Source author record

Chong Wang

Chong Wang appears in the imported research catalog. Authorship, coauthor and topic links are available while profile ownership is still unclaimed.

ResearcherUnclaimed source record

Catalog footprint

What is connected

84works

35topics

4close collaborators

Actions

Connect this record

Open graph Browse works

Inspect adjacent papers, topics, institutions and collaborators without losing the researcher page.

Building this map preview

BZPEER is loading the nearby papers, people, topics and institutions for this page.

preprint2026arXiv

Bridging Generation and Training: A Systematic Review of Quality Issues in LLMs for Code

Large language models (LLMs) frequently generate defective outputs in code generation tasks, ranging from logical bugs to security vulnerabilities. While these generation failures are often treated as model-level limitations, empirical evidence increasingly traces their root causes to imperfections within the training corpora. Yet, the specific mechanisms linking training data quality issues to generated code quality issues remain largely unmapped. This paper presents a systematic literature review of 114 primary studies to investigate how training data quality issues propagate into code generation. We establish a unified taxonomy that categorizes generated code quality issues across nine dimensions and training data quality issues into code and non-code attributes. Based on this taxonomy, we formalize a causal framework detailing 18 typical propagation mapping mechanisms. Furthermore, we synthesize state-of-the-art detection and mitigation techniques across the data, model, and generation lifecycles. The reviewed literature reveals a clear methodological shift: quality assurance is transitioning from reactive, heuristic-based post-generation filtering toward proactive, data-centric governance and closed-loop repair. Finally, we identify open challenges and outline research directions for developing reliable LLMs for code through integrated data curation and continuous evaluation. Our repository is available at https://github.com/SYSUSELab/From-Data-to-Code.

preprint2026arXiv

CheXTemporal: A Dataset for Temporally-Grounded Reasoning in Chest Radiography

Chest radiograph interpretation requires temporal reasoning over prior and current studies, yet most vision-language models are trained on static image-report pairs and lack explicit supervision for modeling longitudinal change. We introduce CheXTemporal, a dataset for temporally grounded reasoning in chest radiography consisting of paired prior-current chest X-rays (CXR) with finding-level temporal and spatial annotations. The dataset includes a five-class progression taxonomy (new, worse, stable, improved, resolved), localized spatial supervision of pathology, explicit spatial-temporal alignment across paired studies, and multi-source coverage for cross-domain evaluation. We additionally construct a 280K-pair silver dataset with automatically derived temporal and anatomical supervision for large-scale evaluation under weaker supervision. Using these resources, we evaluate multiple state-of-the-art vision-language CXR models on grounding and progression-classification tasks in a zero-shot setting. Across both gold and silver evaluations, current models exhibit consistent limitations in spatial grounding, fine-grained temporal reasoning, and robustness under distribution shift. In particular, models perform substantially better on salient progression categories such as worse than on temporally subtle states such as stable and resolved, suggesting limited modeling of longitudinal disease evolution in chest radiography.

preprint2026arXiv

Improving Flexible Image Tokenizers for Autoregressive Image Generation

Flexible image tokenizers aim to represent an image using an ordered 1D variable-length token sequence. This flexible tokenization is typically achieved through nested dropout, where a portion of trailing tokens is randomly truncated during training, and the image is reconstructed using the remaining preceding sequence. However, this tail-truncation strategy inherently concentrates the image information in the early tokens, limiting the effectiveness of downstream AutoRegressive (AR) image generation as the token length increases. To overcome these limitations, we propose \textbf{ReToK}, a flexible tokenizer with \underline{Re}dundant \underline{Tok}en Padding and Hierarchical Semantic Regularization, designed to fully exploit all tokens for enhanced latent modeling. Specifically, we introduce \textbf{Redundant Token Padding} to activate tail tokens more frequently, thereby alleviating information over-concentration in the early tokens. In addition, we apply \textbf{Hierarchical Semantic Regularization} to align the decoding features of earlier tokens with those from a pre-trained vision foundation model, while progressively reducing the regularization strength toward the tail to allow finer low-level detail reconstruction. Extensive experiments demonstrate the effectiveness of ReTok: on ImageNet 256$\times$256, our method achieves superior generation performance compared with both flexible and fixed-length tokenizers. Code will be available at: \href{https://github.com/zfu006/ReTok}{https://github.com/zfu006/ReTok}

preprint2026arXiv

Interleaved Reasoning for Large Language Models via Reinforcement Learning

Long chain-of-thought (CoT) significantly enhances the reasoning capabilities of large language models (LLMs). However, extensive reasoning traces lead to inefficiencies and increased time-to-first-token (TTFT). We propose a training paradigm that uses only reinforcement learning (RL) to guide reasoning LLMs to interleave thinking and answering for multi-hop questions. We observe that models inherently possess the ability to perform interleaved reasoning, which can be further enhanced through RL. We introduce a simple yet effective reward scheme to incentivize correct intermediate steps, guiding the policy model toward correct reasoning paths by leveraging intermediate signals generated during interleaved reasoning. Extensive experiments across five diverse datasets and three RL algorithms (PPO, GRPO, and REINFORCE++) demonstrate consistent improvements over traditional think-answer reasoning, without requiring external tools. Our method improves final task accuracy and overall efficiency by enabling more effective credit assignment during RL. Specifically, our approach achieves a 12.5% improvement in Pass@1 accuracy, while reducing overall reasoning length by 37% and TTFT by over 80% on average. Furthermore, our method, trained solely on question answering and logical reasoning datasets, exhibits strong generalization to complex reasoning datasets such as MATH, GPQA, and MMLU. Additionally, we conduct in-depth analysis to reveal several valuable insights into conditional reward modeling.

preprint2024arXiv

Enhancement of Ising superconductivity in monolayer NbSe$_2$ via surface fluorination

Recently discovered Ising superconductors have garnered considerable interest due to their anomalously large in-plane upper critical fields ($B_{c2}$). However, the requisite strong spin-orbital coupling in the Ising pairing mechanism generally renders these superconductors heavy-element dominant with notably low superconducting transition temperatures ($T_c$). Here, based on the Migdal-Eliashberg theory and the mean-field Bogoliubov-de Gennes Hamiltonian, we demonstrate a significant enhancement of Ising superconductivity in monolayer NbSe$_2$ through surface fluorination, as evidenced by concomitant improvements in $T_c$ and $B_{c2}$. This enhancement arises from three predominant factors. Firstly, fluorine atoms symmetrically and stably adhere to both sides of the monolayer NbSe$_2$, thereby maintaining the out-of-plane mirror symmetry and locking carrier spins out-of-plane. Secondly, fluorination suppresses the charge density wave in monolayer NbSe$_2$ and induces a van Hove singularity in the vicinity of the Fermi level, leading to a marked increase in the number of carriers and, consequently, strengthening the electron-phonon coupling (EPC). Lastly, the appearance of fluorine-related, low-frequency phonon modes further augments the EPC. Our findings suggest a promising avenue to elevate $T_c$ in two-dimensional Ising superconductors without compromising their Ising pairing.

preprint2023arXiv

Asymmetric Co-teaching with Multi-view Consensus for Noisy Label Learning

Learning with noisy-labels has become an important research topic in computer vision where state-of-the-art (SOTA) methods explore: 1) prediction disagreement with co-teaching strategy that updates two models when they disagree on the prediction of training samples; and 2) sample selection to divide the training set into clean and noisy sets based on small training loss. However, the quick convergence of co-teaching models to select the same clean subsets combined with relatively fast overfitting of noisy labels may induce the wrong selection of noisy label samples as clean, leading to an inevitable confirmation bias that damages accuracy. In this paper, we introduce our noisy-label learning approach, called Asymmetric Co-teaching (AsyCo), which introduces novel prediction disagreement that produces more consistent divergent results of the co-teaching models, and a new sample selection approach that does not require small-loss assumption to enable a better robustness to confirmation bias than previous methods. More specifically, the new prediction disagreement is achieved with the use of different training strategies, where one model is trained with multi-class learning and the other with multi-label learning. Also, the new sample selection is based on multi-view consensus, which uses the label views from training labels and model predictions to divide the training set into clean and noisy for training the multi-class model and to re-label the training samples with multiple top-ranked labels for training the multi-label model. Extensive experiments on synthetic and real-world noisy-label datasets show that AsyCo improves over current SOTA methods.

preprint2023arXiv

Knowledge Distillation to Ensemble Global and Interpretable Prototype-Based Mammogram Classification Models

State-of-the-art (SOTA) deep learning mammogram classifiers, trained with weakly-labelled images, often rely on global models that produce predictions with limited interpretability, which is a key barrier to their successful translation into clinical practice. On the other hand, prototype-based models improve interpretability by associating predictions with training image prototypes, but they are less accurate than global models and their prototypes tend to have poor diversity. We address these two issues with the proposal of BRAIxProtoPNet++, which adds interpretability to a global model by ensembling it with a prototype-based model. BRAIxProtoPNet++ distills the knowledge of the global model when training the prototype-based model with the goal of increasing the classification accuracy of the ensemble. Moreover, we propose an approach to increase prototype diversity by guaranteeing that all prototypes are associated with different training images. Experiments on weakly-labelled private and public datasets show that BRAIxProtoPNet++ has higher classification accuracy than SOTA global and prototype-based models. Using lesion localisation to assess model interpretability, we show BRAIxProtoPNet++ is more effective than other prototype-based models and post-hoc explanation of global models. Finally, we show that the diversity of the prototypes learned by BRAIxProtoPNet++ is superior to SOTA prototype-based approaches.

preprint2022arXiv

Contrastive Transformer-based Multiple Instance Learning for Weakly Supervised Polyp Frame Detection

Current polyp detection methods from colonoscopy videos use exclusively normal (i.e., healthy) training images, which i) ignore the importance of temporal information in consecutive video frames, and ii) lack knowledge about the polyps. Consequently, they often have high detection errors, especially on challenging polyp cases (e.g., small, flat, or partially visible polyps). In this work, we formulate polyp detection as a weakly-supervised anomaly detection task that uses video-level labelled training data to detect frame-level polyps. In particular, we propose a novel convolutional transformer-based multiple instance learning method designed to identify abnormal frames (i.e., frames with polyps) from anomalous videos (i.e., videos containing at least one frame with polyp). In our method, local and global temporal dependencies are seamlessly captured while we simultaneously optimise video and snippet-level anomaly scores. A contrastive snippet mining method is also proposed to enable an effective modelling of the challenging polyp cases. The resulting method achieves a detection accuracy that is substantially better than current state-of-the-art approaches on a new large-scale colonoscopy video dataset introduced in this work.

preprint2022arXiv

Differentially Private AUC Computation in Vertical Federated Learning

Federated learning has gained great attention recently as a privacy-enhancing tool to jointly train a machine learning model by multiple parties. As a sub-category, vertical federated learning (vFL) focuses on the scenario where features and labels are split into different parties. The prior work on vFL has mostly studied how to protect label privacy during model training. However, model evaluation in vFL might also lead to potential leakage of private label information. One mitigation strategy is to apply label differential privacy (DP) but it gives bad estimations of the true (non-private) metrics. In this work, we propose two evaluation algorithms that can more accurately compute the widely used AUC (area under curve) metric when using label DP in vFL. Through extensive experiments, we show our algorithms can achieve more accurate AUCs compared to the baselines.

