Search arXiv⌕ Search

arXiv subjects

Wei Wei

Publications and source records attributed to Wei Wei.

At least 289 records · Page 16Linked to original sources
arXiv2023

AttenWalker: Unsupervised Long-Document Question Answering via Attention-based Graph Walking

Annotating long-document question answering (long-document QA) pairs is time-consuming and expensive. To alleviate the problem, it might be possible to generate long-document QA pairs via unsupervised question answering (UQA) methods. However, existing UQA tasks are based on short documents, and can hardly incorporate long-range information. To tackle the problem, we propose a new task, named unsupervised long-document question answering (ULQA), aiming to generate high-quality long-document QA instances in an unsupervised manner. Besides, we propose AttenWalker, a novel unsupervised method to aggregate and generate answers with long-range dependency so as to construct long-document QA pairs. Specifically, AttenWalker is composed of three modules, i.e., span collector, span linker and answer aggregator. Firstly, the span collector takes advantage of constituent parsing and reconstruction loss to select informative candidate spans for constructing answers. Secondly, by going through the attention graph of a pre-trained long-document model, potentially interrelated text spans (that might be far apart) could be linked together via an attention-walking algorithm. Thirdly, in the answer aggregator, linked spans are aggregated into the final answer via the mask-filling ability of a pre-trained model. Extensive experiments show that AttenWalker outperforms previous methods on Qasper and NarrativeQA. In addition, AttenWalker also shows strong performance in the few-shot learning setting.

cs.CL↗
arXiv2023

Large upper critical fields and dimensionality crossover of superconductivity in infinite-layer nickelate La$_{0.8}$Sr$_{0.2}$NiO$_{2}$

The recently emerging superconductivity in infinite-layer nickelates, with isostructure and isoelectron of cuprates, provides a new platform to explore the pairing mechanism of high-temperature superconductors. In this work, we studied the upper critical field ($H_{\rm{c2}}$) of a high-quality La$_{0.8}$Sr$_{0.2}$NiO$_{2}$ thin film with superconducting transition temperature, $T_{\rm{c}}$ = 18.8 K, using high magnetic field up to 56 T. A very large $H_{\rm{c2}}$, $\sim$ 40 T for $H$ $\Arrowvert$ $c$ and $\sim$ 52 T for $H$ $\Arrowvert$ $ab$, was confirmed, which suggests that infinite-layer nickelates also have great application potential. The anisotropy of $H_{\rm{c2}}$ monotonically decreases from $\sim$ 10 near $T_{\rm{c}}$ to $\sim$ 1.5 at 2 K. Angle dependence of $H_{\rm{c2}}$ confirms the crossover of superconductivity from two-dimensional (2D) to three-dimensional (3D) as the temperature decreases. We discussed that the interstitial orbital effect causes the weakening of anisotropy. The observed abnormal upturning of $H_{\rm{c2}}$ at low temperatures is found to be a universal behavior independent of film quality and rare earth elements. Therefore, it should not be the Fulde-Ferrell-Larkin-Ovchinnikov (FFLO) state due to the fact that it is in the dirty limit and insensitive to disorder.

cond-mat.supr-con↗
arXiv2023

Knowledge-Enhanced Relation Extraction Dataset

Recently, knowledge-enhanced methods leveraging auxiliary knowledge graphs have emerged in relation extraction, surpassing traditional text-based approaches. However, to our best knowledge, there is currently no public dataset available that encompasses both evidence sentences and knowledge graphs for knowledge-enhanced relation extraction. To address this gap, we introduce the Knowledge-Enhanced Relation Extraction Dataset (KERED). KERED annotates each sentence with a relational fact, and it provides knowledge context for entities through entity linking. Using our curated dataset, We compared contemporary relation extraction methods under two prevalent task settings: sentence-level and bag-level. The experimental result shows the knowledge graphs provided by KERED can support knowledge-enhanced relation extraction methods. We believe that KERED offers high-quality relation extraction datasets with corresponding knowledge graphs for evaluating the performance of knowledge-enhanced relation extraction methods. Our dataset is available at: \url{https://figshare.com/projects/KERED/134459}

cs.LG↗
arXiv2023

Matching extension and matching exclusion via the size or the spectral radius of graphs

