arXiv subjects

Sheng Zhang

Publications and source records attributed to Sheng Zhang.

At least 181 records · Page 10Linked to original sources

arXiv2023

Benchmarking Diverse-Modal Entity Linking with Generative Models

Entities can be expressed in diverse formats, such as texts, images, or column names and cell values in tables. While existing entity linking (EL) models work well on per modality configuration, such as text-only EL, visual grounding, or schema linking, it is more challenging to design a unified model for diverse modality configurations. To bring various modality configurations together, we constructed a benchmark for diverse-modal EL (DMEL) from existing EL datasets, covering all three modalities including text, image, and table. To approach the DMEL task, we proposed a generative diverse-modal model (GDMM) following a multimodal-encoder-decoder paradigm. Pre-training \Model with rich corpora builds a solid foundation for DMEL without storing the entire KB for inference. Fine-tuning GDMM builds a stronger DMEL baseline, outperforming state-of-the-art task-specific EL models by 8.51 F1 score on average. Additionally, extensive error analyses are conducted to highlight the challenges of DMEL, facilitating future research on this task.

cs.CL↗

arXiv2023

DecAF: Joint Decoding of Answers and Logical Forms for Question Answering over Knowledge Bases

Question answering over knowledge bases (KBs) aims to answer natural language questions with factual information such as entities and relations in KBs. Previous methods either generate logical forms that can be executed over KBs to obtain final answers or predict answers directly. Empirical results show that the former often produces more accurate answers, but it suffers from non-execution issues due to potential syntactic and semantic errors in the generated logical forms. In this work, we propose a novel framework DecAF that jointly generates both logical forms and direct answers, and then combines the merits of them to get the final answers. Moreover, different from most of the previous methods, DecAF is based on simple free-text retrieval without relying on any entity linking tools -- this simplification eases its adaptation to different datasets. DecAF achieves new state-of-the-art accuracy on WebQSP, FreebaseQA, and GrailQA benchmarks, while getting competitive results on the ComplexWebQuestions benchmark.

cs.CL↗

arXiv2023

Machine learning for structure-property relationships: Scalability and limitations

We present a scalable machine learning (ML) framework for predicting intensive properties and particularly classifying phases of many-body systems. Scalability and transferability are central to the unprecedented computational efficiency of ML methods. In general, linear-scaling computation can be achieved through the divide and conquer approach, and the locality of physical properties is key to partitioning the system into sub-domains that can be solved separately. Based on the locality assumption, ML model is developed for the prediction of intensive properties of a finite-size block. Predictions of large-scale systems can then be obtained by averaging results of the ML model from randomly sampled blocks of the system. We show that the applicability of this approach depends on whether the block-size of the ML model is greater than the characteristic length scale of the system. In particular, in the case of phase identification across a critical point, the accuracy of the ML prediction is limited by the diverging correlation length. The two-dimensional Ising model is used to demonstrate the proposed framework. We obtain an intriguing scaling relation between the prediction accuracy and the ratio of ML block size over the spin-spin correlation length. Implications for practical applications are also discussed.

cond-mat.stat-mech↗

arXiv2023

BioGPT: Generative Pre-trained Transformer for Biomedical Text Generation and Mining

Pre-trained language models have attracted increasing attention in the biomedical domain, inspired by their great success in the general natural language domain. Among the two main branches of pre-trained language models in the general language domain, i.e., BERT (and its variants) and GPT (and its variants), the first one has been extensively studied in the biomedical domain, such as BioBERT and PubMedBERT. While they have achieved great success on a variety of discriminative downstream biomedical tasks, the lack of generation ability constrains their application scope. In this paper, we propose BioGPT, a domain-specific generative Transformer language model pre-trained on large scale biomedical literature. We evaluate BioGPT on six biomedical NLP tasks and demonstrate that our model outperforms previous models on most tasks. Especially, we get 44.98%, 38.42% and 40.76% F1 score on BC5CDR, KD-DTI and DDI end-to-end relation extraction tasks respectively, and 78.2% accuracy on PubMedQA, creating a new record. Our case study on text generation further demonstrates the advantage of BioGPT on biomedical literature to generate fluent descriptions for biomedical terms. Code is available at https://github.com/microsoft/BioGPT.