preprint2022arXiv

Differentially Private Label Protection in Split Learning

Split learning is a distributed training framework that allows multiple parties to jointly train a machine learning model over vertically partitioned data (partitioned by attributes). The idea is that only intermediate computation results, rather than private features and labels, are shared between parties so that raw training data remains private. Nevertheless, recent works showed that the plaintext implementation of split learning suffers from severe privacy risks that a semi-honest adversary can easily reconstruct labels. In this work, we propose \textsf{TPSL} (Transcript Private Split Learning), a generic gradient perturbation based split learning framework that provides provable differential privacy guarantee. Differential privacy is enforced on not only the model weights, but also the communicated messages in the distributed computation setting. Our experiments on large-scale real-world datasets demonstrate the robustness and effectiveness of \textsf{TPSL} against label leakage attacks. We also find that \textsf{TPSL} have a better utility-privacy trade-off than baselines.

preprint2022arXiv

Differentially Private Multi-Party Data Release for Linear Regression

Differentially Private (DP) data release is a promising technique to disseminate data without compromising the privacy of data subjects. However the majority of prior work has focused on scenarios where a single party owns all the data. In this paper we focus on the multi-party setting, where different stakeholders own disjoint sets of attributes belonging to the same group of data subjects. Within the context of linear regression that allow all parties to train models on the complete data without the ability to infer private attributes or identities of individuals, we start with directly applying Gaussian mechanism and show it has the small eigenvalue problem. We further propose our novel method and prove it asymptotically converges to the optimal (non-private) solutions with increasing dataset size. We substantiate the theoretical results through experiments on both artificial and real-world datasets.

preprint2022arXiv

Edge physics at the deconfined transition between a quantum spin Hall insulator and a superconductor

We study the edge physics of the deconfined quantum phase transition (DQCP) between a spontaneous quantum spin Hall (QSH) insulator and a spin-singlet superconductor (SC). Although the bulk of this transition is in the same universality class as the paradigmatic deconfined Neel to valence-bond-solid transition, the boundary physics has a richer structure due to proximity to a quantum spin Hall state. We use the parton trick to write down an effective field theory for the QSH-SC transition in the presence of a boundary. We calculate various edge properties in an $N\to\infty$ limit. We show that the boundary Luttinger liquid in the QSH state survives at the phase transition, but only as "fractional" degrees of freedom that carry charge but not spin. The physical fermion remains gapless on the edge at the critical point, with a universal jump in the fermion scaling dimension as the system approaches the transition from the QSH side. The critical point could be viewed as a gapless analogue of the quantum spin Hall state but with the full $SU(2)$ spin rotation symmetry, which cannot be realized if the bulk is gapped.

preprint2022arXiv

Label Leakage and Protection from Forward Embedding in Vertical Federated Learning

Vertical federated learning (vFL) has gained much attention and been deployed to solve machine learning problems with data privacy concerns in recent years. However, some recent work demonstrated that vFL is vulnerable to privacy leakage even though only the forward intermediate embedding (rather than raw features) and backpropagated gradients (rather than raw labels) are communicated between the involved participants. As the raw labels often contain highly sensitive information, some recent work has been proposed to prevent the label leakage from the backpropagated gradients effectively in vFL. However, these work only identified and defended the threat of label leakage from the backpropagated gradients. None of these work has paid attention to the problem of label leakage from the intermediate embedding. In this paper, we propose a practical label inference method which can steal private labels effectively from the shared intermediate embedding even though some existing protection methods such as label differential privacy and gradients perturbation are applied. The effectiveness of the label attack is inseparable from the correlation between the intermediate embedding and corresponding private labels. To mitigate the issue of label leakage from the forward embedding, we add an additional optimization goal at the label party to limit the label stealing ability of the adversary by minimizing the distance correlation between the intermediate embedding and corresponding private labels. We conducted massive experiments to demonstrate the effectiveness of our proposed protection methods.

preprint2022arXiv

Label Leakage and Protection in Two-party Split Learning

Two-party split learning is a popular technique for learning a model across feature-partitioned data. In this work, we explore whether it is possible for one party to steal the private label information from the other party during split training, and whether there are methods that can protect against such attacks. Specifically, we first formulate a realistic threat model and propose a privacy loss metric to quantify label leakage in split learning. We then show that there exist two simple yet effective methods within the threat model that can allow one party to accurately recover private ground-truth labels owned by the other party. To combat these attacks, we propose several random perturbation techniques, including $\texttt{Marvell}$, an approach that strategically finds the structure of the noise perturbation by minimizing the amount of label leakage (measured through our quantification metric) of a worst-case adversary. We empirically demonstrate the effectiveness of our protection techniques against the identified attacks, and show that $\texttt{Marvell}$ in particular has improved privacy-utility tradeoffs relative to baseline approaches.

preprint2022arXiv

Learning to Simulate Unseen Physical Systems with Graph Neural Networks

Simulation of the dynamics of physical systems is essential to the development of both science and engineering. Recently there is an increasing interest in learning to simulate the dynamics of physical systems using neural networks. However, existing approaches fail to generalize to physical substances not in the training set, such as liquids with different viscosities or elastomers with different elasticities. Here we present a machine learning method embedded with physical priors and material parameters, which we term as "Graph-based Physics Engine" (GPE), to efficiently model the physical dynamics of different substances in a wide variety of scenarios. We demonstrate that GPE can generalize to materials with different properties not seen in the training set and perform well from single-step predictions to multi-step roll-out simulations. In addition, introducing the law of momentum conservation in the model significantly improves the efficiency and stability of learning, allowing convergence to better models with fewer training steps.

preprint2022arXiv

Light-Induced Ferromagnetism in Moiré Superlattices

Many-body interactions between carriers lie at the heart of correlated physics. The ability to tune such interactions would open the possibility to access and control complex electronic phase diagrams on demand. Recently, moiré superlattices formed by two-dimensional materials have emerged as a promising platform for quantum engineering such phenomena. The power of the moiré system lies in the high tunability of its physical parameters by tweaking layer twist angle, electrical field, moiré carrier filling, and interlayer coupling. Here, we report that optical excitation can drastically tune the spin-spin interactions between moiré trapped carriers, resulting in ferromagnetic order in WS2/WSe2 moiré superlattices over a small range of doping at elevated temperatures. Near the filling factor v = -1/3 (i.e., one hole per three moiré unit cells), as the excitation power at the exciton resonance increases, a well-developed hysteresis loop emerges in the reflective magnetic circular dichroism (RMCD) signal as a function of magnetic field, a hallmark of ferromagnetism. The hysteresis loop persists down to charge neutrality, and its shape evolves as the moiré superlattice is gradually filled, indicating changes of magnetic ground state properties. The observed phenomenon points to a mechanism in which itinerant photo-excited excitons mediate exchange coupling between moiré trapped holes. This exciton-mediated interaction can be of longer range than direct coupling between moiré trapped holes, and thus magnetic order can arise even in the dilute hole regime under optical excitation. This discovery adds a new and dynamic tuning knob to the rich many-body Hamiltonian of moiré quantum matter.

preprint2022arXiv

Linear Complexity Randomized Self-attention Mechanism

Recently, random feature attentions (RFAs) are proposed to approximate the softmax attention in linear time and space complexity by linearizing the exponential kernel. In this paper, we first propose a novel perspective to understand the bias in such approximation by recasting RFAs as self-normalized importance samplers. This perspective further sheds light on an \emph{unbiased} estimator for the whole softmax attention, called randomized attention (RA). RA constructs positive random features via query-specific distributions and enjoys greatly improved approximation fidelity, albeit exhibiting quadratic complexity. By combining the expressiveness in RA and the efficiency in RFA, we develop a novel linear complexity self-attention mechanism called linear randomized attention (LARA). Extensive experiments across various domains demonstrate that RA and LARA significantly improve the performance of RFAs by a substantial margin.

preprint2022arXiv

Non-zero momentum requires long-range entanglement

We show that a quantum state in a lattice spin (boson) system must be long-range entangled if it has non-zero lattice momentum, i.e. if it is an eigenstate of the translation symmetry with eigenvalue $e^{iP}\neq1$. Equivalently, any state that can be connected with a non-zero momentum state through a finite-depth local unitary transformation must also be long-range entangled. The statement can also be generalized to fermion systems. Some non-trivial consequences follow immediately from our theorem: (1) several different types of Lieb-Schultz-Mattis-Oshikawa-Hastings (LSMOH) theorems, including a previously unknown version involving only a discrete $\mathbb{Z}_n$ symmetry, can be derived in a simple manner from our result; (2) a gapped topological order (in space dimension $d>1$) must weakly break translation symmetry if one of its ground states on torus has nontrivial momentum - this generalizes the familiar physics of Tao-Thouless; (3) our result provides further evidence of the "smoothness" assumption widely used in the classification of crystalline symmetry-protected topological (cSPT) phases.

preprint2022arXiv

NVUM: Non-Volatile Unbiased Memory for Robust Medical Image Classification

Real-world large-scale medical image analysis (MIA) datasets have three challenges: 1) they contain noisy-labelled samples that affect training convergence and generalisation, 2) they usually have an imbalanced distribution of samples per class, and 3) they normally comprise a multi-label problem, where samples can have multiple diagnoses. Current approaches are commonly trained to solve a subset of those problems, but we are unaware of methods that address the three problems simultaneously. In this paper, we propose a new training module called Non-Volatile Unbiased Memory (NVUM), which non-volatility stores running average of model logits for a new regularization loss on noisy multi-label problem. We further unbias the classification prediction in NVUM update for imbalanced learning problem. We run extensive experiments to evaluate NVUM on new benchmarks proposed by this paper, where training is performed on noisy multi-label imbalanced chest X-ray (CXR) training sets, formed by Chest-Xray14 and CheXpert, and the testing is performed on the clean multi-label CXR datasets OpenI and PadChest. Our method outperforms previous state-of-the-art CXR classifiers and previous methods that can deal with noisy labels on all evaluations. Our code is available at https://github.com/FBLADL/NVUM.

preprint2022arXiv

REPNP: Plug-and-Play with Deep Reinforcement Learning Prior for Robust Image Restoration

Image restoration schemes based on the pre-trained deep models have received great attention due to their unique flexibility for solving various inverse problems. In particular, the Plug-and-Play (PnP) framework is a popular and powerful tool that can integrate an off-the-shelf deep denoiser for different image restoration tasks with known observation models. However, obtaining the observation model that exactly matches the actual one can be challenging in practice. Thus, the PnP schemes with conventional deep denoisers may fail to generate satisfying results in some real-world image restoration tasks. We argue that the robustness of the PnP framework is largely limited by using the off-the-shelf deep denoisers that are trained by deterministic optimization. To this end, we propose a novel deep reinforcement learning (DRL) based PnP framework, dubbed RePNP, by leveraging a light-weight DRL-based denoiser for robust image restoration tasks. Experimental results demonstrate that the proposed RePNP is robust to the observation model used in the PnP scheme deviating from the actual one. Thus, RePNP can generate more reliable restoration results for image deblurring and super resolution tasks. Compared with several state-of-the-art deep image restoration baselines, RePNP achieves better results subjective to model deviation with fewer model parameters.

