arXiv subjects

Tianyi Li

Publications and source records attributed to Tianyi Li.

At least 127 records · Page 7Linked to original sources

arXiv2022

ClusterEA: Scalable Entity Alignment with Stochastic Training and Normalized Mini-batch Similarities

Entity alignment (EA) aims at finding equivalent entities in different knowledge graphs (KGs). Embedding-based approaches have dominated the EA task in recent years. Those methods face problems that come from the geometric properties of embedding vectors, including hubness and isolation. To solve these geometric problems, many normalization approaches have been adopted for EA. However, the increasing scale of KGs renders it hard for EA models to adopt the normalization processes, thus limiting their usage in real-world applications. To tackle this challenge, we present ClusterEA, a general framework that is capable of scaling up EA models and enhancing their results by leveraging normalization methods on mini-batches with a high entity equivalent rate. ClusterEA contains three components to align entities between large-scale KGs, including stochastic training, ClusterSampler, and SparseFusion. It first trains a large-scale Siamese GNN for EA in a stochastic fashion to produce entity embeddings. Based on the embeddings, a novel ClusterSampler strategy is proposed for sampling highly overlapped mini-batches. Finally, ClusterEA incorporates SparseFusion, which normalizes local and global similarity and then fuses all similarity matrices to obtain the final similarity matrix. Extensive experiments with real-life datasets on EA benchmarks offer insight into the proposed framework, and suggest that it is capable of outperforming the state-of-the-art scalable EA framework by up to 8 times in terms of Hits@1.

cs.DB↗

arXiv2022

Event Detection Explorer: An Interactive Tool for Event Detection Exploration

Event Detection (ED) is an important task in natural language processing. In the past few years, many datasets have been introduced for advancing ED machine learning models. However, most of these datasets are under-explored because not many tools are available for people to study events, trigger words, and event mention instances systematically and efficiently. In this paper, we present an interactive and easy-to-use tool, namely ED Explorer, for ED dataset and model exploration. ED Explorer consists of an interactive web application, an API, and an NLP toolkit, which can help both domain experts and non-experts to better understand the ED task. We use ED Explorer to analyze a recent proposed large-scale ED datasets (referred to as MAVEN), and discover several underlying problems, including sparsity, label bias, label imbalance, and debatable annotations, which provide us with directions to improve the MAVEN dataset. The ED Explorer can be publicly accessed through http://edx.leafnlp.org/. The demonstration video is available here https://www.youtube.com/watch?v=6QPnxPwxg50.

cs.CL↗

arXiv2022

Unimon qubit

Superconducting qubits are one of the most promising candidates to implement quantum computers. The superiority of superconducting quantum computers over any classical device in simulating random but well-determined quantum circuits has already been shown in two independent experiments and important steps have been taken in quantum error correction. However, the currently wide-spread qubit designs do not yet provide high enough performance to enable practical applications or efficient scaling of logical qubits owing to one or several following issues: sensitivity to charge or flux noise leading to decoherence, too weak non-linearity preventing fast operations, undesirably dense excitation spectrum, or complicated design vulnerable to parasitic capacitance. Here, we introduce and demonstrate a superconducting-qubit type, the unimon, which combines the desired properties of high non-linearity, full insensitivity to dc charge noise, insensitivity to flux noise, and a simple structure consisting only of a single Josephson junction in a resonator. We measure the qubit frequency, $ω_{01}/(2π)$, and anharmonicity $α$ over the full dc-flux range and observe, in agreement with our quantum models, that the qubit anharmonicity is greatly enhanced at the optimal operation point, yielding, for example, 99.9% and 99.8% fidelity for 13-ns single-qubit gates on two qubits with $(ω_{01},α)=(4.49~\mathrm{GHz}, 434~\mathrm{ MHz})\times 2π$ and $(3.55~\mathrm{GHz}, 744~\mathrm{ MHz})\times 2π$, respectively. The energy relaxation time $T_1\lesssim 10~μ\mathrm{s}$ is stable for hours and seems to be limited by dielectric losses. Thus, future improvements of the design, materials, and gate time may promote the unimon to break the 99.99% fidelity target for efficient quantum error correction and possible quantum advantage with noisy systems.