A graph $G$ is said to be $k$-extendable if every matching of size $k$ in $G$ can be extended to a perfect matching of $G$, where $k$ is a positive integer. We say $G$ is $1$-excludable if for every edge $e$ of $G$, there exists a perfect matching excluding $e$. In this paper, we first establish a lower bound on the size (resp. the spectral radius) of $G$ to guarantee that $G$ is $k$-extendable. Then we determine a lower bound on the size (resp. the spectral radius) of $G$ to guarantee that $G$ is $1$-excludable. All the corresponding extremal graphs are characterized.

math.CO↗
arXiv2023

Glocal Energy-based Learning for Few-Shot Open-Set Recognition

Few-shot open-set recognition (FSOR) is a challenging task of great practical value. It aims to categorize a sample to one of the pre-defined, closed-set classes illustrated by few examples while being able to reject the sample from unknown classes. In this work, we approach the FSOR task by proposing a novel energy-based hybrid model. The model is composed of two branches, where a classification branch learns a metric to classify a sample to one of closed-set classes and the energy branch explicitly estimates the open-set probability. To achieve holistic detection of open-set samples, our model leverages both class-wise and pixel-wise features to learn a glocal energy-based score, in which a global energy score is learned using the class-wise features, while a local energy score is learned using the pixel-wise features. The model is enforced to assign large energy scores to samples that are deviated from the few-shot examples in either the class-wise features or the pixel-wise features, and to assign small energy scores otherwise. Experiments on three standard FSOR datasets show the superior performance of our model.

cs.CV↗
arXiv2023

Progressive Transfer Learning for Dexterous In-Hand Manipulation with Multi-Fingered Anthropomorphic Hand

Dexterous in-hand manipulation for a multi-fingered anthropomorphic hand is extremely difficult because of the high-dimensional state and action spaces, rich contact patterns between the fingers and objects. Even though deep reinforcement learning has made moderate progress and demonstrated its strong potential for manipulation, it is still faced with certain challenges, such as large-scale data collection and high sample complexity. Especially, for some slight change scenes, it always needs to re-collect vast amounts of data and carry out numerous iterations of fine-tuning. Remarkably, humans can quickly transfer learned manipulation skills to different scenarios with little supervision. Inspired by human flexible transfer learning capability, we propose a novel dexterous in-hand manipulation progressive transfer learning framework (PTL) based on efficiently utilizing the collected trajectories and the source-trained dynamics model. This framework adopts progressive neural networks for dynamics model transfer learning on samples selected by a new samples selection method based on dynamics properties, rewards and scores of the trajectories. Experimental results on contact-rich anthropomorphic hand manipulation tasks show that our method can efficiently and effectively learn in-hand manipulation skills with a few online attempts and adjustment learning under the new scene. Compared to learning from scratch, our method can reduce training time costs by 95%.

cs.RO↗
arXiv2023

STAGE: Span Tagging and Greedy Inference Scheme for Aspect Sentiment Triplet Extraction

Aspect Sentiment Triplet Extraction (ASTE) has become an emerging task in sentiment analysis research, aiming to extract triplets of the aspect term, its corresponding opinion term, and its associated sentiment polarity from a given sentence. Recently, many neural networks based models with different tagging schemes have been proposed, but almost all of them have their limitations: heavily relying on 1) prior assumption that each word is only associated with a single role (e.g., aspect term, or opinion term, etc. ) and 2) word-level interactions and treating each opinion/aspect as a set of independent words. Hence, they perform poorly on the complex ASTE task, such as a word associated with multiple roles or an aspect/opinion term with multiple words. Hence, we propose a novel approach, Span TAgging and Greedy infErence (STAGE), to extract sentiment triplets in span-level, where each span may consist of multiple words and play different roles simultaneously. To this end, this paper formulates the ASTE task as a multi-class span classification problem. Specifically, STAGE generates more accurate aspect sentiment triplet extractions via exploring span-level information and constraints, which consists of two components, namely, span tagging scheme and greedy inference strategy. The former tag all possible candidate spans based on a newly-defined tagging set. The latter retrieves the aspect/opinion term with the maximum length from the candidate sentiment snippet to output sentiment triplets. Furthermore, we propose a simple but effective model based on the STAGE, which outperforms the state-of-the-arts by a large margin on four widely-used datasets. Moreover, our STAGE can be easily generalized to other pair/triplet extraction tasks, which also demonstrates the superiority of the proposed scheme STAGE.