preprint2022arXiv

The average distance problem with an Euler elastica penalization

We consider the minimization of an average distance functional defined on a two-dimensional domain $Ω$ with an Euler elastica penalization associated with $\pd Ω$, the boundary of $Ω$. The average distance is given by \begin{equation*} \int_Ω \dist^p(x,\pd Ω)\d x \end{equation*} where $p\geq 1$ is a given parameter, and $\dist(x,\pd Ω)$ is the Hausdorff distance between $\{x\}$ and $\pd Ω$. The penalty term is a multiple of the Euler elastica (i.e., the Helfrich bending energy or the Willmore energy) of the boundary curve ${\pd Ω}$, which is proportional to the integrated squared curvature defined on $\pd Ω$, as given by \begin{equation*} \la \int_{\pd Ω} κ_{\pd Ω}^2\d\H_{\llcorner \pd Ω}^1, \end{equation*} where $κ_{\pd Ω}$ denotes the (signed) curvature of $\pd Ω$ and $\la>0$ denotes a penalty constant. The domain $Ω$ is allowed to vary among compact, convex sets of $\mathbb{R}^2$ with Hausdorff dimension equal to $2$\tcr{.} Under no a priori assumptions on the regularity of the boundary $\pd Ω$, we prove the existence of minimizers of $E_{p,\la}$. Moreover, we establish the $C^{1,1}$-regularity of its minimizers. An original construction of a suitable family of competitors plays a decisive role in proving the regularity.

preprint2022arXiv

The average distance problem with perimeter-to-area ratio penalization

In this paper we consider the functional \begin{equation*} E_{p,\la}(Ω):=\int_Ω\dist^p(x,\pd Ω)\d x+\la \frac{\H^1(\pd Ω)}{\H^2(Ω)}. \end{equation*} Here $p\geq 1$, $\la>0$ are given parameters, the unknown $Ω$ varies among compact, convex, Hausdorff two-dimensional sets of $\R^2$, $\pd Ω$ denotes the boundary of $Ω$, and $\dist(x,\pd Ω):=\inf_{y\in\pd Ω}|x-y|$. The integral term $\int_Ω\dist^p(x,\pd Ω)\d x$ quantifies the "easiness" for points in $Ω$ to reach the boundary, while $\frac{\H^1(\pd Ω)}{\H^2(Ω)}$ is the perimeter-to-area ratio. The main aim is to prove existence and $C^{1,1}$-regularity of minimizers of $\E$.

preprint2021arXiv

A Künneth Formula of Hypergraphs

In this paper, based on the embedded homology groups of hypergraphs defined in \cite{h1}, we define the product of hypergraphs and prove the corresponding Künneth formula of hypergraphs which can be generalized to the Künneth formula for the embedded homology of graded subsets of chain complexes with coefficients in a principal ideal domain.

preprint2021arXiv

Abrupt switching of the anomalous Hall effect by field-rotation in nonmagnetic ZrTe5

The Hall effect arises when time reversal symmetry is broken by either intrinsic magnetism or an external magnetic field. The latter contribution dominates in non-magnetic materials, in which the angular dependence of the Hall effect is typically a smooth cosine function because only the out-of-plane projection of the field generates the in-plane transverse motion of electrons. Here, we report the observation of an abrupt switching of the Hall effect by field rotation in a non-magnetic material, ZrTe5. The angular dependence of the Hall resistivity approaches a signum function, persisting down to an extremely low field of 0.03 T. By varying the carrier density of ZrTe5 over three orders of magnitude, we show that this singular behavior is due to the anomalous Hall effect generated by the ultra-dilute massive Dirac carriers in the quantum limit of Pauli paramagnetism when the Zeeman energy exceeds the Fermi energy. Our results elucidate the origin of the anomalous Hall effect in ZrTe5, arising owing to the spin-polarized massive Dirac electrons rather than the separation of Weyl nodes.

preprint2021arXiv

Electric control of a canted-antiferromagnetic Chern insulator

The interplay between band topology and magnetism can give rise to exotic states of matter. For example, magnetically doped topological insulators can realize a Chern insulator that exhibits quantized Hall resistance at zero magnetic field. While prior works have focused on ferromagnetic systems, little is known about band topology and its manipulation in antiferromagnets. Here, we report that MnBi$_2$Te$_4$ is a rare platform for realizing a canted-antiferromagnetic (cAFM) Chern insulator with electrical control. We show that the Chern insulator state with Chern number $C = 1$ appears as soon as the AFM to canted-AFM phase transition happens. The Chern insulator state is further confirmed by observing the unusual transition of the $C = 1$ state in the cAFM phase to the $C = 2$ orbital quantum Hall states in the magnetic field induced ferromagnetic phase. Near the cAFM-AFM phase boundary, we show that the Chern number can be toggled on and off by applying an electric field alone. We attribute this switching effect to the electrical field tuning of the exchange gap alignment between the top and bottom surfaces. Our work paves the way for future studies on topological cAFM spintronics and facilitates the development of proof-of-concept Chern insulator devices.

preprint2021arXiv

Intrinsic nonlinear Hall effect in antiferromagnetic tetragonal CuMnAs

Detecting the orientation of the Néel vector is a major research topic in antiferromagnetic spintronics. Here we recognize the intrinsic nonlinear Hall effect, which is independent of the relaxation time, as a prominent contribution to the time-reversal-odd second order conductivity and can be used to detect the flipping of the Néel vector. In contrast, the Berry-curvature-dipole-induced nonlinear Hall effect depends linear on relaxation time and is time-reversal-even. We study the intrinsic nonlinear Hall effect in an antiferromagnetic metal: tetragonal CuMnAs, and show that its nonlinear Hall conductivity can reach the order of mA/V$^2$. The dependence on the chemical potential of such nonlinear Hall conductivity can be qualitatively explained by a tilted massive Dirac model. Moreover, we demonstrate its strong temperature dependence and briefly discuss its competition with the second order Drude conductivity. Finally, a complete survey of magnetic point groups are presented, providing guidelines for finding more antiferromagnetic materials with the intrinsic nonlinear Hall effect.

preprint2021arXiv

Multi-Domain Neural Machine Translation with Word-Level Adaptive Layer-wise Domain Mixing

Many multi-domain neural machine translation (NMT) models achieve knowledge transfer by enforcing one encoder to learn shared embedding across domains. However, this design lacks adaptation to individual domains. To overcome this limitation, we propose a novel multi-domain NMT model using individual modules for each domain, on which we apply word-level, adaptive and layer-wise domain mixing. We first observe that words in a sentence are often related to multiple domains. Hence, we assume each word has a domain proportion, which indicates its domain preference. Then word representations are obtained by mixing their embedding in individual domains based on their domain proportions. We show this can be achieved by carefully designing multi-head dot-product attention modules for different domains, and eventually taking weighted averages of their parameters by word-level layer-wise domain proportions. Through this, we can achieve effective domain knowledge sharing, and capture fine-grained domain-specific knowledge as well. Our experiments show that our proposed model outperforms existing ones in several NMT tasks.

preprint2021arXiv

Plasmons in the van der Waals charge-density-wave material 2H-TaSe2

Plasmons in two-dimensional (2D) materials beyond graphene have recently gained much attention. However, the experimental investigation is limited due to the lack of suitable materials. Here, we experimentally demonstrate localized plasmons in a correlated 2D charge-density-wave (CDW) material: 2H-TaSe2. The plasmon resonance can cover a broad spectral range from the terahertz (40 μm) to the telecom (1.55 μm) region, which is further tunable by changing thickness and dielectric environments. The plasmon dispersion flattens at large wave vectors, resulted from the universal screening effect of interband transitions. More interestingly, anomalous temperature dependence of plasmon resonances associated with CDW excitations is observed. In the CDW phase, the plasmon peak close to the CDW excitation frequency becomes wider and asymmetric, mimicking two coupled oscillators. Our study not only reveals the universal role of the intrinsic screening on 2D plasmons, but also opens an avenue for tunable plasmons in 2D correlated materials.

preprint2021arXiv

Symmetry-adapted graph neural networks for constructing molecular dynamics force fields

Molecular dynamics is a powerful simulation tool to explore material properties. Most of the realistic material systems are too large to be simulated with first-principles molecular dynamics. Classical molecular dynamics has lower computational cost but requires accurate force fields to achieve chemical accuracy. In this work, we develop a symmetry-adapted graph neural networks framework, named molecular dynamics graph neural networks (MDGNN), to construct force fields automatically for molecular dynamics simulations for both molecules and crystals. This architecture consistently preserves the translation, rotation and permutation invariance in the simulations. We propose a new feature engineering method including higher order contributions and show that MDGNN accurately reproduces the results of both classical and first-principles molecular dynamics. We also demonstrate that force fields constructed by the model has good transferability. Therefore, MDGNN provides an efficient and promising option for molecular dynamics simulations of large scale systems with high accuracy.

preprint2021arXiv

Tunable plasmons in large area WTe2 thin films

The observation of the electrically tunable and highly confined plasmons in graphene has stimulated the exploration of interesting properties of plasmons in other two dimensional materials. Recently, hyperbolic plasmon resonance modes are observed in exfoliated WTe2 films, a type-II Weyl semimetal with layered structure, providing a platform for the assembly of plasmons with hyperbolicity and exotic topological properties. However, the plasmon modes were observed in relatively thick and small-area films, which restrict the tunability and application for plasmons. Here, large-area (~ cm) WTe2 films with different thickness are grown by chemical vapor deposition method, in which plasmon resonance modes are observed in films with different thickness down to about 8 nm. Hybridization of plasmon and surface polar phonons of the substrate is revealed by mapping the plasmon dispersion. The plasmon frequency is demonstrated to be tunable by changing the temperature and film thickness. Our results facilitate the development of a tunable and scalable WTe2 plasmonic system for revealing topological properties and towards various applications in sensing, imaging and light modulation.

preprint2020arXiv

A Discrete Morse Theory for Digraphs

Digraphs are generalizations of graphs in which each edge is assigned with a direction or two directions. In this paper, we define discrete Morse functions on digraphs, and prove that the homology of the Morse complex and the path homology are isomorphic for a transitive digraph. We also study the collapses defined by discrete gradient vector fields. Let $G$ be a digraph and $f$ a discrete Morse function. Assume the out-degree and in-degree of any zero-point of $f$ on $G$ are both 1. We prove that the original digraph $G$ and its $\mathcal{M}$-collapse $\tilde{G}$ have the same path homology groups.

