arXiv subjects

Xiang Huang

Publications and source records attributed to Xiang Huang.

At least 55 records · Page 3Linked to original sources

arXiv2023

Question Decomposition Tree for Answering Complex Questions over Knowledge Bases

Knowledge base question answering (KBQA) has attracted a lot of interest in recent years, especially for complex questions which require multiple facts to answer. Question decomposition is a promising way to answer complex questions. Existing decomposition methods split the question into sub-questions according to a single compositionality type, which is not sufficient for questions involving multiple compositionality types. In this paper, we propose Question Decomposition Tree (QDT) to represent the structure of complex questions. Inspired by recent advances in natural language generation (NLG), we present a two-staged method called Clue-Decipher to generate QDT. It can leverage the strong ability of NLG model and simultaneously preserve the original questions. To verify that QDT can enhance KBQA task, we design a decomposition-based KBQA system called QDTQA. Extensive experiments show that QDTQA outperforms previous state-of-the-art methods on ComplexWebQuestions dataset. Besides, our decomposition method improves an existing KBQA system by 12% and sets a new state-of-the-art on LC-QuAD 1.0.

cs.CL↗

arXiv2023

SE-GSL: A General and Effective Graph Structure Learning Framework through Structural Entropy Optimization

Graph Neural Networks (GNNs) are de facto solutions to structural data learning. However, it is susceptible to low-quality and unreliable structure, which has been a norm rather than an exception in real-world graphs. Existing graph structure learning (GSL) frameworks still lack robustness and interpretability. This paper proposes a general GSL framework, SE-GSL, through structural entropy and the graph hierarchy abstracted in the encoding tree. Particularly, we exploit the one-dimensional structural entropy to maximize embedded information content when auxiliary neighbourhood attributes are fused to enhance the original graph. A new scheme of constructing optimal encoding trees is proposed to minimize the uncertainty and noises in the graph whilst assuring proper community partition in hierarchical abstraction. We present a novel sample-based mechanism for restoring the graph structure via node structural entropy distribution. It increases the connectivity among nodes with larger uncertainty in lower-level communities. SE-GSL is compatible with various GNN models and enhances the robustness towards noisy and heterophily structures. Extensive experiments show significant improvements in the effectiveness and robustness of structure learning and node representation learning.

cs.LG↗

arXiv2023

Exploring high thermal conductivity polymers via interpretable machine learning with physical descriptors

The efficient and economical exploitation of polymers with high thermal conductivity is essential to solve the issue of heat dissipation in organic devices. Currently, the experimental preparation of functional thermal conductivity polymers remains a trial and error process due to the multi-degrees of freedom during the synthesis and characterization process. In this work, we have proposed a high-throughput screening framework for polymer chains with high thermal conductivity via interpretable machine learning and physical-feature engineering. The polymer thermal conductivity datasets for training were first collected by molecular dynamics simulation. Inspired by the drug-like small molecule representation and molecular force field, 320 polymer monomer descriptors were calculated and the 20 optimized descriptors with physical meaning were extracted by hierarchical down-selection. All the machine learning models achieve a prediction accuracy R2 greater than 0.80, which is superior to that of represented by traditional graph descriptors. Further, the cross-sectional area and dihedral stiffness descriptors were identified for positive/negative contribution to thermal conductivity, and 107 promising polymer structures with thermal conductivity greater than 20.00 W/mK were obtained. Mathematical formulas for predicting the polymer thermal conductivity were also constructed by using symbolic regression. The high thermal conductivity polymer structures are mostly π-conjugated, whose overlapping p-orbitals enable easily to maintain strong chain stiffness and large group velocities. The proposed data-driven framework should facilitate the theoretical and experimental design of polymers with desirable properties.