cs.CL↗

arXiv2023

Nonrelativistic and nonmagnetic control of terahertz charge currents via electrical anisotropy in RuO2 and IrO2

Precise and ultrafast control over photo-induced charge currents across nanoscale interfaces could lead to important applications in energy harvesting, ultrafast electronics, and coherent terahertz sources. Recent studies have shown that several relativistic mechanisms, including inverse spin-Hall effect, inverse Rashba-Edelstein effect and inverse spin-orbit-torque effect, can convert longitudinally injected spin-polarized currents from magnetic materials to transverse charge currents, thereby harnessing these currents for terahertz generation. However, these mechanisms typically require external magnetic fields and suffer from low spin-polarization rates and low efficiencies of relativistic spin-to-charge conversion. In this work, we present a novel nonrelativistic and nonmagnetic mechanism that directly utilizes the photo-excited high-density charge currents across the interface. We demonstrate that the electrical anisotropy of conductive oxides RuO2 and IrO2 can effectively deflect injected charge currents to the transverse direction, resulting in efficient and broadband terahertz radiation. Importantly, this new mechanism has the potential to offer much higher conversion efficiency compared to previous methods, as conductive materials with large electrical anisotropy are readily available, whereas further increasing the spin-Hall angle of heavy-metal materials would be challenging. Our new findings offer exciting possibilities for directly utilizing these photo-excited high-density currents across metallic interfaces for ultrafast electronics and terahertz spectroscopy.

physics.optics↗

arXiv2023

Regularized Shallow Image Prior for Electrical Impedance Tomography

Untrained Neural Network Prior (UNNP) based algorithms have gained increasing popularity in tomographic imaging, as they offer superior performance compared to hand-crafted priors and do not require training. UNNP-based methods usually rely on deep architectures which are known for their excellent feature extraction ability compared to shallow ones. Contrary to common UNNP-based approaches, we propose a regularized shallow image prior method that combines UNNP with hand-crafted prior for Electrical Impedance Tomography (EIT). Our approach employs a 3-layer Multi-Layer Perceptron (MLP) as the UNNP in regularizing 2D and 3D EIT inversion. We demonstrate the influence of two typical hand-crafted regularizations when representing the conductivity distribution with shallow MLPs. We show considerably improved EIT image quality compared to conventional regularization algorithms, especially in structure preservation. The results suggest that combining the shallow image prior and the hand-crafted regularization can achieve similar performance to the Deep Image Prior (DIP) but with less architectural dependency and complexity of the neural network.

eess.IV↗

arXiv2023

PromptCAL: Contrastive Affinity Learning via Auxiliary Prompts for Generalized Novel Category Discovery

Although existing semi-supervised learning models achieve remarkable success in learning with unannotated in-distribution data, they mostly fail to learn on unlabeled data sampled from novel semantic classes due to their closed-set assumption. In this work, we target a pragmatic but under-explored Generalized Novel Category Discovery (GNCD) setting. The GNCD setting aims to categorize unlabeled training data coming from known and novel classes by leveraging the information of partially labeled known classes. We propose a two-stage Contrastive Affinity Learning method with auxiliary visual Prompts, dubbed PromptCAL, to address this challenging problem. Our approach discovers reliable pairwise sample affinities to learn better semantic clustering of both known and novel classes for the class token and visual prompts. First, we propose a discriminative prompt regularization loss to reinforce semantic discriminativeness of prompt-adapted pre-trained vision transformer for refined affinity relationships.Besides, we propose contrastive affinity learning to calibrate semantic representations based on our iterative semi-supervised affinity graph generation method for semantically-enhanced supervision. Extensive experimental evaluation demonstrates that our PromptCAL method is more effective in discovering novel classes even with limited annotations and surpasses the current state-of-the-art on generic and fine-grained benchmarks (e.g., with nearly 11% gain on CUB-200, and 9% on ImageNet-100) on overall accuracy. Our code is available at https://github.com/sheng-eatamath/PromptCAL.