preprint2020arXiv

A Discrete Morse Theory for Hypergraphs

A hypergraph can be obtained from a simplicial complex by deleting some non-maximal simplices. By [11], a hypergraph gives an associated simplicial complex. By [4], the embedded homology of a hypergraph is the homology of the infimum chain complex, or equivalently, the homology of the supremum chain complex. In this paper, we generalize the discrete Morse theory for simplicial complexes by R. Forman [5-7] and give a discrete Morse theory for hypergraphs. We use the critical simplices of the associated simplicial complex to construct a sub-chain complex of the infimum chain complex and a sub-chain complex of the supremum chain complex, then prove that the embedded homology of a hypergraph is isomorphic to the homology of the constructed chain complexes. Moreover, we define discrete Morse functions on hypergraphs and compute the embedded homology in terms of the critical hyperedges. As by-products, we derive some Morse inequalities and collapse results for hypergraphs.

preprint2020arXiv

AutoEmb: Automated Embedding Dimensionality Search in Streaming Recommendations

Deep learning based recommender systems (DLRSs) often have embedding layers, which are utilized to lessen the dimensionality of categorical variables (e.g. user/item identifiers) and meaningfully transform them in the low-dimensional space. The majority of existing DLRSs empirically pre-define a fixed and unified dimension for all user/item embeddings. It is evident from recent researches that different embedding sizes are highly desired for different users/items according to their popularity. However, manually selecting embedding sizes in recommender systems can be very challenging due to the large number of users/items and the dynamic nature of their popularity. Thus, in this paper, we propose an AutoML based end-to-end framework (AutoEmb), which can enable various embedding dimensions according to the popularity in an automated and dynamic manner. To be specific, we first enhance a typical DLRS to allow various embedding dimensions; then we propose an end-to-end differentiable framework that can automatically select different embedding dimensions according to user/item popularity; finally we propose an AutoML based optimization algorithm in a streaming recommendation setting. The experimental results based on widely used benchmark datasets demonstrate the effectiveness of the AutoEmb framework.

preprint2020arXiv

Correlated Fragile Topology: a Parton Approach

A fragile topological insulator (FTI) can be viewed as an almost-atomic insulator, with emergent negative charges localized at certain real space points, even though the underlying lattice Hilbert space contains only positively charged states. Fragile topology in free fermion systems has been fruitfully studied using modern topological band theory. However, the concept is well defined even for strongly correlated systems, and fragile states that cannot be realized within free fermion band theory exist abundantly. In this work we propose a rather general parton construction for such correlation-enabled FTIs. In our parton construction the associated gauge symmetries are completely Higgsed, resulting in only short-range entangled states. The effective negative charges in the FTIs emerge naturally as the remnants of negatively charged partons. For spinful electrons with $SU(2)$ spin-rotation symmetry, the fragile phases can be viewed as FTIs of charge-$2$, spin singlet bosonic Cooper pairs. We discuss a few examples of correlated FTIs for both spinless and spinful fermions, including some "featureless Mott insulators" on the honeycomb lattice previously discussed in the literature.

preprint2020arXiv

Enhancement of superconductivity in organic-inorganic hybrid topological materials

Inducing or enhancing superconductivity in topological materials is an important route toward topological superconductivity. Reducing the thickness of transition metal dichalcogenides (e.g. WTe2 and MoTe2) has provided an important pathway to engineer superconductivity in topological matters; for instance, emergent superconductivity with Tc=0.82 K was observed in monolayer WTe2 which also hosts intriguing quantum spin Hall effect, although the bulk crystal is nonsuperconducting. However, such monolayer sample is difficult to obtain, unstable in air, and with extremely low Tc, which could pose a grand challenge for practical applications. Here we report an experimentally convenient approach to control the interlayer coupling to achieve tailored topological properties, enhanced superconductivity and good sample stability through organic cation intercalation of the Weyl semimetals MoTe2 and WTe2. The as-formed organic-inorganic hybrid crystals are weak topological insulators with enhanced Tc of 7.0 K for intercalated MoTe2 (0.25 K for pristine crystal) and 2.3 K for intercalated WTe2 (2.8 times compared to monolayer WTe2). Such organic-cationintercalation method can be readily applied to many other layered crystals, providing a new pathway for manipulating their electronic, topological and superconducting properties.

preprint2020arXiv

Fractional Quantum Hall Effect in Weyl Semimetals

Weyl semimetal may be thought of as a gapless topological phase protected by the chiral anomaly, where the symmetries involved in the anomaly are the $U(1)$ charge conservation and the crystal translational symmetry. The absence of a band gap in a weakly-interacting Weyl semimetal is mandated by the electronic structure topology and is guaranteed as long as the symmetries and the anomaly are intact. The nontrivial topology also manifests in the Fermi arc surface states and topological response, in particular taking the form of an anomalous Hall effect in magnetic Weyl semimetals, whose magnitude is only determined by the location of the Weyl nodes in the Brillouin zone. Here we consider the situation when the interactions are not weak and ask whether it is possible to open a gap in a magnetic Weyl semimetal while preserving its nontrivial electronic structure topology along with the translational and the charge conservation symmetries. Surprisingly, the answer turns out to be yes. The resulting topologically ordered state provides a nontrivial realization of the fractional quantum Hall effect in three spatial dimensions in the absence of an external magnetic field, which cannot be viewed as a stack of two dimensional states. Our state contains loop excitations with nontrivial braiding statistics when linked with lattice dislocations.

preprint2020arXiv

Learning to Structure Long-term Dependence for Sequential Recommendation

Sequential recommendation recommends items based on sequences of users' historical actions. The key challenge in it is how to effectively model the influence from distant actions to the action to be predicted, i.e., recognizing the long-term dependence structure; and it remains an underexplored problem. To better model the long-term dependence structure, we propose a GatedLongRec solution in this work. To account for the long-term dependence, GatedLongRec extracts distant actions of top-$k$ related categories to the user's ongoing intent with a top-$k$ gating network, and utilizes a long-term encoder to encode the transition patterns among these identified actions. As user intent is not directly observable, we take advantage of available side-information about the actions, i.e., the category of their associated items, to infer the intents. End-to-end training is performed to estimate the intent representation and predict the next action for sequential recommendation. Extensive experiments on two large datasets show that the proposed solution can recognize the structure of long-term dependence, thus greatly improving the sequential recommendation.

preprint2020arXiv

Self-formed 2D/3D Heterostructure on the Edge of 2D Ruddlesden-Popper Hybrid Perovskites Responsible for Intriguing Optoelectronic Properties and Higher Cell Efficiency

The observation of low energy edge photoluminescence and its beneficial effect on the solar cell efficiency of Ruddlesden-Popper perovskites has unleashed an intensive research effort to reveal its origin. This effort, however, has been met with more challenges as the underlying material structure has still not been identified; new modellings and observations also do not seem to converge. Using 2D (BA)2(MA)2Pb3Br10 as an example, we show that 3D MAPbBr3 is formed due to the loss of BA on the edge. This self-formed MAPbBr3 can explain the reported edge emission under various conditions, while the reported intriguing optoelectronic properties such as fast exciton trapping from the interior 2D perovskite, rapid exciton dissociation and long carrier lifetime can be understood via the self-formed 2D/3D lateral perovskite heterostructure. The 3D perovskite is identified by submicron infrared spectroscopy, the emergence of XRD signature from freezer-milled nanometer-sized 2D perovskite and its photoluminescence response to external hydrostatic pressure. The revelation of this edge emission mystery and the identification of a self-formed 2D/3D heterostructure provide a new approach to engineering 2D perovskites for high-performance optoelectronic devices.

preprint2020arXiv

Stacking Domain Wall Magnons in Twisted van der Waals Magnets

Using bilayer CrI$_3$ as an example, we demonstrate that stacking domain walls in van der Waals magnets can host one dimensional (1D) magnon channels, which have lower energies than bulk magnons. Interestingly, some magnon channels are hidden in magnetically homogeneous background and can only be inferred with the knowledge of stacking domain walls. Compared to 1D magnons confined in magnetic domain walls, 1D magnons in stacking domain walls are more stable against external perturbations. We show that the relaxed moiré superlattices of small-angle twisted bilayer CrI$_3$ is a natural realization of stacking domain walls and host interconnected moiré magnon network. Our work reveals the importance of stacking domain walls in understanding magnetic properties of van der Waals magnets, and extends the scope of stacking engineering to magnetic dynamics.

preprint2020arXiv

The optical conductivity of few-layer black phosphorus by infrared spectroscopy

The strength of light-matter interaction is of central importance in photonics and optoelectronics. For many widely studied two-dimensional semiconductors, such as MoS2, the optical absorption due to exciton resonances increases with thickness. However, here we will show, few-layer black phosphorus exhibits an opposite trend. We determine the optical conductivity of few-layer black phosphorus with thickness down to bilayer by infrared spectroscopy. On the contrary to our expectations, the frequency-integrated exciton absorption is found to be enhanced in thinner samples. Moreover, the continuum absorption near the band edge is almost a constant, independent of the thickness. We will show such scenario is related to the quanta of the universal optical conductivity of graphene, with a prefactor originating from the band anisotropy.

preprint2020arXiv

The optical properties of few-layer InSe

Few-layer InSe draws tremendous research interests owing to the superior electronic and optical properties. It exhibits high carrier mobility up to more than 1000 cm2/Vs at room temperature. The strongly layer-tunable band gap spans a large spectral range from near-infrared to the visible. In this perspective, we systematically review the optical properties of few-layer InSe. Firstly, the intrinsic optical and electronic properties are introduced. Compared to other two-dimensional (2D) materials, the light-matter interaction of few-layer InSe is unusual. The band gap transition is inactive or extremely weak for in-plane polarized light, and the emission light is mainly polarized along the out-of-plane direction. Secondly, we will present several schemes to tune the optical properties of few-layer InSe such as external strain, surface chemical doping and van der Waals (vdW) interfacing. Thirdly, we survey the applications of few-layer InSe in photodetection and heterostructures. Overall, few-layer InSe exhibits great potential not only in fundamental research, but also in electronic and optoelectronic applications.

preprint2020arXiv

Towards Earnings Call and Stock Price Movement

Earnings calls are hosted by management of public companies to discuss the company's financial performance with analysts and investors. Information disclosed during an earnings call is an essential source of data for analysts and investors to make investment decisions. Thus, we leverage earnings call transcripts to predict future stock price dynamics. We propose to model the language in transcripts using a deep learning framework, where an attention mechanism is applied to encode the text data into vectors for the discriminative network classifier to predict stock price movements. Our empirical experiments show that the proposed model is superior to the traditional machine learning baselines and earnings call information can boost the stock price prediction performance.

preprint2020arXiv

Tunable Graphene Split-Ring Resonators

A split-ring resonator is a prototype of meta-atom in metamaterials. Though noble metal-based split-ring resonators have been extensively studied, up to date, there is no experimental demonstration of split-ring resonators made from graphene, an emerging intriguing plasmonic material. Here, we experimentally demonstrate graphene split-ring resonators with deep subwavelength (about one hundredth of the excitation wavelength) magnetic dipole response in the terahertz regime. Meanwhile, the quadrupole and electric dipole are observed,depending on the incident light polarization. All modes can be tuned via chemical doping or stacking multiple graphene layers. The strong interaction with surface polar phonons of the SiO2 substrate also significantly modifies the response. Finite-element frequency domain simulations nicely reproduce experimental results. Our study moves one stride forward toward the multi-functional graphene metamaterials, beyond simple graphene ribbon or disk arrays with electrical dipole resonances only.