quant-ph↗

arXiv2022

Cross-lingual Inference with A Chinese Entailment Graph

Predicate entailment detection is a crucial task for question-answering from text, where previous work has explored unsupervised learning of entailment graphs from typed open relation triples. In this paper, we present the first pipeline for building Chinese entailment graphs, which involves a novel high-recall open relation extraction (ORE) method and the first Chinese fine-grained entity typing dataset under the FIGER type ontology. Through experiments on the Levy-Holt dataset, we verify the strength of our Chinese entailment graph, and reveal the cross-lingual complementarity: on the parallel Levy-Holt dataset, an ensemble of Chinese and English entailment graphs outperforms both monolingual graphs, and raises unsupervised SOTA by 4.7 AUC points.

cs.CL↗

arXiv2022

Study of Nonlinear Interaction between Waves and Ocean Currents Using High-Fidelity Simulation and Machine Learning

Modeling ocean surface waves under complex ocean current conditions is of crucial importance to many naval applications. For example, traveling ships and underwater vehicles generate spatially heterogeneous currents behind them through their drag and propeller motions. The strong currents can influence the surface wave pattern in the ship wake. In this study, the nonlinear interactions between waves and complex wake currents are investigated using numerical simulations. An in-house code is developed for high-fidelity simulations of a nonlinear phase-resolved ocean wavefield interacting with subsurface currents. Several typical wake patterns are simulated using the present numerical method, and the influence of complex currents on the waves is analyzed quantitatively using theoretical solutions of wave-current interactions. We also present a method for solving the inverse problem of deducing the current field based on surface-wave data using machine-learning techniques. A deep neural network is designed for processing spatial-temporal surface wave data. Detailed analyses on the distributions of regression errors and the training dataset-dependency show that the proposed neural network can effectively deduce the current field.

physics.flu-dyn↗

arXiv2021

Evolutionary Clustering of Streaming Trajectories

The widespread deployment of smartphones and location-enabled, networked in-vehicle devices renders it increasingly feasible to collect streaming trajectory data of moving objects. The continuous clustering of such data can enable a variety of real-time services, such as identifying representative paths or common moving trends among objects in real-time. However, little attention has so far been given to the quality of clusters -- for example, it is beneficial to smooth short-term fluctuations in clusters to achieve robustness to exceptional data. We propose the notion of evolutionary clustering of streaming trajectories, abbreviated ECO, that enhances streaming-trajectory clustering quality by means of temporal smoothing that prevents abrupt changes in clusters across successive timestamps. Employing the notions of snapshot and historical trajectory costs, we formalize ECO and then formulate ECO as an optimization problem and prove that ECO can be performed approximately in linear time, thus eliminating the iterative processes employed in previous studies. Further, we propose a minimal-group structure and a seed point shifting strategy to facilitate temporal smoothing. Finally, we present all algorithms underlying ECO along with a set of optimization techniques. Extensive experiments with two real-life datasets offer insight into ECO and show that it outperforms state-of-the-art solutions in terms of both clustering quality and efficiency.

cs.DB↗

arXiv2021

Defect Extremal Surface for Reflected Entropy

Defect extremal surface is defined by extremizing the Ryu-Takayanagi formula corrected by the quantum defect theory. This is interesting when the AdS bulk contains a defect brane (or string). We introduce a defect extremal surface formula for reflected entropy, which is a mixed state generalization of entanglement entropy measure. Based on a decomposition procedure of an AdS bulk with a brane, we demonstrate the equivalence between defect extremal surface formula and island formula for reflected entropy in AdS$_3$/BCFT$_2$. We also compute the evolution of reflected entropy in evaporating black hole model and find that defect extremal surface formula agrees with island formula.

hep-th↗

arXiv2021

DeepQTMT: A Deep Learning Approach for Fast QTMT-based CU Partition of Intra-mode VVC