cs.CV↗

arXiv2023

Compositional Zero-Shot Domain Transfer with Text-to-Text Models

Label scarcity is a bottleneck for improving task performance in specialised domains. We propose a novel compositional transfer learning framework (DoT5 - domain compositional zero-shot T5) for zero-shot domain transfer. Without access to in-domain labels, DoT5 jointly learns domain knowledge (from MLM of unlabelled in-domain free text) and task knowledge (from task training on more readily available general-domain data) in a multi-task manner. To improve the transferability of task training, we design a strategy named NLGU: we simultaneously train NLG for in-domain label-to-data generation which enables data augmentation for self-finetuning and NLU for label prediction. We evaluate DoT5 on the biomedical domain and the resource-lean subdomain of radiology, focusing on NLI, text summarisation and embedding learning. DoT5 demonstrates the effectiveness of compositional transfer learning through multi-task learning. In particular, DoT5 outperforms the current SOTA in zero-shot transfer by over 7 absolute points in accuracy on RadNLI. We validate DoT5 with ablations and a case study demonstrating its ability to solve challenging NLI examples requiring in-domain expertise.

cs.CL↗

arXiv2023

Machine learning for phase ordering dynamics of charge density waves

We present a machine learning (ML) framework for large-scale dynamical simulations of charge density wave (CDW) states. The charge modulation in a CDW state is often accompanied by a concomitant structural distortion, and the adiabatic evolution of a CDW order is governed by the dynamics of the lattice distortion. Calculation of the electronic contribution to the driving forces, however, is computationally very expensive for large systems. Assuming the principle of locality for electron systems, a neural-network model is developed to accurately and efficiently predict local electronic forces with input from neighborhood configurations. Importantly, the ML model makes possible a linear complexity algorithm for dynamical simulations of CDWs. As a demonstration, we apply our approach to investigate the phase ordering dynamics of the Holstein model, a canonical system of CDW order. Our large-scale simulations uncover an intriguing growth of the CDW domains that deviates significantly from the expected Allen-Cahn law for phase ordering of Ising-type order parameter field. This anomalous domain-growth could be attributed to the complex structure of domain-walls in this system. Our work highlights the promising potential of ML-based force-field models for dynamical simulations of functional electronic materials.

cond-mat.str-el↗

arXiv2023

Optimizing Bi-Encoder for Named Entity Recognition via Contrastive Learning

We present a bi-encoder framework for named entity recognition (NER), which applies contrastive learning to map candidate text spans and entity types into the same vector representation space. Prior work predominantly approaches NER as sequence labeling or span classification. We instead frame NER as a representation learning problem that maximizes the similarity between the vector representations of an entity mention and its type. This makes it easy to handle nested and flat NER alike, and can better leverage noisy self-supervision signals. A major challenge to this bi-encoder formulation for NER lies in separating non-entity spans from entity mentions. Instead of explicitly labeling all non-entity spans as the same class $\texttt{Outside}$ ($\texttt{O}$) as in most prior methods, we introduce a novel dynamic thresholding loss. Experiments show that our method performs well in both supervised and distantly supervised settings, for nested and flat NER alike, establishing new state of the art across standard datasets in the general domain (e.g., ACE2004, ACE2005) and high-value verticals such as biomedicine (e.g., GENIA, NCBI, BC5CDR, JNLPBA). We release the code at github.com/microsoft/binder.