preprint2020arXiv

Type-II Ising superconductivity and anomalous metallic state in macro-size ambient-stable ultrathin crystalline films

Recent emergence of two-dimensional (2D) crystalline superconductors has provided a promising platform to investigate novel quantum physics and potential applications. To reveal essential quantum phenomena therein, ultralow temperature transport investigation on high quality ultrathin superconducting films is critically required, although it has been quite challenging experimentally. Here we report a systematic transport study on the ultrathin crystalline PdTe2 films grown by molecular beam epitaxy (MBE). Interestingly, a new type of Ising superconductivity in 2D centrosymmetric materials is revealed by the detection of large in-plane critical field more than 7 times Pauli limit. Remarkably, in perpendicular magnetic field, we provide solid evidence of anomalous metallic state characterized by the resistance saturation at low temperatures with high quality filters. The robust superconductivity with intriguing quantum phenomena in the macro-size ambient-stable ultrathin PdTe2 films remains almost the same for 20 months, showing great potentials in electronic and spintronic applications.

preprint2020arXiv

Van der Waals thin films of WTe2 for natural hyperbolic plasmonic surfaces

A hyperbolic plasmonic surface supports highly directional propagating polaritons with extremely large density of states. Such plasmon polaritons have been realized in artificially structured metasurfaces. However, the upper bound of the achievable plasmon wave vector is limited by the structure size, which calls for a natural hyperbolic surface without any structuring. Here, we experimentally demonstrate a natural hyperbolic plasmonic surface based on thin films of WTe2 in the light wavelength range of 16 to 23 microns by far infrared absorption spectroscopy. The topological transition from the elliptic to the hyperbolic regime is further manifested by mapping the iso-frequency contours of the plasmon. Moreover, the anisotropy character and plasmon frequency exhibit prominent temperature dependence. Our study demonstrates the first natural platform to host 2D hyperbolic plasmons, which opens exotic avenues for the manipulation of plasmon propagation, light-matter interaction and light emission in planar photonics.

preprint2019arXiv

A theory of deconfined pseudo-criticality

It has been proposed that the deconfined criticality in $(2+1)d$ -- the quantum phase transition between a Neel anti-ferromagnet and a valence-bond-solid (VBS) -- may actually be pseudo-critical, in the sense that it is a weakly first-order transition with a generically long correlation length. The underlying field theory of the transition would be a slightly complex (non-unitary) fixed point as a result of fixed points annihilation. This proposal was motivated by existing numerical results from large scale Monte-Carlo simulations as well as conformal bootstrap. However, an actual theory of such complex fixed point, incorporating key features of the transition such as the emergent $SO(5)$ symmetry, is so far absent. Here we propose a Wess-Zumino-Witten (WZW) nonlinear sigma model with level $k=1$, defined in $2+ε$ dimensions, with target space $S^{3+ε}$ and global symmetry $SO(4+ε)$. This gives a formal interpolation between the deconfined criticality at $d=3$ and the $SU(2)_1$ WZW theory at $d=2$ describing the spin-$1/2$ Heisenberg chain. The theory can be formally controlled, at least to leading order, in terms of the inverse of the WZW level $1/k$. We show that at leading order, there is a fixed point annihilation at $d^*\approx2.77$, with complex fixed points above this dimension including the physical $d=3$ case. The pseudo-critical properties such as correlation length, scaling dimensions and the drifts of scaling dimensions as the system size increases, calculated crudely to leading order, are qualitatively consistent with existing numerics.

preprint2019arXiv

Crystallographic splitting theorem for band representations and fragile topological photonic crystals

The fundamental building blocks in band theory are band representations (BRs): bands whose infinitely-numbered Wannier functions are generated (by action of a space group) from a finite number of symmetric Wannier functions centered on a point in space. This work aims to simplify questions on a multi-rank BR by splitting it into unit-rank bands, via the following crystallographic splitting theorem: being a rank-$N$ BR is equivalent to being splittable into a finite sum of bands indexed by $\{1,2,\ldots,N\}$, such that each band is spanned by a single, analytic Bloch function of $k$, and any symmetry in the space group acts by permuting $\{1,2,\ldots,N\}$. Applying this theorem, we develop computationally efficient methods to determine whether a given energy band (of a tight-binding or Schrödinger Hamiltonian) is a BR, and, if so, how to numerically construct the corresponding symmetric Wannier functions. Thus we prove that rotation-symmetric topological insulators in class AI are fragile, meaning that the obstruction to symmetric Wannier functions is removable by addition of BRs. An implication of fragility is that its boundary states, while robustly covering the bulk energy gap in finite-rank tight-binding models, are unstable if the Hilbert space is expanded to include all symmetry-allowed representations. These fragile insulators have photonic analogs that we identify; in particular, we prove that an existing photonic crystal built by Yang et al. [Nature 565, 622 (2019)] is fragile topological with removable surface states, which disproves a widespread perception of 'topologically-protected' surface states in time-reversal-invariant, gapped photonic/phononic crystals. Our theorem is finally applied to derive various symmetry obstructions on the Wannier functions of topological insulators, and to prove their equivalence with the nontrivial holonomy of Bloch functions.

preprint2019arXiv

From spinon band topology to the symmetry quantum numbers of monopoles in Dirac spin liquids

The interplay of symmetry and topology has been at the forefront of recent progress in quantum matter. Here we uncover an unexpected connection between band topology and the description of competing orders in a quantum magnet. Specifically we show that aspects of band topology protected by crystalline symmetries determine key properties of the Dirac spin liquid (DSL) which can be defined on the honeycomb, square, triangular and kagomé lattices. At low energies, the DSL on all these lattices is described by an emergent Quantum Electrodynamics (QED$_3$) with $N_f=4$ flavors of Dirac fermions coupled to a $U(1)$ gauge field. However the symmetry properties of the magnetic monopoles, an important class of critical degrees of freedom, behave very differently on different lattices. In particular, we show that the lattice momentum and angular momentum of monopoles can be determined from the charge (or Wannier) centers of the corresponding spinon insulator. We also show that for DSLs on bipartite lattices, there always exists a monopole that transforms trivially under all microscopic symmetries owing to the existence of a parent SU(2) gauge theory. We connect our results to generalized Lieb-Schultz-Mattis theorems and also derive the time-reversal and reflection properties of monopoles. Our results indicate that recent insights into free fermion band topology can also guide the description of strongly correlated quantum matter.

preprint2019arXiv

Short-Term Temporal Convolutional Networks for Dynamic Hand Gesture Recognition

The purpose of gesture recognition is to recognize meaningful movements of human bodies, and gesture recognition is an important issue in computer vision. In this paper, we present a multimodal gesture recognition method based on 3D densely convolutional networks (3D-DenseNets) and improved temporal convolutional networks (TCNs). The key idea of our approach is to find a compact and effective representation of spatial and temporal features, which orderly and separately divide task of gesture video analysis into two parts: spatial analysis and temporal analysis. In spatial analysis, we adopt 3D-DenseNets to learn short-term spatio-temporal features effectively. Subsequently, in temporal analysis, we use TCNs to extract temporal features and employ improved Squeeze-and-Excitation Networks (SENets) to strengthen the representational power of temporal features from each TCNs' layers. The method has been evaluated on the VIVA and the NVIDIA Gesture Dynamic Hand Gesture Datasets. Our approach obtains very competitive performance on VIVA benchmarks with the classification accuracies of 91.54%, and achieve state-of-the art performance with 86.37% accuracy on NVIDIA benchmark.

preprint2019arXiv

Type-II Ising Pairing in Few-Layer Stanene

Spin-orbit coupling has proven indispensable in realizing topological materials and more recently Ising pairing in two-dimensional superconductors. This pairing mechanism relies on inversion symmetry breaking and sustains anomalously large in-plane polarizing magnetic fields whose upper limit is expected to diverge at low temperatures, although experimental demonstration of this has remained elusive due to the required fields. In this work, the recently discovered superconductor few-layer stanene, i.e. epitaxially strained $α$-Sn, is shown to exhibit a new type of Ising pairing between carriers residing in bands with different orbital indices near the $Γ$-point. The bands are split as a result of spin-orbit locking without the participation of inversion symmetry breaking. The in-plane upper critical field is strongly enhanced at ultra-low temperature and reveals the sought for upturn.

preprint2018arXiv

Duality between $(2+1)d$ Quantum Critical Points

Duality refers to two equivalent descriptions of the same theory from different points of view. Recently there has been tremendous progress in formulating and understanding possible dualities of quantum many body theories in $2+1$-spacetime dimensions. Of particular interest are dualities that describe conformally invariant quantum field theories in $(2+1)d$. These arise as descriptions of quantum critical points in condensed matter physics. The appreciation of the possible dual descriptions of such theories has greatly enhanced our understanding of some challenging questions about such quantum critical points. Perhaps surprisingly the same dualities also underlie recent progress in our understanding of other problems such as the half-filled Landau level and correlated surface states of topological insulators. Here we provide a pedagogical review of these recent developments from a point of view geared toward condensed matter physics.

preprint2016arXiv

A Duality Web in 2+1 Dimensions and Condensed Matter Physics

Building on earlier work in the high energy and condensed matter communities, we present a web of dualities in $2+1$ dimensions that generalize the known particle/vortex duality. Some of the dualities relate theories of fermions to theories of bosons. Others relate different theories of fermions. For example, the long distance behavior of the $2+1$-dimensional analog of QED with a single Dirac fermion (a theory known as $U(1)_{1/2}$) is identified with the $O(2)$ Wilson-Fisher fixed point. The gauged version of that fixed point with a Chern-Simons coupling at level one is identified as a free Dirac fermion. The latter theory also has a dual version as a fermion interacting with some gauge fields. Assuming some of these dualities, other dualities can be derived. Our analysis resolves a number of confusing issues in the literature including how time reversal is realized in these theories. It also has many applications in condensed matter physics like the theory of topological insulators (and their gapped boundary states) and the problem of electrons in the lowest Landau level at half filling. (Our techniques also clarify some points in the fractional Hall effect and its description using flux attachment.) In addition to presenting several consistency checks, we also present plausible (but not rigorous) derivations of the dualities and relate them to $3+1$-dimensional $S$-duality.

preprint2016arXiv

A General Method for Robust Bayesian Modeling

Robust Bayesian models are appealing alternatives to standard models, providing protection from data that contains outliers or other departures from the model assumptions. Historically, robust models were mostly developed on a case-by-case basis; examples include robust linear regression, robust mixture models, and bursty topic models. In this paper we develop a general approach to robust Bayesian modeling. We show how to turn an existing Bayesian model into a robust model, and then develop a generic strategy for computing with it. We use our method to study robust variants of several models, including linear regression, Poisson regression, logistic regression, and probabilistic topic models. We discuss the connections between our methods and existing approaches, especially empirical Bayes and James-Stein estimation.