cond-mat.mtrl-sci↗

arXiv2022

Meta-Auto-Decoder for Solving Parametric Partial Differential Equations

Many important problems in science and engineering require solving the so-called parametric partial differential equations (PDEs), i.e., PDEs with different physical parameters, boundary conditions, shapes of computation domains, etc. Recently, building learning-based numerical solvers for parametric PDEs has become an emerging new field. One category of methods such as the Deep Galerkin Method (DGM) and Physics-Informed Neural Networks (PINNs) aim to approximate the solution of the PDEs. They are typically unsupervised and mesh-free, but require going through the time-consuming network training process from scratch for each set of parameters of the PDE. Another category of methods such as Fourier Neural Operator (FNO) and Deep Operator Network (DeepONet) try to approximate the solution mapping directly. Being fast with only one forward inference for each PDE parameter without retraining, they often require a large corpus of paired input-output observations drawn from numerical simulations, and most of them need a predefined mesh as well. In this paper, we propose Meta-Auto-Decoder (MAD), a mesh-free and unsupervised deep learning method that enables the pre-trained model to be quickly adapted to equation instances by implicitly encoding (possibly heterogenous) PDE parameters as latent vectors. The proposed method MAD can be interpreted by manifold learning in infinite-dimensional spaces, granting it a geometric insight. Extensive numerical experiments show that the MAD method exhibits faster convergence speed without losing accuracy than other deep learning-based methods. The project page with code is available: https://gitee.com/mindspore/mindscience/tree/master/MindElec/.

cs.LG↗

arXiv2022

Signature of topological band crossing in ferromagnetic Cr1/3NbSe2 epitaxial thin film

In intercalated transition metal dichalcogenides (I-TMDC), transition metal intercalation introduces magnetic phases which in some cases induce topological band crossing. However, evidence of the topological properties remains elusive in such materials. Here we employ angle-resolved photoemission spectroscopy to reveal the band structure of epitaxially grown ferromagnetic Cr1/3NbSe2. Experimental evidence of the Weyl crossing shows Cr1/3NbSe2 to be a topological ferromagnet. This work highlights I-TMDC as platform towards the interplay of magnetic and topological physics in low-dimensional systems.

cond-mat.mtrl-sci↗

arXiv2022

Enhancing thermoelectric properties of isotope graphene nanoribbons via machine learning guided manipulation of disordered antidots and interfaces

Structural manipulation at the nanoscale breaks the intrinsic correlations among different energy carrier transport properties, achieving high thermoelectric performance. However, the coupled multifunctional (phonon and electron) transport in the design of nanomaterials makes the optimization of thermoelectric properties challenging. Machine learning brings convenience to the design of nanostructures with large degree of freedom. Herein, we conducted comprehensive thermoelectric optimization of isotopic armchair graphene nanoribbons (AGNRs) with antidots and interfaces by combining Green's function approach with machine learning algorithms. The optimal AGNR with ZT of 0.894 by manipulating antidots was obtained at the interfaces of the aperiodic isotope superlattices, which is 5.69 times larger than that of the pristine structure. The proposed optimal structure via machine learning provides physical insights that the carbon-13 atoms tend to form a continuous interface barrier perpendicular to the carrier transport direction to suppress the propagation of phonons through isotope AGNRs. The antidot effect is more effective than isotope substitution in improving the thermoelectric properties of AGNRs. The proposed approach coupling energy carrier transport property analysis with machine learning algorithms offers highly efficient guidance on enhancing the thermoelectric properties of low-dimensional nanomaterials, as well as to explore and gain non-intuitive physical insights.

cond-mat.mes-hall↗

arXiv2022

Computing Real Numbers with Large-Population Protocols Having a Continuum of Equilibria

Bournez, Fraigniaud, and Koegler defined a number in [0,1] as computable by their Large-Population Protocol (LPP) model, if the proportion of agents in a set of marked states converges to said number over time as the population grows to infinity. The notion, however, restricts the ordinary differential equations (ODEs) associated with an LPP to have only finitely many equilibria. This restriction places an intrinsic limitation on the model. As a result, a number is computable by an LPP if and only if it is algebraic, namely, not a single transcendental number can be computed under this notion. In this paper, we lift the finitary requirement on equilibria. That is, we consider systems with a continuum of equilibria. We show that essentially all numbers in [0,1] that are computable by bounded general-purpose analog computers (GPACs) or chemical reaction networks (CRNs) can also be computed by LPPs under this new definition. This implies a rich series of numbers (e.g., the reciprocal of Euler's constant, $π/4$, Euler's $γ$, Catalan's constant, and Dottie number) are all computable by LPPs. Our proof is constructive: We develop an algorithm that transfers bounded GPACs/CRNs into LPPs. Our algorithm also fixes a gap in Bournez et al.'s construction of LPPs designed to compute any arbitrary algebraic number in [0,1].