cs.CL↗

arXiv2023

Dr.Spider: A Diagnostic Evaluation Benchmark towards Text-to-SQL Robustness

Neural text-to-SQL models have achieved remarkable performance in translating natural language questions into SQL queries. However, recent studies reveal that text-to-SQL models are vulnerable to task-specific perturbations. Previous curated robustness test sets usually focus on individual phenomena. In this paper, we propose a comprehensive robustness benchmark based on Spider, a cross-domain text-to-SQL benchmark, to diagnose the model robustness. We design 17 perturbations on databases, natural language questions, and SQL queries to measure the robustness from different angles. In order to collect more diversified natural question perturbations, we utilize large pretrained language models (PLMs) to simulate human behaviors in creating natural questions. We conduct a diagnostic study of the state-of-the-art models on the robustness set. Experimental results reveal that even the most robust model suffers from a 14.0% performance drop overall and a 50.7% performance drop on the most challenging perturbation. We also present a breakdown analysis regarding text-to-SQL model designs and provide insights for improving model robustness.

cs.CL↗

arXiv2023

Particle migration in porous media: from the mesoscopic perspective

Convection-diffusion equation is used to describe particle migration process in many fields, while it is proposed based on the empirical Fick's law. In this paper, with the help of the percolation model, we theoretically investigate the particle migration law in porous media from the mesoscopic perspective, and base on the probabilistic migration characteristic of particles to strictly reformulate the convection-diffusion equation. Meanwhile the quantitative relations between the convection, diffusion coefficients and the mesoscopic parameters of particle-motion and percolation-configuration are revealed. Furthermore, via the Monte-Carlo numerical simulation, we verify the proposed mesoscopic particle migration theory and modify the expressions of convection, diffusion coefficients for global applicability. In addition, applicable qualification of the proposed mesoscopic theory is given, and relation between the blocking effect parameter and connecting probability of the percolation model is obtained.

cond-mat.soft↗

arXiv2022

Hydrogen isotope separation using graphene-based membranes in liquid water

Hydrogen isotope separation has been effectively achieved using gaseous H2/D2 filtered through graphene/Nafion composite membranes. Nevertheless, deuteron nearly does not exist in the form of gaseous D2 in nature but in liquid water. Thus, it is a more feasible way to separate and enrich deuterium from water. Herein we have successfully transferred monolayer graphene to a rigid and porous polymer substrate PITEM (polyimide tracked film), which could avoid the swelling problem of the Nafion substrate, as well as keep the integrity of graphene. Meanwhile, defects in large area of CVD graphene could be successfully repaired by interfacial polymerization resulting in high separation factor. Moreover, a new model was proposed for the proton transport mechanism through monolayer graphene based on the kinetic isotope effect (KIE). In this model, graphene plays the significant role in the H/D separation process by completely breaking the O-H/O-D bond, which can maximize the KIE leading to prompted H/D separation performance. This work suggests a promising application of using monolayer graphene in industry and proposes a pronounced understanding of proton transport in graphene

physics.chem-ph↗

arXiv2022

Importance of Synthesizing High-quality Data for Text-to-SQL Parsing

Recently, there has been increasing interest in synthesizing data to improve downstream text-to-SQL tasks. In this paper, we first examined the existing synthesized datasets and discovered that state-of-the-art text-to-SQL algorithms did not further improve on popular benchmarks when trained with augmented synthetic data. We observed two shortcomings: illogical synthetic SQL queries from independent column sampling and arbitrary table joins. To address these issues, we propose a novel synthesis framework that incorporates key relationships from schema, imposes strong typing, and conducts schema-distance-weighted column sampling. We also adopt an intermediate representation (IR) for the SQL-to-text task to further improve the quality of the generated natural language questions. When existing powerful semantic parsers are pre-finetuned on our high-quality synthesized data, our experiments show that these models have significant accuracy boosts on popular benchmarks, including new state-of-the-art performance on Spider.