cs.CL↗
arXiv2023

Generalized Anthropomorphic Functional Grasping with Minimal Demonstrations

This article investigates the challenge of achieving functional tool-use grasping with high-DoF anthropomorphic hands, with the aim of enabling anthropomorphic hands to perform tasks that require human-like manipulation and tool-use. However, accomplishing human-like grasping in real robots present many challenges, including obtaining diverse functional grasps for a wide variety of objects, handling generalization ability for kinematically diverse robot hands and precisely completing object shapes from a single-view perception. To tackle these challenges, we propose a six-step grasp synthesis algorithm based on fine-grained contact modeling that generates physically plausible and human-like functional grasps for category-level objects with minimal human demonstrations. With the contact-based optimization and learned dense shape correspondence, the proposed algorithm is adaptable to various objects in same category and a board range of robot hand models. To further demonstrate the robustness of the framework, over 10K functional grasps are synthesized to train our neural network, named DexFG-Net, which generates diverse sets of human-like functional grasps based on the reconstructed object model produced by a shape completion module. The proposed framework is extensively validated in simulation and on a real robot platform. Simulation experiments demonstrate that our method outperforms baseline methods by a large margin in terms of grasp functionality and success rate. Real robot experiments show that our method achieved an overall success rate of 79\% and 68\% for tool-use grasp on 3-D printed and real test objects, respectively, using a 5-Finger Schunk Hand. The experimental results indicate a step towards human-like grasping with anthropomorphic hands.

cs.RO↗
arXiv2023

Beyond Empirical Risk Minimization: Local Structure Preserving Regularization for Improving Adversarial Robustness

It is broadly known that deep neural networks are susceptible to being fooled by adversarial examples with perturbations imperceptible by humans. Various defenses have been proposed to improve adversarial robustness, among which adversarial training methods are most effective. However, most of these methods treat the training samples independently and demand a tremendous amount of samples to train a robust network, while ignoring the latent structural information among these samples. In this work, we propose a novel Local Structure Preserving (LSP) regularization, which aims to preserve the local structure of the input space in the learned embedding space. In this manner, the attacking effect of adversarial samples lying in the vicinity of clean samples can be alleviated. We show strong empirical evidence that with or without adversarial training, our method consistently improves the performance of adversarial robustness on several image classification datasets compared to the baselines and some state-of-the-art approaches, thus providing promising direction for future research.

cs.LG↗
arXiv2023

ALUM: Adversarial Data Uncertainty Modeling from Latent Model Uncertainty Compensation

It is critical that the models pay attention not only to accuracy but also to the certainty of prediction. Uncertain predictions of deep models caused by noisy data raise significant concerns in trustworthy AI areas. To explore and handle uncertainty due to intrinsic data noise, we propose a novel method called ALUM to simultaneously handle the model uncertainty and data uncertainty in a unified scheme. Rather than solely modeling data uncertainty in the ultimate layer of a deep model based on randomly selected training data, we propose to explore mined adversarial triplets to facilitate data uncertainty modeling and non-parametric uncertainty estimations to compensate for the insufficiently trained latent model layers. Thus, the critical data uncertainty and model uncertainty caused by noisy data can be readily quantified for improving model robustness. Our proposed ALUM is model-agnostic which can be easily implemented into any existing deep model with little extra computation overhead. Extensive experiments on various noisy learning tasks validate the superior robustness and generalization ability of our method. The code is released at https://github.com/wwzjer/ALUM.

cs.LG↗
arXiv2023

Significant enhancement of critical current density in H+-intercalated FeSe single crystal

Superconducting transition temperature (Tc) and critical current density (Jc) are two key factors that are not only crucial for probing high-temperature superconducting mechanisms, but also for practical applications. The simple crystal structure of FeSe is very favorable for the fabrication of thin films and wires, but its application is limited by the relatively low Tc and small Jc. A previous study has found that the Tc of FeSe can be significantly enhanced over 40 K by using the protonation method. Here, we present a systematic study of Jc and vortex properties of H+-intercalated FeSe (Hx-FeSe) single crystals. The value of Jc for Hx-FeSe single crystal is significantly enhanced, exceeding 1.3*10^6 A/cm2 at 4 K, which is more than two orders of magnitude larger than 1.1*10^4 A/cm2 of pristine FeSe. The vortex pinning mechanism of Hx-FeSe is found to be surface pinning, which is different from the dominant strong point-like pinning in pristine FeSe. Moreover, the systematic study of the vortex phase transition and the underlying mechanism provides a wealth of information for the vortex phase diagram of Hx-FeSe single crystal. Our results confirm that the introduction of H+ intercalations into FeSe not only enhances the Tc, but also significantly increases the value of Jc, which is favorable for practical applications.

