arXiv subjects

Yang Wang

Publications and source records attributed to Yang Wang.

At least 613 records · Page 34Linked to original sources

arXiv2021

SmartTriage: A system for personalized patient data capture, documentation generation, and decision support

Symptom checkers have emerged as an important tool for collecting symptoms and diagnosing patients, minimizing the involvement of clinical personnel. We developed a machine-learning-backed system, SmartTriage, which goes beyond conventional symptom checking through a tight bi-directional integration with the electronic medical record (EMR). Conditioned on EMR-derived patient history, our system identifies the patient's chief complaint from a free-text entry and then asks a series of discrete questions to obtain relevant symptomatology. The patient-specific data are used to predict detailed ICD-10-CM codes as well as medication, laboratory, and imaging orders. Patient responses and clinical decision support (CDS) predictions are then inserted back into the EMR. To train the machine learning components of SmartTriage, we employed novel data sets of over 25 million primary care encounters and 1 million patient free-text reason-for-visit entries. These data sets were used to construct: (1) a long short-term memory (LSTM) based patient history representation, (2) a fine-tuned transformer model for chief complaint extraction, (3) a random forest model for question sequencing, and (4) a feed-forward network for CDS predictions. In total, our system supports 337 patient chief complaints, which together make up $>90\%$ of all primary care encounters at Kaiser Permanente.

cs.CL↗

arXiv2021

Visualizing Graph Neural Networks with CorGIE: Corresponding a Graph to Its Embedding

Graph neural networks (GNNs) are a class of powerful machine learning tools that model node relations for making predictions of nodes or links. GNN developers rely on quantitative metrics of the predictions to evaluate a GNN, but similar to many other neural networks, it is difficult for them to understand if the GNN truly learns characteristics of a graph as expected. We propose an approach to corresponding an input graph to its node embedding (aka latent space), a common component of GNNs that is later used for prediction. We abstract the data and tasks, and develop an interactive multi-view interface called CorGIE to instantiate the abstraction. As the key function in CorGIE, we propose the K-hop graph layout to show topological neighbors in hops and their clustering structure. To evaluate the functionality and usability of CorGIE, we present how to use CorGIE in two usage scenarios, and conduct a case study with five GNN experts.

cs.LG↗

arXiv2021

Inferring galaxy dark halo properties from visible matter with Machine Learning

Next-generation surveys will provide photometric and spectroscopic data of millions to billions of galaxies with unprecedented precision. This offers a unique chance to improve our understanding of the galaxy evolution and the unresolved nature of dark matter (DM). At galaxy scales, the density distribution of DM is strongly affected by the astrophysical feedback processes, which are difficult to fully account for in classical techniques to derive mass models. In this work, we explore the capability of supervised learning algorithms to predict the DM content of galaxies from luminous observational-like parameters, using the public catalog of the TNG100 simulation. In particular, we use Photometric, Structural and Kinematic parameters to predict the total DM mass, DM half-mass radius, DM mass inside one and two stellar half-mass radii. We adopt the coefficient of determination, $R^2$, as a reference metric to evaluate the accuracy of these predictions. We find that the Photometric features alone are able to predict the total DM mass with fair accuracy, while Structural and Photometric features together are more effective to determine the DM inside the stellar half mass radius, and the DM within twice the stellar half mass radius. However, using all observational quantities together (Photometry, Structural and Kinematics) incredibly improves the overall accuracy for all DM quantities. This first test shows that Machine Learning tools are promising approaches to derive predictions of the DM in real galaxies. The next steps will be to improve observational realism of the training sets, by closely select samples which accurately reproduce the typical observed luminous scaling relations. The trained pipelines will be suitable for real galaxy data collected from the next-generation surveys like Rubin/LSST, Euclid, CSST, 4MOST, DESI, to derive, e.g., the properties of their central DM fractions.

astro-ph.GA↗

arXiv2021

Unrestricted Adversarial Attacks on ImageNet Competition

Many works have investigated the adversarial attacks or defenses under the settings where a bounded and imperceptible perturbation can be added to the input. However in the real-world, the attacker does not need to comply with this restriction. In fact, more threats to the deep model come from unrestricted adversarial examples, that is, the attacker makes large and visible modifications on the image, which causes the model classifying mistakenly, but does not affect the normal observation in human perspective. Unrestricted adversarial attack is a popular and practical direction but has not been studied thoroughly. We organize this competition with the purpose of exploring more effective unrestricted adversarial attack algorithm, so as to accelerate the academical research on the model robustness under stronger unbounded attacks. The competition is held on the TianChi platform (\url{https://tianchi.aliyun.com/competition/entrance/531853/introduction}) as one of the series of AI Security Challengers Program.

