unlabeled expansion microscopy (ExM) data. Our CySGAN outperforms both pretrained generalist models and the baselines that sequentially conduct image translation and segmentation. Our implementation and the newly collected, densely annotated ExM nuclei dataset, named NucExM, are available at https://connectomics-bazaar.github.io/proj/CySGAN/index.html.

cs.CV↗

arXiv2022

Soft-CP: A Credible and Effective Data Augmentation for Semantic Segmentation of Medical Lesions

The medical datasets are usually faced with the problem of scarcity and data imbalance. Moreover, annotating large datasets for semantic segmentation of medical lesions is domain-knowledge and time-consuming. In this paper, we propose a new object-blend method(short in soft-CP) that combines the Copy-Paste augmentation method for semantic segmentation of medical lesions offline, ensuring the correct edge information around the lession to solve the issue above-mentioned. We proved the method's validity with several datasets in different imaging modalities. In our experiments on the KiTS19[2] dataset, Soft-CP outperforms existing medical lesions synthesis approaches. The Soft-CP augementation provides gains of +26.5% DSC in the low data regime(10% of data) and +10.2% DSC in the high data regime(all of data), In offline training data, the ratio of real images to synthetic images is 3:1.

eess.IV↗

arXiv2021

Invertible Concept-based Explanations for CNN Models with Non-negative Concept Activation Vectors

Convolutional neural network (CNN) models for computer vision are powerful but lack explainability in their most basic form. This deficiency remains a key challenge when applying CNNs in important domains. Recent work on explanations through feature importance of approximate linear models has moved from input-level features (pixels or segments) to features from mid-layer feature maps in the form of concept activation vectors (CAVs). CAVs contain concept-level information and could be learned via clustering. In this work, we rethink the ACE algorithm of Ghorbani et~al., proposing an alternative invertible concept-based explanation (ICE) framework to overcome its shortcomings. Based on the requirements of fidelity (approximate models to target models) and interpretability (being meaningful to people), we design measurements and evaluate a range of matrix factorization methods with our framework. We find that non-negative concept activation vectors (NCAVs) from non-negative matrix factorization provide superior performance in interpretability and fidelity based on computational and human subject experiments. Our framework provides both local and global concept-level explanations for pre-trained CNN models.

cs.CV↗

arXiv2021

Noncommutative $H^p$ spaces associated with type 1 subdiagonal algebras

Let $\mathfrak A$ be a type 1 subdiagonal algebra in a $σ$-finite von Neumann algebra $\mathcal M$ with respect to a faithful normal conditional expectation $Φ$. We consider a Riesz type factorization theorem in noncommutative $H^p$ spaces associated with $\mathfrak A$. It is shown that if $1\leq r,p,q<\infty$ such that $\frac1r=\frac1p+\frac1q$, then for any $h\in H^r$, there exist $h_p\in H^p$ and $h_q\in H^q$ such that $h=h_ph_q$. Beurling type invariant subspace theorem for noncommutative $L^p(1< p<\infty)$ space is obtained. Furthermore, we show that a $σ$-weakly closed subalgebra containing $\mathfrak A$ of $\mathcal M$ is also a type 1 subdiagonal algebra. As an application, We prove that the relative invariant subspace lattice $Lat_{\mathcal M}\mathfrak A$ of $\mathfrak A$ in $\mathcal M$ is commutative.

math.OA↗