cond-mat.supr-con↗
arXiv2023

Heterogeneous Graph Contrastive Learning for Recommendation

Graph Neural Networks (GNNs) have become powerful tools in modeling graph-structured data in recommender systems. However, real-life recommendation scenarios usually involve heterogeneous relationships (e.g., social-aware user influence, knowledge-aware item dependency) which contains fruitful information to enhance the user preference learning. In this paper, we study the problem of heterogeneous graph-enhanced relational learning for recommendation. Recently, contrastive self-supervised learning has become successful in recommendation. In light of this, we propose a Heterogeneous Graph Contrastive Learning (HGCL), which is able to incorporate heterogeneous relational semantics into the user-item interaction modeling with contrastive learning-enhanced knowledge transfer across different views. However, the influence of heterogeneous side information on interactions may vary by users and items. To move this idea forward, we enhance our heterogeneous graph contrastive learning with meta networks to allow the personalized knowledge transformer with adaptive contrastive augmentation. The experimental results on three real-world datasets demonstrate the superiority of HGCL over state-of-the-art recommendation methods. Through ablation study, key components in HGCL method are validated to benefit the recommendation performance improvement. The source code of the model implementation is available at the link https://github.com/HKUDS/HGCL.

cs.IR↗
arXiv2023

Learning Tri-mode Grasping for Ambidextrous Robot Picking

Object picking in cluttered scenes is a widely investigated field of robot manipulation, however, ambidextrous robot picking is still an important and challenging issue. We found the fusion of different prehensile actions (grasp and suction) can expand the range of objects that can be picked by robot, and the fusion of prehensile action and nonprehensile action (push) can expand the picking space of ambidextrous robot. In this paper, we propose a Push-Grasp-Suction (PGS) tri-mode grasping learning network for ambidextrous robot picking through the fusion of different prehensile actions and the fusion of prehensile action and nonprehensile aciton. The prehensile branch of PGS takes point clouds as input, and the 6-DoF picking configuration of grasp and suction in cluttered scenes are generated by multi-task point cloud learning. The nonprehensile branch with depth image input generates instance segmentation map and push configuration, cooperating with the prehensile actions to complete the picking of objects out of single-arm space. PGS generalizes well in real scene and achieves state-of-the-art picking performance.

cs.RO↗
arXiv2023

Yaglom's law and conserved quantity dissipation in turbulence

In this paper, we are concerned with the local exact relationship for third-order structure functions in the temperature equation, the inviscid MHD equations and the Euler equations in the sense of Duchon-Robert type and Eyink type. It is shown that the local version of Yaglom's $4/3$ law is valid for the dissipation rates of conserved quantities such as the energy, cross-helicity and helicity in these systems. In the spirit of Duchon-Robert's classical work, we derive the dissipation term resulted from the lack of smoothness of the solutions in corresponding conservation relation. It seems that these results suggest that the Yaglom's law of the hydrodynamic equations holds if an analogue of dissipation term as Duchon-Robert's is obtained. Base on this, the first Yaglom's relation for the Oldroyd-B model and, inspired by the very recent work due to Boutros-Titi, six new 4/3 laws for subgrid scale $α$-models of turbulence are also presented.

math.AP↗
arXiv2023

Straightforward Bias and Frequency Dependent Small-Signal Model Extraction for Single-Layer Graphene FETs

We propose an explicit small-signal graphene field-effect transistor (GFET) parameter extraction procedure based on a charge-based quasi-static model. The dependence of the small-signal parameters on both gate voltage and frequency is precisely validated by high-frequency (up to 18 GHz) on-wafer measurements from a 300 nm device. These parameters are studied simultaneously, in contrast to other works which focus exclusively on few. Efficient procedures have been applied to GFETs for the first time to remove contact and gate resistances from the Y-parameters. The use of these methods yields straightforward equations for extracting the small-signal model parameters, which is extremely useful for radio-frequency circuit design. Furthermore, we show for the first time experimental validation vs. both gate voltage and frequency of the intrinsic GFET non-reciprocal capacitance model. Accurate models are also presented for the gate voltage-dependence of the measured unity-gain and maximum oscillation frequencies as well as of the current and power gains.