cs.CV↗

arXiv2021

Non-Asymptotic Error Bounds for Bidirectional GANs

We derive nearly sharp bounds for the bidirectional GAN (BiGAN) estimation error under the Dudley distance between the latent joint distribution and the data joint distribution with appropriately specified architecture of the neural networks used in the model. To the best of our knowledge, this is the first theoretical guarantee for the bidirectional GAN learning approach. An appealing feature of our results is that they do not assume the reference and the data distributions to have the same dimensions or these distributions to have bounded support. These assumptions are commonly assumed in the existing convergence analysis of the unidirectional GANs but may not be satisfied in practice. Our results are also applicable to the Wasserstein bidirectional GAN if the target distribution is assumed to have a bounded support. To prove these results, we construct neural network functions that push forward an empirical distribution to another arbitrary empirical distribution on a possibly different-dimensional space. We also develop a novel decomposition of the integral probability metric for the error analysis of bidirectional GANs. These basic theoretical results are of independent interest and can be applied to other related learning problems.

cs.LG↗

arXiv2021

On the capacity of deep generative networks for approximating distributions

We study the efficacy and efficiency of deep generative networks for approximating probability distributions. We prove that neural networks can transform a low-dimensional source distribution to a distribution that is arbitrarily close to a high-dimensional target distribution, when the closeness are measured by Wasserstein distances and maximum mean discrepancy. Upper bounds of the approximation error are obtained in terms of the width and depth of neural network. Furthermore, it is shown that the approximation error in Wasserstein distance grows at most linearly on the ambient dimension and that the approximation order only depends on the intrinsic dimension of the target distribution. On the contrary, when $f$-divergences are used as metrics of distributions, the approximation property is different. We show that in order to approximate the target distribution in $f$-divergences, the dimension of the source distribution cannot be smaller than the intrinsic dimension of the target distribution.

cs.LG↗

arXiv2021

Importance of Mock Observations in Validating Galaxy Properties for Cosmological Simulations

The galaxy luminosity function and galaxy stellar mass function are fundamental statistics in the testing of galaxy formation models. Theoretical predictions based on cosmological simulations can deviate from observations, especially at the bright and faint ends. In this case, the mismatch may come from missing physics, oversimplified or inaccurate model recipes, or inappropriate methods of extracting basic astrophysical quantities from simulations. The latter is a crucial aspect to consider to avoid misleading conclusions when comparing simulations with observations. In this paper, we have applied a new method to produce `observed' galaxies identified in mock imaging of hydrodynamical simulations. We generate low-redshift mock galaxies from the TNG100-1 simulation of IllustrisTNG and analyse them using standard `observational' techniques to extract their main structural parameters. We show that our technique can produce realistic surface-brightness distributions of the simulated galaxies, including classical morphological substructures, such as spiral arms and bars. In particular, we find a very good agreement of the total luminosity and stellar mass versus halo mass relationships, and the galaxy stellar mass versus size relationship between mock observations and real galaxies. We also compare the luminosity function and the mass function of the mock galaxy sample with literature data and find a good agreement at all luminosity and mass scales. In particular, we find no significant tension at the bright end of the galaxy luminosity function, as reported in many analyses using simplified recipes to identify galaxy haloes, which in fact miscount the contribution of the extended galaxy haloes around large galaxies. This demonstrates the critical impact of using observational driven approaches to the simulation analyses to produce realistic predictions to compare to observations.

astro-ph.CO↗

arXiv2021

NOAHQA: Numerical Reasoning with Interpretable Graph Question Answering Dataset

While diverse question answering (QA) datasets have been proposed and contributed significantly to the development of deep learning models for QA tasks, the existing datasets fall short in two aspects. First, we lack QA datasets covering complex questions that involve answers as well as the reasoning processes to get the answers. As a result, the state-of-the-art QA research on numerical reasoning still focuses on simple calculations and does not provide the mathematical expressions or evidences justifying the answers. Second, the QA community has contributed much effort to improving the interpretability of QA models. However, these models fail to explicitly show the reasoning process, such as the evidence order for reasoning and the interactions between different pieces of evidence. To address the above shortcomings, we introduce NOAHQA, a conversational and bilingual QA dataset with questions requiring numerical reasoning with compound mathematical expressions. With NOAHQA, we develop an interpretable reasoning graph as well as the appropriate evaluation metric to measure the answer quality. We evaluate the state-of-the-art QA models trained using existing QA datasets on NOAHQA and show that the best among them can only achieve 55.5 exact match scores, while the human performance is 89.7. We also present a new QA model for generating a reasoning graph where the reasoning graph metric still has a large gap compared with that of humans, e.g., 28 scores.