cs.FL↗

arXiv2021

Solving Partial Differential Equations with Point Source Based on Physics-Informed Neural Networks

In recent years, deep learning technology has been used to solve partial differential equations (PDEs), among which the physics-informed neural networks (PINNs) emerges to be a promising method for solving both forward and inverse PDE problems. PDEs with a point source that is expressed as a Dirac delta function in the governing equations are mathematical models of many physical processes. However, they cannot be solved directly by conventional PINNs method due to the singularity brought by the Dirac delta function. We propose a universal solution to tackle this problem with three novel techniques. Firstly the Dirac delta function is modeled as a continuous probability density function to eliminate the singularity; secondly a lower bound constrained uncertainty weighting algorithm is proposed to balance the PINNs losses between point source area and other areas; and thirdly a multi-scale deep neural network with periodic activation function is used to improve the accuracy and convergence speed of the PINNs method. We evaluate the proposed method with three representative PDEs, and the experimental results show that our method outperforms existing deep learning-based methods with respect to the accuracy, the efficiency and the versatility.

cs.LG↗

arXiv2019

Coulomb focusing in retrapped ionization with near-circularly polarized laser field

The full three-dimensional photoelectron momentum distributions of argon are measured in intense near-circularly polarized laser fields. We observed that the transverse momentum distribution of ejected electrons by 410-nm near-circularly polarized field is unexpectedly narrowed with increasing laser intensity, which is contrary to the conventional rules predicted by adiabatic theory. By analyzing the momentum-resolved angular momentum distribution measured experimentally and the corresponding trajectories of ejected electrons semiclassically, the narrowing can be attributed to a temporary trapping and thereby focusing of a photoelectron by the atomic potential in a quasibound state. With the near-circularly polarized laser field, the strong Coulomb interaction with the rescattering electrons is avoided, thus the Coulomb focusing in the retrapped process is highlighted. We believe that these findings will facilitate understanding and steering electron dynamics in the Coulomb coupled system.

physics.atom-ph↗

arXiv2019

Asymptotic Divergences and Strong Dichotomy

The Schnorr-Stimm dichotomy theorem concerns finite-state gamblers that bet on infinite sequences of symbols taken from a finite alphabet $Σ$. In this paper we use the Kullback-Leibler divergence to formulate the $\textit{lower asymptotic divergence}$ $\text{div}(S||α)$ of a probability measure $α$ on $Σ$ from a sequence $S$ over $Σ$ and the $\textit{upper asymptotic divergence}$ $\text{Div}(S||α)$ of $α$ from $S$ in such a way that a sequence $S$ is $α$-normal (meaning that every string $w$ has asymptotic frequency $α(w)$ in $S$) if and only if $\text{Div}(S||α)=0$. We also use the Kullback-Leibler divergence to quantify the $\textit{total risk }$ $\text{Risk}_G(w)$ that a finite-state gambler $G$ takes when betting along a prefix $w$ of $S$. Our main theorem is a $\textit{strong dichotomy theorem}$ that uses the above notions to $\textit{quantify}$ the exponential rates of winning and losing on the two sides of the Schnorr-Stimm dichotomy theorem (with the latter routinely extended from normality to $α$-normality). Modulo asymptotic caveats in the paper, our strong dichotomy theorem says that the following two things hold for prefixes $w$ of $S$. (1) The infinitely-often exponential rate of winning is $2^{\text{Div}(S||α)|w|}$. (2) The exponential rate of loss is $2^{-\text{Risk}_G(w)}$. We also use (1) to show that $1-\text{Div}(S||α)/c$, where $c= \log(1/ \min_{a\inΣ}α(a))$, is an upper bound on the finite-state $α$-dimension of $S$ and prove the dual fact that $1-\text{div}(S||α)/c$ is an upper bound on the finite-state strong $α$-dimension of $S$.