cond-mat.mes-hall↗
arXiv2023

Extensions on spectral extrema of $C_5/C_6$-free graphs with given size

Let $\mathcal{F}$ denote a set of graphs. A graph $G$ is said to be $\mathcal{F}$-free if it does not contain any element of $\mathcal{F}$ as a subgraph. The Turán number is the maximum possible number of edges in an $\mathcal{F}$-free graph with $n$ vertices. It is well known that classical Turán type extremal problem aims to study the Turán number of fixed graphs. In 2010, Nikiforov \cite{Nik2} proposed analogously a spectral Turán type problem which asks to determine the maximum spectral radius of an $\mathcal{F}$-free graph with $n$ vertices. It attracts much attention and many such problems remained elusive open even after serious attempts, and so they are considered as one of the most intriguing problems in spectral extremal graph theory. It is interesting to consider another spectral Turán type problem which asks to determine the maximum spectral radius of an $\mathcal{F}$-free graph with $m$ edges. Denote by $\mathcal{G}(m,\mathcal{F})$ the set of $\mathcal{F}$-free graphs with $m$ edges having no isolated vertices. Each of the graphs among $\mathcal{G}(m,\mathcal{F})$ having the largest spectral radius is called a maximal graph. Let $θ_{p,q,r}$ be a theta graph formed by connecting two distinct vertices with three independent paths of length $p,q$ and $r,$ respectively (length refers to the number of edges). In this paper, we firstly determine the unique maximal graph among $\mathcal{G}(m,θ_{1,2,3})$ and $\mathcal{G}(m,θ_{1,2,4}),$ respectively. Then we determine all the maximal graphs among $\mathcal{G}(m,C_5)$ (resp. $\mathcal{G}(m,C_6)$) excluding the book graph. These results extend some earlier results.

math.CO↗
arXiv2023

Mind the Gap: Polishing Pseudo labels for Accurate Semi-supervised Object Detection

Exploiting pseudo labels (e.g., categories and bounding boxes) of unannotated objects produced by a teacher detector have underpinned much of recent progress in semi-supervised object detection (SSOD). However, due to the limited generalization capacity of the teacher detector caused by the scarce annotations, the produced pseudo labels often deviate from ground truth, especially those with relatively low classification confidences, thus limiting the generalization performance of SSOD. To mitigate this problem, we propose a dual pseudo-label polishing framework for SSOD. Instead of directly exploiting the pseudo labels produced by the teacher detector, we take the first attempt at reducing their deviation from ground truth using dual polishing learning, where two differently structured polishing networks are elaborately developed and trained using synthesized paired pseudo labels and the corresponding ground truth for categories and bounding boxes on the given annotated objects, respectively. By doing this, both polishing networks can infer more accurate pseudo labels for unannotated objects through sufficiently exploiting their context knowledge based on the initially produced pseudo labels, and thus improve the generalization performance of SSOD. Moreover, such a scheme can be seamlessly plugged into the existing SSOD framework for joint end-to-end learning. In addition, we propose to disentangle the polished pseudo categories and bounding boxes of unannotated objects for separate category classification and bounding box regression in SSOD, which enables introducing more unannotated objects during model training and thus further improve the performance. Experiments on both PASCAL VOC and MS COCO benchmarks demonstrate the superiority of the proposed method over existing state-of-the-art baselines.

cs.CV↗
arXiv2023

Privacy Preserving Ultra-Short-term Wind Power Prediction Based on Secure Multi Party Computation

Mining the spatial and temporal correlation of wind farm output data is beneficial for enhancing the precision of ultra-short-term wind power prediction. However, if the wind farms are owned by separate entities, they may be reluctant to share their data directly due to privacy concerns as well as business management regulation policies. Although cryptographic approaches have been designed to protect privacy in the process of data sharing, it is still a challenging problem to encrypt the original data while extracting the nonlinear relationship among multiple wind farms in the machine learning process. This paper presents pwXGBoost, a technique based on the machine learning tree model and secure multi-party computation (SMPC) that can successfully extract complicated relationships while preserving data privacy. A maximum mean discrepancy (MMD) based scheme is proposed to effectively choose adjacent candidate wind farms to participate in the collaborative model training, therefore improving the accuracy and reducing the burden of data acquisition. The proposed method was evaluated on real world data collected from a cluster of wind farms in Inner Mongolia, China, demonstrating that it is capable of achieving considerable efficiency and performance improvements while preserving privacy

cs.CR↗