cs.CL↗

arXiv2021

Knowledge Distillation for Human Action Anticipation

We consider the task of training a neural network to anticipate human actions in video. This task is challenging given the complexity of video data, the stochastic nature of the future, and the limited amount of annotated training data. In this paper, we propose a novel knowledge distillation framework that uses an action recognition network to supervise the training of an action anticipation network, guiding the latter to attend to the relevant information needed for correctly anticipating the future actions. This framework is possible thanks to a novel loss function to account for positional shifts of semantic concepts in a dynamic video. The knowledge distillation framework is a form of self-supervised learning, and it takes advantage of unlabeled data. Experimental results on JHMDB and EPIC-KITCHENS dataset show the effectiveness of our approach.

cs.CV↗

arXiv2021

Linear convergence of randomized Kaczmarz method for solving complex-valued phaseless equations

A randomized Kaczmarz method was recently proposed for phase retrieval, which has been shown numerically to exhibit empirical performance over other state-of-the-art phase retrieval algorithms both in terms of the sampling complexity and in terms of computation time. While the rate of convergence has been studied well in the real case where the signals and measurement vectors are all real-valued, there is no guarantee for the convergence in the complex case. In fact, the linear convergence of the randomized Kaczmarz method for phase retrieval in the complex setting is left as a conjecture by Tan and Vershynin. In this paper, we provide the first theoretical guarantees for it. We show that for random measurements $\mathbf{a}_j \in \mathbb{C}^n, j=1,\ldots,m $ which are drawn independently and uniformly from the complex unit sphere, or equivalent are independent complex Gaussian random vectors, when $m \ge Cn$ for some universal positive constant $C$, the randomized Kaczmarz scheme with a good initialization converges linearly to the target solution (up to a global phase) in expectation with high probability. This gives a positive answer to that conjecture.

math.NA↗

arXiv2021

On-Surface Decarboxylation Coupling Facilitated by Lock-to-Unlock Variation of Molecules upon the Reaction

On-surface synthesis (OSS) involving relatively high energy barriers remains challenging due to a typical dilemma: firm molecular anchor is required to prevent molecular desorption upon the reaction, whereas sufficient lateral mobility is crucial for subsequent coupling and assembly. By locking the molecular precursors on the substrate then unlocking them during the reaction, we present a strategy to address this challenge. High-yield synthesis based on well-defined decarboxylation, intermediate transition and hexamerization is demonstrated, resulting in an extended and ordered network exclusively composed of the newly-synthesized macrocyclic compound. Thanks to the steric hindrance of its maleimide group, we attain a preferential selection of the coupling. This work unlocks a promising path to enrich the reaction types and improve the coupling selectivity hence the structual homogeneity of the final product for OSS.

physics.chem-ph↗

arXiv2021

Preparing Dicke states in a spin ensemble using phase estimation

We present a Dicke state preparation scheme which uses global control of $N$ spin qubits: our scheme is based on the standard phase estimation algorithm, which estimates the eigenvalue of a unitary operator. The scheme prepares a Dicke state non-deterministically by collectively coupling the spins to an ancilla qubit via a $ZZ$-interaction, using $\ceil*{\log_2 N} + 1$ ancilla qubit measurements. The preparation of such Dicke states can be useful if the spins in the ensemble are used for magnetic sensing: we discuss a possible realization using an ensemble of electronic spins located at diamond Nitrogen-Vacancy (NV) centers coupled to a single superconducting flux qubit. We also analyze the effect of noise and limitations in our scheme.

quant-ph↗

arXiv2021

Dual-constrained Deep Semi-Supervised Coupled Factorization Network with Enriched Prior

Nonnegative matrix factorization is usually powerful for learning the "shallow" parts-based representation, but it clearly fails to discover deep hierarchical information within both the basis and representation spaces. In this paper, we technically propose a new enriched prior based Dual-constrained Deep Semi-Supervised Coupled Factorization Network, called DS2CF-Net, for learning the hierarchical coupled representations. To ex-tract hidden deep features, DS2CF-Net is modeled as a deep-structure and geometrical structure-constrained neural network. Specifically, DS2CF-Net designs a deep coupled factorization architecture using multi-layers of linear transformations, which coupled updates the bases and new representations in each layer. To improve the discriminating ability of learned deep representations and deep coefficients, our network clearly considers enriching the supervised prior by the joint deep coefficients-regularized label prediction, and incorporates enriched prior information as additional label and structure constraints. The label constraint can enable the samples of the same label to have the same coordinate in the new feature space, while the structure constraint forces the coefficient matrices in each layer to be block-diagonal so that the enhanced prior using the self-expressive label propagation are more accurate. Our network also integrates the adaptive dual-graph learning to retain the local manifold structures of both the data manifold and feature manifold by minimizing the reconstruction errors in each layer. Extensive experiments on several real databases demonstrate that our DS2CF-Net can obtain state-of-the-art performance for representation learning and clustering.

cs.LG↗

arXiv2021

OSKR/OKAI: Systematic Optimization of Key Encapsulation Mechanisms from Module Lattice