cs.IT↗

arXiv2019

A Distributed Synchronous SGD Algorithm with Global Top-$k$ Sparsification for Low Bandwidth Networks

Distributed synchronous stochastic gradient descent (S-SGD) has been widely used in training large-scale deep neural networks (DNNs), but it typically requires very high communication bandwidth between computational workers (e.g., GPUs) to exchange gradients iteratively. Recently, Top-$k$ sparsification techniques have been proposed to reduce the volume of data to be exchanged among workers. Top-$k$ sparsification can zero-out a significant portion of gradients without impacting the model convergence. However, the sparse gradients should be transferred with their irregular indices, which makes the sparse gradients aggregation difficult. Current methods that use AllGather to accumulate the sparse gradients have a communication complexity of $O(kP)$, where $P$ is the number of workers, which is inefficient on low bandwidth networks with a large number of workers. We observe that not all top-$k$ gradients from $P$ workers are needed for the model update, and therefore we propose a novel global Top-$k$ (gTop-$k$) sparsification mechanism to address the problem. Specifically, we choose global top-$k$ largest absolute values of gradients from $P$ workers, instead of accumulating all local top-$k$ gradients to update the model in each iteration. The gradient aggregation method based on gTop-$k$ sparsification reduces the communication complexity from $O(kP)$ to $O(k\log P)$. Through extensive experiments on different DNNs, we verify that gTop-$k$ S-SGD has nearly consistent convergence performance with S-SGD, and it has only slight degradations on generalization performance. In terms of scaling efficiency, we evaluate gTop-$k$ on a cluster with 32 GPU machines which are interconnected with 1 Gbps Ethernet. The experimental results show that our method achieves $2.7-12\times$ higher scaling efficiency than S-SGD and $1.1-1.7\times$ improvement than the existing Top-$k$ S-SGD.

cs.DC↗

arXiv2018

Computational multifocal microscopy

Despite recent advances, high performance single-shot 3D microscopy remains an elusive task. By introducing designed diffractive optical elements (DOEs), one is capable of converting a microscope into a 3D "kaleidoscope", in which case the snapshot image consists of an array of tiles and each tile focuses on different depths. However, the acquired multifocal microscopic (MFM) image suffers from multiple sources of degradation, which prevents MFM from further applications. We propose a unifying computational framework which simplifies the imaging system and achieves 3D reconstruction via computation. Our optical configuration omits chromatic correction grating and redesigns the multifocal grating to enlarge the tracking area. Our proposed setup features only one single grating in addition to a regular microscope. The aberration correction, along with Poisson and background denoising, are incorporated in our deconvolution-based fully-automated algorithm, which requires no empirical parameter-tuning. In experiments, we achieve the spatial resolutions of $0.35$um (lateral) and $0.5$um (axial), which are comparable to the resolution that can be achieved with confocal deconvolution microscopy. We demonstrate a 3D video of moving bacteria recorded at $25$ frames per second using our proposed computational multifocal microscopy technique.

physics.optics↗

arXiv2018

U-SLADS: Unsupervised Learning Approach for Dynamic Dendrite Sampling

Novel data acquisition schemes have been an emerging need for scanning microscopy based imaging techniques to reduce the time in data acquisition and to minimize probing radiation in sample exposure. Varies sparse sampling schemes have been studied and are ideally suited for such applications where the images can be reconstructed from a sparse set of measurements. Dynamic sparse sampling methods, particularly supervised learning based iterative sampling algorithms, have shown promising results for sampling pixel locations on the edges or boundaries during imaging. However, dynamic sampling for imaging skeleton-like objects such as metal dendrites remains difficult. Here, we address a new unsupervised learning approach using Hierarchical Gaussian Mixture Mod- els (HGMM) to dynamically sample metal dendrites. This technique is very useful if the users are interested in fast imaging the primary and secondary arms of metal dendrites in solidification process in materials science.