cs.CL↗

arXiv2022

REKnow: Enhanced Knowledge for Joint Entity and Relation Extraction

Relation extraction is an important but challenging task that aims to extract all hidden relational facts from the text. With the development of deep language models, relation extraction methods have achieved good performance on various benchmarks. However, we observe two shortcomings of previous methods: first, there is no unified framework that works well under various relation extraction settings; second, effectively utilizing external knowledge as background information is absent. In this work, we propose a knowledge-enhanced generative model to mitigate these two issues. Our generative model is a unified framework to sequentially generate relational triplets under various relation extraction settings and explicitly utilizes relevant knowledge from Knowledge Graph (KG) to resolve ambiguities. Our model achieves superior performance on multiple benchmarks and settings, including WebNLG, NYT10, and TACRED.

cs.CL↗

arXiv2022

A Dual Accelerated Method for Online Stochastic Distributed Averaging: From Consensus to Decentralized Policy Evaluation

Motivated by decentralized sensing and policy evaluation problems, we consider a particular type of distributed stochastic optimization problem over a network, called the online stochastic distributed averaging problem. We design a dual-based method for this distributed consensus problem with Polyak--Ruppert averaging and analyze its behavior. We show that the proposed algorithm attains an accelerated deterministic error depending optimally on the condition number of the network, and also that it has an order-optimal stochastic error. This improves on the guarantees of state-of-the-art distributed stochastic optimization algorithms when specialized to this setting, and yields -- among other things -- corollaries for decentralized policy evaluation. Our proofs rely on explicitly studying the evolution of several relevant linear systems, and may be of independent interest. Numerical experiments are provided, which validate our theoretical results and demonstrate that our approach outperforms existing methods in finite-sample scenarios on several natural network topologies.

math.OC↗

arXiv2022

Shilling Black-box Recommender Systems by Learning to Generate Fake User Profiles

Due to the pivotal role of Recommender Systems (RS) in guiding customers towards the purchase, there is a natural motivation for unscrupulous parties to spoof RS for profits. In this paper, we study Shilling Attack where an adversarial party injects a number of fake user profiles for improper purposes. Conventional Shilling Attack approaches lack attack transferability (i.e., attacks are not effective on some victim RS models) and/or attack invisibility (i.e., injected profiles can be easily detected). To overcome these issues, we present Leg-UP, a novel attack model based on the Generative Adversarial Network. Leg-UP learns user behavior patterns from real users in the sampled ``templates'' and constructs fake user profiles. To simulate real users, the generator in Leg-UP directly outputs discrete ratings. To enhance attack transferability, the parameters of the generator are optimized by maximizing the attack performance on a surrogate RS model. To improve attack invisibility, Leg-UP adopts a discriminator to guide the generator to generate undetectable fake user profiles. Experiments on benchmarks have shown that Leg-UP exceeds state-of-the-art Shilling Attack methods on a wide range of victim RS models. The source code of our work is available at: https://github.com/XMUDM/ShillingAttack.

cs.IR↗

arXiv2022

Knowledge-Rich Self-Supervision for Biomedical Entity Linking

Entity linking faces significant challenges such as prolific variations and prevalent ambiguities, especially in high-value domains with myriad entities. Standard classification approaches suffer from the annotation bottleneck and cannot effectively handle unseen entities. Zero-shot entity linking has emerged as a promising direction for generalizing to new entities, but it still requires example gold entity mentions during training and canonical descriptions for all entities, both of which are rarely available outside of Wikipedia. In this paper, we explore Knowledge-RIch Self-Supervision ($\tt KRISS$) for biomedical entity linking, by leveraging readily available domain knowledge. In training, it generates self-supervised mention examples on unlabeled text using a domain ontology and trains a contextual encoder using contrastive learning. For inference, it samples self-supervised mentions as prototypes for each entity and conducts linking by mapping the test mention to the most similar prototype. Our approach can easily incorporate entity descriptions and gold mention labels if available. We conducted extensive experiments on seven standard datasets spanning biomedical literature and clinical notes. Without using any labeled information, our method produces $\tt KRISSBERT$, a universal entity linker for four million UMLS entities that attains new state of the art, outperforming prior self-supervised methods by as much as 20 absolute points in accuracy.

cs.CL↗