In this work, we make \emph{systematic} optimizations of key encapsulation mechanisms (KEM) based on module learning-with-errors (MLWE), covering algorithmic design, fundamental operation of number-theoretic transform (NTT), approaches to expanding encapsulated key size, and optimized implementation coding. We focus on Kyber (now in the Round-3 finalist of NIST PQC standardization) and Aigis (a variant of Kyber proposed at PKC 2020). By careful analysis, we first observe that the algorithmic design of Kyber and Aigis can be optimized by the mechanism of asymmetric key consensus with noise (AKCN) proposed in \cite{JZ16,JZ19}. Specifically, the decryption process can be simplified with AKCN, leading to a both faster and less error-prone decryption process. Moreover, the AKCN-based optimized version has perfect compatibility with the deployment of Kyber/Aigis in reality, as they can run on the same parameters, the same public key, and the same encryption process. We make a systematic study of the variants of NTT proposed in recent years for extending its applicability scope, make concrete analysis of their exact computational complexity, and in particular show their equivalence. We then present a new variant named hybrid-NTT (H-NTT), combining the advantages of existing NTT methods, and derive its optimality in computational complexity. The H-NTT technique not only has larger applicability scope but also allows for modular and unified implementation codes of NTT operations even with varying module dimensions. We analyze and compare the different approaches to expand the size of key to be encapsulated (specifically, 512-bit key for dimension of 1024), and conclude with the most economic approach. To mitigate the compatibility issue in implementations we adopt the proposed H-NTT method.

cs.CR↗

arXiv2021

Error Analysis of Deep Ritz Methods for Elliptic Equations

Using deep neural networks to solve PDEs has attracted a lot of attentions recently. However, why the deep learning method works is falling far behind its empirical success. In this paper, we provide a rigorous numerical analysis on deep Ritz method (DRM) \cite{Weinan2017The} for second order elliptic equations with Drichilet, Neumann and Robin boundary condition, respectively. We establish the first nonasymptotic convergence rate in $H^1$ norm for DRM using deep networks with smooth activation functions including logistic and hyperbolic tangent functions. Our results show how to set the hyper-parameter of depth and width to achieve the desired convergence rate in terms of number of training samples.

math.NA↗

arXiv2021

An investigation of high entropy alloy conductivity using first-principles calculations

The Kubo-Greenwood equation, in combination with the first-principles Korringa-Kohn-Rostoker Coherent Potential Approximation (KKR-CPA) can be used to calculate the DC residual resistivity of random alloys at T = 0 K. We implemented this method in a multiple scattering theory based ab initio package, MuST, and applied it to the ab initio study of the residual resistivity of the high entropy alloy Al$_x$CoCrFeNi as a function of $x$. The calculated resistivities are compared with experimental data. We also predict the residual resistivity of refractory high entropy alloy MoNbTaV$_x$W. The calculated resistivity trends are also explained using theoretical arguments.

cond-mat.mtrl-sci↗

arXiv2021

Variational quantum process tomography

Quantum process tomography is an experimental technique to fully characterize an unknown quantum process. Standard quantum process tomography suffers from exponentially scaling of the number of measurements with the increasing system size. In this work, we put forward a quantum machine learning algorithm which approximately encodes the unknown unitary quantum process into a relatively shallow depth parametric quantum circuit. We demonstrate our method by reconstructing the unitary quantum processes resulting from the quantum Hamiltonian evolution and random quantum circuits up to $8$ qubits. Results show that those quantum processes could be reconstructed with high fidelity, while the number of input states required are at least $2$ orders of magnitude less than required by the standard quantum process tomography.

quant-ph↗

arXiv2021

Three-Dimensional Trajectory Design for Multi-User MISO UAV Communications: A Deep Reinforcement Learning Approach

In this paper, we investigate a multi-user downlink multiple-input single-output (MISO) unmanned aerial vehicle (UAV) communication system, where a multi-antenna UAV is employed to serve multiple ground terminals. Unlike existing approaches focus only on a simplified two-dimensional scenario, this paper considers a three-dimensional (3D) urban environment, where the UAV's 3D trajectory is designed to minimize data transmission completion time subject to practical throughput and flight movement constraints. Specifically, we propose a deep reinforcement learning (DRL)-based trajectory design for completion time minimization (DRL-TDCTM), which is developed from a deep deterministic policy gradient algorithm. In particular, to represent the state information of UAV and environment, we set an additional information, i.e., the merged pheromone, as a reference of reward which facilitates the algorithm design. By interacting with the external environment in the corresponding Markov decision process, the proposed algorithm can continuously and adaptively learn how to adjust the UAV's movement strategy. Finally, simulation results show the superiority of the proposed DRL-TDCTM algorithm over the conventional baseline methods.

cs.IT↗