eess.IV↗

arXiv2018

3D Snapshot Microscopy of Extended Objects

Volumetric biological imaging often involves compromising high temporal resolution at the expense of high spatial resolution when popular scanning methods are used to capture 3D information. We introduce an integrated experimental and image reconstruction method for capturing dynamic 3D fluorescent extended objects as a series of synchronously measured 3D snapshots taken at the frame rate of the imaging camera. We employ multifocal microscopy (MFM) to simultaneously image at 25 focal planes and process this depth-encoded image to recover the 3D structure of extended objects, such as bacteria, using a sparsity-based reconstruction approach. The combined experimental and computational method produces image quality similar to confocal microscopy in a fraction of the acquisition time. In addition, our computational image reconstruction approach allows a simplified MFM optical design by correcting aberrations using the measured response to point sources. This "compressive" MFM acquisition and reconstruction method, where an image volume with roughly 8 million voxels is recovered from a single 1-megapixel captured image, enables straightforward study of dynamic processes in 3D, and as a simultaneous snapshot advances the state of the art in dynamic 3D microscopy.

physics.ins-det↗

arXiv2018

Real-Time Computability of Real Numbers by Chemical Reaction Networks

We explore the class of real numbers that are computed in real time by deterministic chemical reaction networks that are integral in the sense that all their reaction rate constants are positive integers. We say that such a reaction network computes a real number $α$ in real time if it has a designated species $X$ such that, when all species concentrations are set to zero at time $t = 0$, the concentration $x(t)$ of $X$ is within $2^{-t}$ of $|α|$ at all times $t \ge 1$, and the concentrations of all other species are bounded. We show that every algebraic number and some transcendental numbers are real time computable by chemical reaction networks in this sense. We discuss possible implications of this for the 1965 Hartmanis-Stearns conjecture, which says that no irrational algebraic number is real time computable by a Turing machine.

cs.CC↗

arXiv2018

Snapshot 3D tracking of insulin granules in live cells

Rapid and accurate volumetric imaging remains a challenge, yet has the potential to enhance understanding of cell function. We developed and used a multifocal microscope (MFM) for 3D snapshot imaging to allow 3D tracking of insulin granules labeled with mCherry in MIN6 cells. MFM employs a special diffractive optical element (DOE) to simultaneously image multiple focal planes. This simultaneous acquisition of information determines the 3D location of single objects at a speed only limited by the frame rate of array detector . We validated the accuracy of MFM imaging and tracking with fluorescence beads; the 3D positions and trajectories of single fluorescence beads can be determined accurately over a wide range of spatial and temporal scales. The 3D positions and trajectories of single insulin granules in a 3.2 micro meter deep volume were determined with imaging processing that combines 3D decovolution, shift correction, and finally tracking using the Imaris software package. We find that the motion of the granules is super-diffusive, but less so in 3D than 2D for cells grown on coverslip surfaces, suggesting an anisotropy in the cytoskeleton (e.g. microtubules and action).

physics.optics↗

arXiv2016

Polynomial Space Randomness in Analysis

We study the interaction between polynomial space randomness and a fundamental result of analysis, the Lebesgue differentiation theorem. We generalize Ko's framework for polynomial space computability in $\mathbb{R}^n$ to define \textit{weakly pspace-random} points, a new variant of polynomial space randomness. We show that the Lebesgue differentiation theorem holds for every weakly pspace-random point.

cs.CC↗

arXiv2015

Massively Parallel Ray Tracing Algorithm Using GPU

Ray tracing is a technique for generating an image by tracing the path of light through pixels in an image plane and simulating the effects of high-quality global illumination at a heavy computational cost. Because of the high computation complexity, it can't reach the requirement of real-time rendering. The emergence of many-core architectures, makes it possible to reduce significantly the running time of ray tracing algorithm by employing the powerful ability of floating point computation. In this paper, a new GPU implementation and optimization of the ray tracing to accelerate the rendering process is presented.

cs.GR↗