Versatile Video Coding (VVC), as the latest standard, significantly improves the coding efficiency over its ancestor standard High Efficiency Video Coding (HEVC), but at the expense of sharply increased complexity. In VVC, the quad-tree plus multi-type tree (QTMT) structure of coding unit (CU) partition accounts for over 97% of the encoding time, due to the brute-force search for recursive rate-distortion (RD) optimization. Instead of the brute-force QTMT search, this paper proposes a deep learning approach to predict the QTMT-based CU partition, for drastically accelerating the encoding process of intra-mode VVC. First, we establish a large-scale database containing sufficient CU partition patterns with diverse video content, which can facilitate the data-driven VVC complexity reduction. Next, we propose a multi-stage exit CNN (MSE-CNN) model with an early-exit mechanism to determine the CU partition, in accord with the flexible QTMT structure at multiple stages. Then, we design an adaptive loss function for training the MSE-CNN model, synthesizing both the uncertain number of split modes and the target on minimized RD cost. Finally, a multi-threshold decision scheme is developed, achieving desirable trade-off between complexity and RD performance. Experimental results demonstrate that our approach can reduce the encoding time of VVC by 44.65%-66.88% with the negligible Bjøntegaard delta bit-rate (BD-BR) of 1.322%-3.188%, which significantly outperforms other state-of-the-art approaches.

eess.IV↗

arXiv2021

Minimum Long-Loop Feedback Vertex Set and Network Dismantling

Network dismantling aims at breaking a network into disconnected components, and attacking vertices that intersect with many loops has proven to be a most efficient strategy. But the existing loop-focusing methods treat the short loops within densely connected local clusters (e.g., cliques) as equally important as the long loops connecting different clusters. Here we propose for highly clustered artificial and real-world networks a bipartite factor-graph formulation that retains all the long loops while simplifies the local dense clusters as individual factor nodes. We develop a mean-field theory for the associated long-loop feedback vertex set problem and apply its message-passing equation as a solver for network dismantling. The proposed factor-graph loop algorithm outperforms the current state-of-the-art graph loop algorithms by a considerable margin on various real networks. Further improvement in dismantling performance is achievable by optimizing the choice of the local dense clusters.

cs.SI↗

arXiv2021

Urban Epidemic Hazard Index for Chinese Cities: Why Did Small Cities Become Epidemic Hotspots?

Multiple small- to middle-scale cities, mostly located in northern China, became epidemic hotspots during the second wave of the spread of COVID-19 in early 2021. Despite qualitative discussions of potential social-economic causes, it remains unclear how this pattern could be accounted for from a quantitative approach. Through the development of an urban epidemic hazard index (EpiRank), we came up with a mathematical explanation for this phenomenon. The index is constructed from epidemic simulations on a multi-layer transportation network model on top of local SEIR transmission dynamics, which characterizes intra- and inter-city compartment population flow with a detailed mathematical description. Essentially, we argue that these highlighted cities possess greater epidemic hazards due to the combined effect of large regional population and small inter-city transportation. The proposed index, dynamic and applicable to different epidemic settings, could be a useful indicator for the risk assessment and response planning of urban epidemic hazards in China; the model framework is modularized and can be adapted for other nations without much difficulty.

physics.soc-ph↗

arXiv2020

Early Exit or Not: Resource-Efficient Blind Quality Enhancement for Compressed Images

Lossy image compression is pervasively conducted to save communication bandwidth, resulting in undesirable compression artifacts. Recently, extensive approaches have been proposed to reduce image compression artifacts at the decoder side; however, they require a series of architecture-identical models to process images with different quality, which are inefficient and resource-consuming. Besides, it is common in practice that compressed images are with unknown quality and it is intractable for existing approaches to select a suitable model for blind quality enhancement. In this paper, we propose a resource-efficient blind quality enhancement (RBQE) approach for compressed images. Specifically, our approach blindly and progressively enhances the quality of compressed images through a dynamic deep neural network (DNN), in which an early-exit strategy is embedded. Then, our approach can automatically decide to terminate or continue enhancement according to the assessed quality of enhanced images. Consequently, slight artifacts can be removed in a simpler and faster process, while the severe artifacts can be further removed in a more elaborate process. Extensive experiments demonstrate that our RBQE approach achieves state-of-the-art performance in terms of both blind quality enhancement and resource efficiency. The code is available at https://github.com/RyanXingQL/RBQE.