preprint2016arXiv

All-fiber upconversion high spectral resolution wind lidar using a Fabry-Perot interferometer

An all-fiber, micro-pulse and eye-safe high spectral resolution wind lidar (HSRWL) at 1550nm is proposed and demonstrated by using a pair of upconversion single-photon detectors and a fiber Fabry-Perot scanning interferometer (FFP-SI). In order to improve the optical detection efficiency, both the transmission spectrum and the reflection spectrum of the FFP-SI are used for spectral analyses of the aerosol backscatter and the reference laser pulse. The reference signal is tapped from the outgoing laser and served as a zero velocity indicator. The Doppler shift is retrieved from a frequency response function Q, which is defined as the ratio of difference of the transmitted signal and the reflected signal to their sum. Taking advantages of high signal-to-noise ratio of the detectors and high spectral resolution of the FFP-SI, the Q spectra of the aerosol backscatter are reconstructed along the line-of-sight (LOS) of the telescope. By applying a least squares fit procedure to the measured Q spectra, the center frequencies and the bandwidths are obtained simultaneously. And then the Doppler shifts are determined relative to the center frequency of the reference signal. To eliminate the influence of temperature fluctuations on the FFP-SI, the FFP-SI is cased in a chamber with temperature stability of 0.001 during the measurement. Continuous LOS wind observations are carried out on two days at Hefei (31.843 N, 117.265 E), China. In the meantime, LOS wind measurements from the HSRWL show good agreement with the results from an ultrasonic wind sensor (Vaisala windcap WMT52). Due to the computational expensive of the convolution operation of the Q function, an empirical method is adopted to evaluate the quality of the measurements. The standard deviation of the wind speed is 0.76 m/s at the 1.8 km. The standard deviation of the retrieved bandwidth variation is 2.07 MHz at the 1.8 km.

preprint2016arXiv

Composite fermi liquids in the lowest Landau level

We study composite fermi liquid (CFL) states in the lowest Landau level (LLL) limit at a generic filling $ν= \frac{1}{n}$. We begin with the old observation that, in compressible states, the composite fermion in the lowest Landau level should be viewed as a charge-neutral particle carrying vorticity. This leads to the absence of a Chern-Simons term in the effective theory of the CFL. We argue here that instead a Berry curvature should be enclosed by the fermi surface of composite fermions, with the total Berry phase fixed by the filling fraction $ϕ_B=-2πν$. We illustrate this point with the CFL of fermions at filling fractions $ν=1/2q$ and (single and two-component) bosons at $ν=1/(2q+1)$. The Berry phase leads to sharp consequences in the transport properties including thermal and spin Hall conductances, which in the RPA approximation are distinct from the standard Halperin-Lee-Read predictions. We emphasize that these results only rely on the LLL limit, and do not require particle-hole symmetry, which is present microscopically only for fermions at $ν=1/2$. Nevertheless, we show that the existing LLL theory of the composite fermi liquid for bosons at $ν=1$ does have an emergent particle-hole symmetry. We interpret this particle-hole symmetry as a transformation between the empty state at $ν=0$ and the boson integer quantum hall state at $ν=2$. This understanding enables us to define particle-hole conjugates of various bosonic quantum Hall states which we illustrate with the bosonic Jain and Pfaffian states. The bosonic particle-hole symmetry can be realized exactly on the surface of a three-dimensional boson topological insulator. We also show that with the particle-hole and spin $SU(2)$ rotation symmetries, there is no gapped topological phase for bosons at $ν=1$.

preprint2016arXiv

Micro-pulse upconversion Doppler lidar for wind and visibility detection in the atmospheric boundary layer

For the first time, a versatile, eyesafe, compact and direct detection Doppler lidar is developed using upconversion single-photon detection method. An all-fiber and polarization maintaining architecture is realized to guarantee the high optical coupling efficiency and the system stability. Using integrated-optic components, the conservation of etendue of the optical receiver is achieved by manufacturing a fiber-coupled periodically poled Lithium niobate waveguide and an all-fiber Fabry-Perot interferometer (FPI). The so-called double-edge direct detection is implemented using a single-channel FPI and a single upconversion detector, incorporating time-division multiplexing method. The relative error of the system is lower than 0.1% over 9 weeks. To show the robust of the system, atmospheric wind and visibility over 48 hours are detected in the boundary layer. In the intercomparison experiments, lidar shows good agreement with the ultrasonic wind sensor (Vaisala windcap WMT52), with standard deviation of 1.04 m/s in speed and 12.3° in direction.

preprint2016arXiv

Time-reversal symmetric U(1) quantum spin liquids

We study possible quantum $U(1)$ spin liquids in three dimensions with time-reversal symmetry. We find a total of 7 families of such $U(1)$ spin liquids, distinguished by the properties of their emergent electric/magnetic charges. We show how these spin liquids are related to each other. Two of these classes admit nontrivial protected surface states which we describe. We show how to access all of the 7 spin liquids through slave particle (parton) constructions. We also provide intuitive loop gas descriptions of their ground state wave functions. One of these phases is the `topological Mott insulator' conventionally described as a topological insulator of an emergent fermionic `spinon'. We show that this phase admits a remarkable dual description as a topological insulator of emergent fermionic magnetic monopoles. This results in a new (possibly natural) surface phase for the topological Mott insulator and a new slave particle construction. We describe some of the continuous quantum phase transitions between the different $U(1)$ spin liquids. Each of these seven families of states admits a finer distinction in terms of their surface properties which we determine by combining these spin liquids with symmetry protected topological phases. We discuss lessons for materials such as pyrochlore quantum spin ices which may harbor a $U(1)$ spin liquid. We suggest the topological Mott insulator as a possible ground state in some range of parameters for the quantum spin ice Hamiltonian.

preprint2015arXiv

Bound states of three fermions forming symmetry-protected topological phases

We propose a simple theoretical construction of certain short-range entangled phases of interacting fermions, by putting the bound states of three fermions (which we refer to as clustons) into topological bands. We give examples in two and three dimensions, and show that they are distinct from any free fermion state. We further argue that these states can be viewed as combinations of certain free fermion topological states and bosonic symmetry-protected topological (SPT) states. This provides a conceptually simple understanding of various SPT phases, and the possibility of realizing them in cold atom systems. New parton constructions of these SPT phases in purely bosonic systems are proposed. We also discuss a related anomaly in two dimensional Dirac theories, which is the gravitational analogue of the parity anomaly.

preprint2015arXiv

Continuous Time Dynamic Topic Models

In this paper, we develop the continuous time dynamic topic model (cDTM). The cDTM is a dynamic topic model that uses Brownian motion to model the latent topics through a sequential collection of documents, where a "topic" is a pattern of word use that we expect to evolve over the course of the collection. We derive an efficient variational approximate inference algorithm that takes advantage of the sparsity of observations in text, a property that lets us easily handle many time points. In contrast to the cDTM, the original discrete-time dynamic topic model (dDTM) requires that time be discretized. Moreover, the complexity of variational inference for the dDTM grows quickly as time granularity increases, a drawback which limits fine-grained discretization. We demonstrate the cDTM on two news corpora, reporting both predictive perplexity and the novel task of time stamp prediction.

preprint2015arXiv

Deep Speech 2: End-to-End Speech Recognition in English and Mandarin

We show that an end-to-end deep learning approach can be used to recognize either English or Mandarin Chinese speech--two vastly different languages. Because it replaces entire pipelines of hand-engineered components with neural networks, end-to-end learning allows us to handle a diverse variety of speech including noisy environments, accents and different languages. Key to our approach is our application of HPC techniques, resulting in a 7x speedup over our previous system. Because of this efficiency, experiments that previously took weeks now run in days. This enables us to iterate more quickly to identify superior architectures and algorithms. As a result, in several cases, our system is competitive with the transcription of human workers when benchmarked on standard datasets. Finally, using a technique called Batch Dispatch with GPUs in the data center, we show that our system can be inexpensively deployed in an online setting, delivering low latency when serving users at scale.

preprint2015arXiv

Dual Dirac liquid on the surface of the electron topological insulator

We discuss a non-fermi liquid gapless metallic surface state of the topological band insulator. It has an odd number of gapless Dirac fermions coupled to a non-compact U(1) gauge field. This can be viewed as a vortex dual to the conventional Dirac fermion surface state. This surface duality is a reflection of a bulk dual description discussed recently for the gauged topological insulator. All the other known surface states can be conveniently accessed from the dual Dirac liquid, including the surface quantum hall state, the Fu-Kane superconductor, the gapped symmetric topological order and the "composite Dirac liquid". We also discuss the physical properties of the dual Dirac liquid, and its connection to the half-filled Landau level.

preprint2015arXiv

Embarrassingly Parallel Variational Inference in Nonconjugate Models

We develop a parallel variational inference (VI) procedure for use in data-distributed settings, where each machine only has access to a subset of data and runs VI independently, without communicating with other machines. This type of "embarrassingly parallel" procedure has recently been developed for MCMC inference algorithms; however, in many cases it is not possible to directly extend this procedure to VI methods without requiring certain restrictive exponential family conditions on the form of the model. Furthermore, most existing (nonparallel) VI methods are restricted to use on conditionally conjugate models, which limits their applicability. To combat these issues, we make use of the recently proposed nonparametric VI to facilitate an embarrassingly parallel VI procedure that can be applied to a wider scope of models, including to nonconjugate models. We derive our embarrassingly parallel VI algorithm, analyze our method theoretically, and demonstrate our method empirically on a few nonconjugate models.

preprint2015arXiv

Fractionalized gapless quantum vortex liquids

The standard theoretical approach to gapless spin liquid phases of two-dimensional frustrated quantum antiferromagnets invokes the concept of fermionic slave particles into which the spin fractionalizes. As an alternate we explore new kinds of gapless spin liquid phases in frustrated quantum magnets with XY anisotropy where the vortex of the spin fractionalizes into gapless itinerant fermions. The resulting gapless fractionalized vortex liquid phases are studied within a slave particle framework that is dual to the usual one. We demonstrate the stability of some such phases and describe their properties. We give an explicit construction in an XY-spin-1 system on triangular lattice, and interpret it as a critical phase in the vicinity of spin-nematic states.

preprint2015arXiv

Half-filled Landau level, topological insulator surfaces, and three dimensional quantum spin liquids