eess.IV↗

arXiv2020

Reflected Entropy for an Evaporating Black Hole

We study reflected entropy as a correlation measure in black hole evaporation. As a measure for bipartite mixed states, reflected entropy can be computed between black hole and radiation, radiation and radiation. We compute reflected entropy curves in three different models: 3-side wormhole model, End-of-the-World (EOW) brane model in three dimensions and two-dimensional eternal black hole plus CFT model. For 3-side wormhole model, we find that reflected entropy is dual to island cross sections. The reflected entropy between radiation and black hole increases at early time and then decreases to zero, similar to Page curve, but with a later transition time. The reflected entropy between radiation and radiation first increases and then saturates. For the EOW brane model, similar behaviors of reflected entropy are found. We propose a quantum extremal surface for reflected entropy, which we call quantum extremal cross section. In the eternal black hole plus CFT model, we find a generalized formula for reflected entropy with island cross section as its area term by considering the right half as the canonical purification of the left. Interestingly, the reflected entropy curve between the left black hole and the left radiation is nothing but the Page curve. We also find that reflected entropy between the left black hole and the right black hole decreases and goes to zero at late time. The reflected entropy between radiation and radiation increases at early time and saturates at late time.

hep-th↗

arXiv2020

Detecting Problem Statements in Peer Assessments

Effective peer assessment requires students to be attentive to the deficiencies in the work they rate. Thus, their reviews should identify problems. But what ways are there to check that they do? We attempt to automate the process of deciding whether a review comment detects a problem. We use over 18,000 review comments that were labeled by the reviewees as either detecting or not detecting a problem with the work. We deploy several traditional machine-learning models, as well as neural-network models using GloVe and BERT embeddings. We find that the best performer is the Hierarchical Attention Network classifier, followed by the Bidirectional Gated Recurrent Units (GRU) Attention and Capsule model with scores of 93.1% and 90.5% respectively. The best non-neural network model was the support vector machine with a score of 89.71%. This is followed by the Stochastic Gradient Descent model and the Logistic Regression model with 89.70% and 88.98%.

cs.IR↗

arXiv2020

Structural Control Analysis of System Dynamics Models

Structural control theory could be applied to study the control principles of social, economic and managerial systems. System Dynamics (SD) is the target field in social-economic sciences for endogenizing this theory, a subject that provides modeling solutions to real-world problems. SD models adopt diagrammatic representations, making it an ideal ground for transplanting structural control theory which utilizes similar graphic representations. This study sets up the theoretical ground for conducting structural control analysis (SCA) on SD models, summarized as a post-modeling workflow for SD practitioners, which serves as a specific application of the general structural control theory in social-economic sciences. Theoretical and practical establishments for SCA components are developed coordinately. Specifically, this study addresses the following questions: (1) How do SD models differ from physical control systems in graphic representations, and how do these differences affect the way of applying structural control theories to SD? (2) How could one identify control inputs in SD models, and how could different levels of system control in SD models be conceptualized? (3) What are the structural control properties for important SD components, and how could these properties and control principles help justify modeling heuristics in SD practice? (4) What are the procedures for conducting Structural Control Analysis (SCA) in SD models, and what are the implications of SCA results for model calibration and decision making? Overall, this study provides general insights for system control analysis of nonlinear dynamic simulation models, which may go beyond SD and extend to various disciplines in social-economic sciences.

math.OC↗

arXiv2020

Broadband Tunable Phase Shifter For Microwaves

We implement a broadly tunable phase shifter for microwaves based on superconducting quantum interference devices (SQUIDs) and study it both experimentally and theoretically. At different frequencies, a unit transmission coefficient, $|S_{21}|=1$, can be theoretically achieved along a curve where the phase shift is controllable by magnetic flux. The fabricated device consists of three equidistant SQUIDs interrupting a transmission line. We model each SQUID embedded at different positions along the transmission line with two parameters, capacitance and inductance, the values of which we extract from the experiments. In our experiments, the tunability of the phase shift varies from from $0.07\timesπ$ to $0.14\timesπ$ radians along the full-transmission curve with the input frequency ranging from 6.00 to 6.28~GHz. The reported measurements are in good agreement with simulations, which is promising for future design work of phase shifters for different applications.