We synthesize and partly review recent developments relating the physics of the half-filled Landau level in two dimensions to correlated surface states of topological insulators in three dimensions. The latter are in turn related to the physics of certain three dimensional quantum spin liquid states. The resulting insights provide an interesting answer to the old question of how particle-hole symmetry is realized in composite fermion liquids. Specifically the metallic state at filling $ν= \frac{1}{2}$ - described originally in pioneering work by Halperin , Lee, and Read as a liquid of composite fermions - was proposed recently by Son to be described by a particle-hole symmetric effective field theory distinct from that in the prior literature. We show how the relation to topological insulator surface states leads to a physical understanding of the correctness of this proposal. We develop a simple picture of the particle-hole symmetric composite fermion through a modification of older pictures as electrically neutral "dipolar" particles. We revisit the phenomenology of composite fermi liquids (with or without particle-hole symmetry), and show that their heat/electrical transport dramatically violates the conventional Wiedemann-Franz law but satisfies a modified one. We also discuss the implications of these insights for finding physical realizations of correlated topological insulator surfaces.

preprint2015arXiv

Interacting fermionic topological insulators/superconductors in three dimensions

Symmetry Protected Topological (SPT) phases are a minimal generalization of the concept of topological insulators to interacting systems. In this paper we describe the classification and properties of such phases for three dimensional(3D) electronic systems with a number of different symmetries. For symmetries representative of all classes in the famous 10-fold way of free fermion topological insulators/superconductors, we determine the stability to interactions. By combining with results on bosonic SPT phases we obtain a classification of electronic 3D SPT phases for these symmetries. In cases with a normal U(1) subgroup we show that this classification is complete. We describe the non-trivial surface and bulk properties of these states. In particular we discuss interesting correlated surface states that are not captured in a free fermion description. We show that in many, but not all cases, the surface can be gapped while preserving symmetry if it develops intrinsic topological order.

preprint2015arXiv

Photon-counting Brillouin optical time-domain reflectometry based on up-conversion detector and fiber Fabry-Perot scanning interferometer

A direct-detection Brillouin optical time-domain reflectometry (BOTDR) is proposed and demonstrated by using an up-conversion single-photon detector and a fiber Fabry-Perot scanning interferometer (FFP-SI). Taking advantage of high signal-to-noise ratio of the detector and high spectrum resolution of the FFP-SI, the Brillouin spectrum along a polarization maintaining fiber (PMF) is recorded on a multiscaler with a small data size directly. In contrast with conventional BOTDR adopting coherent detection, photon-counting BOTDR is simpler in structure and easier in data processing. In the demonstration experiment, characteristic parameters of the Brillouin spectrum including its power, spectral width and frequency center are analyzed simultaneously along a 10 km PMF at different temperature and stain conditions.

preprint2015arXiv

Scalable Modeling of Conversational-role based Self-presentation Characteristics in Large Online Forums

Online discussion forums are complex webs of overlapping subcommunities (macrolevel structure, across threads) in which users enact different roles depending on which subcommunity they are participating in within a particular time point (microlevel structure, within threads). This sub-network structure is implicit in massive collections of threads. To uncover this structure, we develop a scalable algorithm based on stochastic variational inference and leverage topic models (LDA) along with mixed membership stochastic block (MMSB) models. We evaluate our model on three large-scale datasets, Cancer-ThreadStarter (22K users and 14.4K threads), Cancer-NameMention(15.1K users and 12.4K threads) and StackOverFlow (1.19 million users and 4.55 million threads). Qualitatively, we demonstrate that our model can provide useful explanations of microlevel and macrolevel user presentation characteristics in different communities using the topics discovered from posts. Quantitatively, we show that our model does better than MMSB and LDA in predicting user reply structure within threads. In addition, we demonstrate via synthetic data experiments that the proposed active sub-network discovery model is stable and recovers the original parameters of the experimental setup with high probability.

preprint2015arXiv

Topological Paramagnetism in Frustrated Spin-One Mott Insulators

Time reversal protected three dimensional (3D) topological paramagnets are magnetic analogs of the celebrated 3D topological insulators. Such paramagnets have a bulk gap, no exotic bulk excitations, but non-trivial surface states protected by symmetry. We propose that frustrated spin-1 quantum magnets are a natural setting for realising such states in 3D. We describe a physical picture of the ground state wavefunction for such a spin-1 topological paramagnet in terms of loops of fluctuating Haldane chains with non-trivial linking phases. We illustrate some aspects of such loop gases with simple exactly solvable models. We also show how 3D topological paramagnets can be very naturally accessed within a slave particle description of a spin-1 magnet. Specifically we construct slave particle mean field states which are naturally driven into the topological paramagnet upon including fluctuations. We propose bulk projected wave functions for the topological paramagnet based on this slave particle description. An alternate slave particle construction leads to a stable U(1) quantum spin liquid from which a topological paramagnet may be accessed by condensing the emergent magnetic monopole excitation of the spin liquid.

preprint2014arXiv

Asymptotically Exact, Embarrassingly Parallel MCMC

Communication costs, resulting from synchronization requirements during learning, can greatly slow down many parallel machine learning algorithms. In this paper, we present a parallel Markov chain Monte Carlo (MCMC) algorithm in which subsets of data are processed independently, with very little communication. First, we arbitrarily partition data onto multiple machines. Then, on each machine, any classical MCMC method (e.g., Gibbs sampling) may be used to draw samples from a posterior distribution given the data subset. Finally, the samples from each machine are combined to form samples from the full posterior. This embarrassingly parallel algorithm allows each machine to act independently on a subset of the data (without communication) until the final combination stage. We prove that our algorithm generates asymptotically exact samples and empirically demonstrate its ability to parallelize burn-in and sampling in several models.

preprint2014arXiv

Gapped Symmetry Preserving Surface-State for the Electron Topological Insulator

It is well known that the 3D electronic topological insulator (TI) with charge-conservation and time-reversal symmetry cannot have a trivial insulating surface that preserves symmetry. It is often implicitly assumed that if the TI surface preserves both symmetries then it must be gapless. Here we show that it is possible for the TI surface to be both gapped and symmetry-preserving, at the expense of having surface-topological order. In contrast to analogous bosonic topological insulators, this symmetric surface topological order is intrinsically non-Abelian. We show that the surface-topological order provides a complete non-perturbative definition of the electron TI that transcends a free-particle band-structure picture, and could provide a useful perspective for studying strongly correlated topological Mott insulators.

preprint2014arXiv

How are identifiers named in open source software? About popularity and consistency

With the rapid increasing of software project size and maintenance cost, adherence to coding standards especially by managing identifier naming, is attracting a pressing concern from both computer science educators and software managers. Software developers mainly use identifier names to represent the knowledge recorded in source code. However, the popularity and adoption consistency of identifier naming conventions have not been revealed yet in this field. Taking forty-eight popular open source projects written in three top-ranking programming languages Java, C and C++ as examples, an identifier extraction tool based on regular expression matching is developed. In the subsequent investigation, some interesting findings are obtained. For the identifier naming popularity, it is found that Camel and Pascal naming conventions are leading the road while Hungarian notation is vanishing. For the identifier naming consistency, we have found that the projects written in Java have a much better performance than those written in C and C++. Finally, academia and software industry are urged to adopt the most popular naming conventions consistently in their practices so as to lead the identifier naming to a standard, unified and high-quality road.

preprint2014arXiv

Long-range micro-pulse aerosol lidar at 1.5 um with an up-conversion single-photon detector

A micro-pulse lidar at eye-safe wavelength is constructed based on an up-conversion single-photon detector. The ultralow noise detector enables using integration technique to improve the signal-to-noise ratio of the atmospheric backscattering even at daytime. With the pulse energy of 110uJ, the pulse repetition rate of 15 kHz, the optical antenna diameter of 100 mm and integration time of 5 minutes, a horizontal detection range of 7 km is realized. In the demonstration experiment, atmospheric visibility over 24 hours is monitored continuously, with results in accordance with the weather forecasts.

preprint2014arXiv

Nested Hierarchical Dirichlet Processes

We develop a nested hierarchical Dirichlet process (nHDP) for hierarchical topic modeling. The nHDP is a generalization of the nested Chinese restaurant process (nCRP) that allows each word to follow its own path to a topic node according to a document-specific distribution on a shared tree. This alleviates the rigid, single-path formulation of the nCRP, allowing a document to more easily express thematic borrowings as a random effect. We derive a stochastic variational inference algorithm for the model, in addition to a greedy subtree selection method for each document, which allows for efficient inference using massive collections of text documents. We demonstrate our algorithm on 1.8 million documents from The New York Times and 3.3 million documents from Wikipedia.

preprint2014arXiv

Soft vibrational mode associated with incommensurate orbital order in multiferroic CaMn$_7$O$_{12}$

We report inelastic light scattering measurements of lattice dynamics related to the incommensurate orbital order in $\mathrm{CaMn_7O_{12}}$. Below the ordering temperature $T_\mathrm{o} \approx 250 \,\mathrm{K}$, we observe extra phonon peaks as a result of Brillouin-zone folding, as well as a soft vibrational mode with a power-law $T$-dependent energy, $Ω= Ω_{0}(1 - T/T_{\mathrm{o}})^{1/2}$. This temperature dependence demonstrates the second-order nature of the transition at $T_\mathrm{o}$, and it indicates that the soft mode can be regarded as the amplitude excitation of the composite order parameter. Our result strongly suggests that the lattice degrees of freedom are actively involved in the orbital-ordering mechanism.

preprint2013arXiv

A Nested HDP for Hierarchical Topic Models

preprint2013arXiv

Boson topological insulators: A window into highly entangled quantum phases

We study several aspects of the realization of global symmetries in highly entangled phases of quantum matter. Examples include gapped topological ordered phases, gapless quantum spin liquids and non-fermi liquid phases. An insightful window into such phases is provided by recent developments in the theory of short ranged entangled Symmetry Protected Topological (SPT) phases . First they generate useful no-go constraints on how global symmetry may be implemented in a highly entangled phase. Possible symmetry implementation in gapped topological phases and some proposed gapless spin/bose liquids are examined in this light. We show that some previously proposed spin liquid states for 2d quantum magnets do not in fact have consistent symmetry implementation unless they occur as the surface of a 3d SPT phase. A second SPT-based insight into highly entangled states is the development of a view point of such states as SPT phases of one of the emergent excitations. We describe this in the specific context of time reversal symmetric 3d U(1) quantum spin liquids with an emergent photon. Different such spin liquids are shown to be equivalent to different SPT insulating phases of the emergent monopole excitation of such phases. The highly entangled states also in turn enrich our understanding of SPT phases. We use the insights obtained from our results to provide an explicit construction of bosonic SPT phases in 3d in a system of coupled layers. This includes construction of a time reversal symmetric SPT state that is not currently part of the cohomology classification of such states.

preprint2013arXiv

Classification of interacting electronic topological insulators in three dimensions

A fundamental open problem in condensed matter physics is how the dichotomy between conventional and topological band insulators is modified in the presence of strong electron interactions. We show that there are 6 new electronic topological insulators that have no non-interacting counterpart. Combined with the previously known band-insulators, these produce a total of 8 topologically distinct phases. Two of the new topological insulators have a simple physical description as Mott insulators in which the electron spins form spin analogs of the familiar topological band-insulator. The remaining are obtained as combinations of these two `topological paramagnets' and the topological band insulator. We prove that these 8 phases form a complete list of all possible interacting topological insulators, and are classified by a Z_2^3 group-structure. Experimental signatures are also discussed for these phases.