physics.app-ph↗

arXiv2020

Simulating the Spread of Epidemics in China on the Multi-layer Transportation Network: Beyond the Coronavirus in Wuhan

Based on the SEIR model and the modeling of urban transportation networks, a general-purpose simulator for the spread of epidemics in Chinese cities is built. The Chinese public transportation system between over 340 prefectural-level cities is modeled as a multi-layer bi-partite network, with layers representing different means of transportation (airlines, railways, sail routes and buses), and nodes divided into two categories (central cities, peripheral cities). At each city, an open-system SEIR model tracks the local spread of the disease, with population in- and out-flow exchanging with the overlying transportation network. The model accounts for (1) different transmissivities of the epidemic on different transportation media, (2) the transit of inbound flow at cities, (3) cross-infection on public transportation vehicles due to path overlap, and the realistic considerations that (4) the infected population are not entering public transportation and (5) the recovered population are not subject to repeated infections. The model could be used to simulate the city-level spread in China (and potentially other countries) of an arbitrary epidemic, characterized by its basic reproduction number, incubation period, infection period and zoonotic force, originated from any Chinese prefectural-level city(s), during the period before effective government interventions are implemented. Flowmaps are input into the system to trigger inter-city dynamics, assuming different flow strength, determined from empirical observation, within/between the bi-partite divisions of nodes. The model is used to simulate the 2019 Coronavirus epidemic in Wuhan; it shows that the framework is robust and reliable, and simulated results match public city-level datasets to an extraordinary extent.

physics.soc-ph↗

arXiv2020

Effective routing design for remote entanglement generation on quantum networks

Quantum network is a promising platform for many ground-breaking applications that lie beyond the capability of its classical counterparts. Efficient entanglement generation on quantum networks with relatively limited resources such as quantum memories is essential to fully realize the network's capabilities, the solution to which calls for delicate network design and is currently at the primitive stage. In this study we propose an effective routing scheme to enable automatic responses for multiple requests of entanglement generation between source-terminal stations on a quantum lattice network with finite edge capacities. Multiple connection paths are exploited for each connection request while entanglement fidelity is ensured for each path by performing entanglement purification. The routing scheme is highly modularized with a flexible nature, embedding quantum operations within the algorithmic workflow, whose performance is evaluated from multiple perspectives. In particular, three algorithms are proposed and compared for the scheduling of capacity allocation on the edges of quantum network. Embodying the ideas of proportional share and progressive filling that have been well-studied in classical routing problems, we design a new scheduling algorithm, the propagatory update method, which in certain aspects overrides the two algorithms based on classical heuristics in scheduling performances. The general solution scheme paves the road for effective design of efficient routing and flow control protocols on applicational quantum networks.

quant-ph↗

arXiv2019

Phase transitions and optimal algorithms for semi-supervised classifications on graphs: from belief propagation to graph convolution network

We perform theoretical and algorithmic studies for the problem of clustering and semi-supervised classification on graphs with both pairwise relational information and single-point feature information, upon a joint stochastic block model for generating synthetic graphs with both edges and node features. Asymptotically exact analysis based on the Bayesian inference of the underlying model are conducted, using the cavity method in statistical physics. Theoretically, we identify a phase transition of the generative model, which puts fundamental limits on the ability of all possible algorithms in the clustering task of the underlying model. Algorithmically, we propose a belief propagation algorithm that is asymptotically optimal on the generative model, and can be further extended to a belief propagation graph convolution neural network (BPGCN) for semi-supervised classification on graphs. For the first time, well-controlled benchmark datasets with asymptotially exact properties and optimal solutions could be produced for the evaluation of graph convolution neural networks, and for the theoretical understanding of their strengths and weaknesses. In particular, on these synthetic benchmark networks we observe that existing graph convolution neural networks are subject to an sparsity issue and an ovefitting issue in practice, both of which are successfully overcome by our BPGCN. Moreover, when combined with classic neural network methods, BPGCN yields extraordinary classification performances on some real-world datasets that have never been achieved before.

cond-mat.stat-mech↗