preprint2013arXiv

Community Specific Temporal Topic Discovery from Social Media

Studying temporal dynamics of topics in social media is very useful to understand online user behaviors. Most of the existing work on this subject usually monitors the global trends, ignoring variation among communities. Since users from different communities tend to have varying tastes and interests, capturing community-level temporal change can improve the understanding and management of social content. Additionally, it can further facilitate the applications such as community discovery, temporal prediction and online marketing. However, this kind of extraction becomes challenging due to the intricate interactions between community and topic, and intractable computational complexity. In this paper, we take a unified solution towards the community-level topic dynamic extraction. A probabilistic model, CosTot (Community Specific Topics-over-Time) is proposed to uncover the hidden topics and communities, as well as capture community-specific temporal dynamics. Specifically, CosTot considers text, time, and network information simultaneously, and well discovers the interactions between community and topic over time. We then discuss the approximate inference implementation to enable scalable computation of model parameters, especially for large social data. Based on this, the application layer support for multi-scale temporal analysis and community exploration is also investigated. We conduct extensive experimental studies on a large real microblog dataset, and demonstrate the superiority of proposed model on tasks of time stamp prediction, link prediction and topic perplexity.

preprint2013arXiv

Longitudinal spin excitations and magnetic anisotropy in antiferromagnetically ordered BaFe2As2

We report on a spin-polarized inelastic neutron scattering study of spin waves in the antiferromagnetically ordered state of BaFe2As2. Three distinct excitation components are identified, with spins fluctuating along the c-axis, perpendicular to the ordering direction in the ab-plane, and parallel to the ordering direction. While the first two "transverse" components can be described by a linear spin-wave theory with magnetic anisotropy and inter-layer coupling, the third "longitudinal" component is generically incompatible with the local moment picture. It points towards a contribution of itinerant electrons to the magnetism already in the parent compound of this family of Fe-based superconductors.

preprint2013arXiv

Stochastic Variational Inference

We develop stochastic variational inference, a scalable algorithm for approximating posterior distributions. We develop this technique for a large class of probabilistic models and we demonstrate it with two probabilistic topic models, latent Dirichlet allocation and the hierarchical Dirichlet process topic model. Using stochastic variational inference, we analyze several large collections of documents: 300K articles from Nature, 1.8M articles from The New York Times, and 3.8M articles from Wikipedia. Stochastic inference can easily handle data sets of this size and outperforms traditional variational inference, which can only handle a smaller subset. (We also show that the Bayesian nonparametric topic model outperforms its parametric counterpart.) Stochastic variational inference lets us apply complex Bayesian models to massive data sets.

preprint2013arXiv

Variational Inference in Nonconjugate Models

Mean-field variational methods are widely used for approximate posterior inference in many probabilistic models. In a typical application, mean-field methods approximately compute the posterior with a coordinate-ascent optimization algorithm. When the model is conditionally conjugate, the coordinate updates are easily derived and in closed form. However, many models of interest---like the correlated topic model and Bayesian logistic regression---are nonconjuate. In these models, mean-field methods cannot be directly applied and practitioners have had to develop variational algorithms on a case-by-case basis. In this paper, we develop two generic methods for nonconjugate models, Laplace variational inference and delta method variational inference. Our methods have several advantages: they allow for easily derived variational algorithms with a wide class of nonconjugate models; they extend and unify some of the existing algorithms that have been derived for specific models; and they work well on real-world datasets. We studied our methods on the correlated topic model, Bayesian logistic regression, and hierarchical Bayesian logistic regression.

preprint2012arXiv

A Split-Merge MCMC Algorithm for the Hierarchical Dirichlet Process

The hierarchical Dirichlet process (HDP) has become an important Bayesian nonparametric model for grouped data, such as document collections. The HDP is used to construct a flexible mixed-membership model where the number of components is determined by the data. As for most Bayesian nonparametric models, exact posterior inference is intractable---practitioners use Markov chain Monte Carlo (MCMC) or variational inference. Inspired by the split-merge MCMC algorithm for the Dirichlet process (DP) mixture model, we describe a novel split-merge MCMC sampling algorithm for posterior inference in the HDP. We study its properties on both synthetic data and text corpora. We find that split-merge MCMC for the HDP can provide significant improvements over traditional Gibbs sampling, and we give some understanding of the data properties that give rise to larger improvements.

preprint2012arXiv

Latent Collaborative Retrieval

Retrieval tasks typically require a ranking of items given a query. Collaborative filtering tasks, on the other hand, learn to model user's preferences over items. In this paper we study the joint problem of recommending items to a user with respect to a given query, which is a surprisingly common task. This setup differs from the standard collaborative filtering one in that we are given a query x user x item tensor for training instead of the more traditional user x item matrix. Compared to document retrieval we do have a query, but we may or may not have content features (we will consider both cases) and we can also take account of the user's profile. We introduce a factorized model for this new task that optimizes the top-ranked items returned for the given query and user. We report empirical results where it outperforms several baselines.

preprint2012arXiv

The Discrete Infinite Logistic Normal Distribution

We present the discrete infinite logistic normal distribution (DILN), a Bayesian nonparametric prior for mixed membership models. DILN is a generalization of the hierarchical Dirichlet process (HDP) that models correlation structure between the weights of the atoms at the group level. We derive a representation of DILN as a normalized collection of gamma-distributed random variables, and study its statistical properties. We consider applications to topic modeling and derive a variational inference algorithm for approximate posterior inference. We study the empirical performance of the DILN topic model on four corpora, comparing performance with the HDP and the correlated topic model (CTM). To deal with large-scale data sets, we also develop an online inference algorithm for DILN and compare with online HDP and online LDA on the Nature magazine, which contains approximately 350,000 articles.

Chong Wang

What is connected

Connect this record

See the researcher in context

Building this map preview

84 published item(s)

Bridging Generation and Training: A Systematic Review of Quality Issues in LLMs for Code

CheXTemporal: A Dataset for Temporally-Grounded Reasoning in Chest Radiography

Improving Flexible Image Tokenizers for Autoregressive Image Generation

Interleaved Reasoning for Large Language Models via Reinforcement Learning

Enhancement of Ising superconductivity in monolayer NbSe$_2$ via surface fluorination

Asymmetric Co-teaching with Multi-view Consensus for Noisy Label Learning

Knowledge Distillation to Ensemble Global and Interpretable Prototype-Based Mammogram Classification Models

Contrastive Transformer-based Multiple Instance Learning for Weakly Supervised Polyp Frame Detection

Differentially Private AUC Computation in Vertical Federated Learning

Differentially Private Label Protection in Split Learning

Differentially Private Multi-Party Data Release for Linear Regression

Edge physics at the deconfined transition between a quantum spin Hall insulator and a superconductor

Label Leakage and Protection from Forward Embedding in Vertical Federated Learning

Label Leakage and Protection in Two-party Split Learning

Learning to Simulate Unseen Physical Systems with Graph Neural Networks

Light-Induced Ferromagnetism in Moiré Superlattices

Linear Complexity Randomized Self-attention Mechanism

Non-zero momentum requires long-range entanglement

NVUM: Non-Volatile Unbiased Memory for Robust Medical Image Classification

REPNP: Plug-and-Play with Deep Reinforcement Learning Prior for Robust Image Restoration

The average distance problem with an Euler elastica penalization

The average distance problem with perimeter-to-area ratio penalization

A Künneth Formula of Hypergraphs

Abrupt switching of the anomalous Hall effect by field-rotation in nonmagnetic ZrTe5

Electric control of a canted-antiferromagnetic Chern insulator

Intrinsic nonlinear Hall effect in antiferromagnetic tetragonal CuMnAs

Multi-Domain Neural Machine Translation with Word-Level Adaptive Layer-wise Domain Mixing

Plasmons in the van der Waals charge-density-wave material 2H-TaSe2

Symmetry-adapted graph neural networks for constructing molecular dynamics force fields

Tunable plasmons in large area WTe2 thin films

A Discrete Morse Theory for Digraphs

A Discrete Morse Theory for Hypergraphs

AutoEmb: Automated Embedding Dimensionality Search in Streaming Recommendations

Correlated Fragile Topology: a Parton Approach

Enhancement of superconductivity in organic-inorganic hybrid topological materials

Fractional Quantum Hall Effect in Weyl Semimetals

Learning to Structure Long-term Dependence for Sequential Recommendation

Self-formed 2D/3D Heterostructure on the Edge of 2D Ruddlesden-Popper Hybrid Perovskites Responsible for Intriguing Optoelectronic Properties and Higher Cell Efficiency

Stacking Domain Wall Magnons in Twisted van der Waals Magnets

The optical conductivity of few-layer black phosphorus by infrared spectroscopy

The optical properties of few-layer InSe

Towards Earnings Call and Stock Price Movement

Tunable Graphene Split-Ring Resonators

Type-II Ising superconductivity and anomalous metallic state in macro-size ambient-stable ultrathin crystalline films

Van der Waals thin films of WTe2 for natural hyperbolic plasmonic surfaces

A theory of deconfined pseudo-criticality

Crystallographic splitting theorem for band representations and fragile topological photonic crystals

From spinon band topology to the symmetry quantum numbers of monopoles in Dirac spin liquids

Short-Term Temporal Convolutional Networks for Dynamic Hand Gesture Recognition

Type-II Ising Pairing in Few-Layer Stanene

Duality between $(2+1)d$ Quantum Critical Points

A Duality Web in 2+1 Dimensions and Condensed Matter Physics

A General Method for Robust Bayesian Modeling

All-fiber upconversion high spectral resolution wind lidar using a Fabry-Perot interferometer

Composite fermi liquids in the lowest Landau level

Micro-pulse upconversion Doppler lidar for wind and visibility detection in the atmospheric boundary layer

Time-reversal symmetric U(1) quantum spin liquids

Bound states of three fermions forming symmetry-protected topological phases

Continuous Time Dynamic Topic Models

Deep Speech 2: End-to-End Speech Recognition in English and Mandarin

Dual Dirac liquid on the surface of the electron topological insulator

Embarrassingly Parallel Variational Inference in Nonconjugate Models

Fractionalized gapless quantum vortex liquids

Half-filled Landau level, topological insulator surfaces, and three dimensional quantum spin liquids

Interacting fermionic topological insulators/superconductors in three dimensions

Photon-counting Brillouin optical time-domain reflectometry based on up-conversion detector and fiber Fabry-Perot scanning interferometer

Scalable Modeling of Conversational-role based Self-presentation Characteristics in Large Online Forums

Topological Paramagnetism in Frustrated Spin-One Mott Insulators

Asymptotically Exact, Embarrassingly Parallel MCMC

Gapped Symmetry Preserving Surface-State for the Electron Topological Insulator

How are identifiers named in open source software? About popularity and consistency

Long-range micro-pulse aerosol lidar at 1.5 um with an up-conversion single-photon detector

Nested Hierarchical Dirichlet Processes

Soft vibrational mode associated with incommensurate orbital order in multiferroic CaMn$_7$O$_{12}$