arXiv subjects

Dong Zhang

Publications and source records attributed to Dong Zhang.

At least 181 records · Page 10Linked to original sources

arXiv2022

Hessian filter-assisted full diameter at half maximum (FDHM) segmentation and quantification method for optical-resolution photoacoustic microscopy

Optical-resolution photoacoustic microscopy has been validated as a high-resolution and high-sensitivity imaging modality for angiographic studies in the past decades. Quantitative vascular analysis reveals critical information of physiological changes, where vessel segmentation is the key step. In this work, we developed a Hessian filter-assisted, adaptive thresholding vessel segmentation algorithm. Its performance is validated by a digital phantom and in vivo images. Its capability of capturing subtle vessel changes is further tested in two longitudinal studies on vascular responses to blood pressure agents. The results are compared with the widely used Hessian filter method. In the antihypotensive case, the proposed method detected a twice larger vasoconstriction than the Hessian filter method. In the antihypertensive case, the proposed method detected a vasodilation of 18.8 %, while the Hessian filter method failed in change detection. The proposed algorithm could correct errors caused by conventional segmentation methods and improve quantitative accuracy for angiographic applications.

physics.med-ph↗

arXiv2022

FedMix: Mixed Supervised Federated Learning for Medical Image Segmentation

The purpose of federated learning is to enable multiple clients to jointly train a machine learning model without sharing data. However, the existing methods for training an image segmentation model have been based on an unrealistic assumption that the training set for each local client is annotated in a similar fashion and thus follows the same image supervision level. To relax this assumption, in this work, we propose a label-agnostic unified federated learning framework, named FedMix, for medical image segmentation based on mixed image labels. In FedMix, each client updates the federated model by integrating and effectively making use of all available labeled data ranging from strong pixel-level labels, weak bounding box labels, to weakest image-level class labels. Based on these local models, we further propose an adaptive weight assignment procedure across local clients, where each client learns an aggregation weight during the global model update. Compared to the existing methods, FedMix not only breaks through the constraint of a single level of image supervision, but also can dynamically adjust the aggregation weight of each local client, achieving rich yet discriminative feature representations. To evaluate its effectiveness, experiments have been carried out on two challenging medical image segmentation tasks, i.e., breast tumor segmentation and skin lesion segmentation. The results validate that our proposed FedMix outperforms the state-of-the-art method by a large margin.

cs.CV↗

arXiv2022

Learning Calibrated-Guidance for Object Detection in Aerial Images

Object detection is one of the most fundamental yet challenging research topics in the domain of computer vision. Recently, the study on this topic in aerial images has made tremendous progress. However, complex background and worse imaging quality are obvious problems in aerial object detection. Most state-of-the-art approaches tend to develop elaborate attention mechanisms for the space-time feature calibrations with arduous computational complexity, while surprisingly ignoring the importance of feature calibrations in channel-wise. In this work, we propose a simple yet effective Calibrated-Guidance (CG) scheme to enhance channel communications in a feature transformer fashion, which can adaptively determine the calibration weights for each channel based on the global feature affinity correlations. Specifically, for a given set of feature maps, CG first computes the feature similarity between each channel and the remaining channels as the intermediary calibration guidance. Then, re-representing each channel by aggregating all the channels weighted together via the guidance operation. Our CG is a general module that can be plugged into any deep neural networks, which is named as CG-Net. To demonstrate its effectiveness and efficiency, extensive experiments are carried out on both oriented object detection task and horizontal object detection task in aerial images. Experimental results on two challenging benchmarks (DOTA and HRSC2016) demonstrate that our CG-Net can achieve the new state-of-the-art performance in accuracy with a fair computational overhead. The source code has been open sourced at https://github.com/WeiZongqi/CG-Net

cs.CV↗

arXiv2022

FaceAtlasAR: Atlas of Facial Acupuncture Points in Augmented Reality

Acupuncture is a technique in which practitioners stimulate specific points on the body. Those points, called acupuncture points (or acupoints), anatomically define areas on the skin relative to specific landmarks on the body. However, mapping the acupoints to individuals could be challenging for inexperienced acupuncturists. In this project, we proposed a system to localize and visualize facial acupoints for individuals in an augmented reality (AR) context. This system combines a face alignment model and a hair segmentation model to provide dense reference points for acupoints localization in real-time (60FPS). The localization process takes the proportional bone (B-cun or skeletal) measurement method, which is commonly operated by specialists; however, in the real practice, operators sometimes find it inaccurate due to skill-related error. With this system, users, even without any skills, can locate the facial acupoints as a part of the self-training or self-treatment process.

cs.GR↗

arXiv2022

The Machine Learning for Combinatorial Optimization Competition (ML4CO): Results and Insights

Combinatorial optimization is a well-established area in operations research and computer science. Until recently, its methods have focused on solving problem instances in isolation, ignoring that they often stem from related data distributions in practice. However, recent years have seen a surge of interest in using machine learning as a new approach for solving combinatorial problems, either directly as solvers or by enhancing exact solvers. Based on this context, the ML4CO aims at improving state-of-the-art combinatorial optimization solvers by replacing key heuristic components. The competition featured three challenging tasks: finding the best feasible solution, producing the tightest optimality certificate, and giving an appropriate solver configuration. Three realistic datasets were considered: balanced item placement, workload apportionment, and maritime inventory routing. This last dataset was kept anonymous for the contestants.

cs.LG↗

arXiv2022

Discrete-to-Continuous Extensions: Lovász extension and Morse theory

This is the first of a series of papers that develop a systematic bridge between constructions in discrete mathematics and the corresponding continuous analogs. In this paper, we establish an equivalence between Forman's discrete Morse theory on a simplicial complex and the continuous Morse theory (in the sense of any known non-smooth Morse theory) on the associated order complex via the Lovász extension. Furthermore, we propose a new version of the Lusternik-Schnirelman category on abstract simplicial complexes to bridge the classical Lusternik-Schnirelman theorem and its discrete analog on finite complexes. More generally, we can suggest a discrete Morse theory on hypergraphs by employing piecewise-linear (PL) Morse theory and Lovász extension, hoping to provide new tools for exploring the structure of hypergraphs.

math.CO↗

arXiv2022

Efficient primal heuristics for mixed-integer linear programs

This paper is a short report about our work for the primal task in the Machine Learning for Combinatorial Optimization NeurIPS 2021 Competition. For each dataset of our interest in the competition, we propose customized primal heuristic methods to efficiently identify high-quality feasible solutions. The computational studies demonstrate the superiority of our proposed approaches over the competitors'.

math.OC↗

arXiv2021

Discrete-to-Continuous Extensions: piecewise multilinear extension, min-max theory and spectral theory

We introduce the homogeneous and piecewise multilinear extensions and the eigenvalue problem for locally Lipschitz function pairs, in order to develop a systematic framework for relating discrete and continuous min-max problems. This also enables us to investigate spectral properties for pairs of $p$-homogeneous functions and to propose a critical point theory for zero-homogeneous functions. The main contributions are: (1) We provide several min-max relations between an original discrete formulation and its piecewise multilinear extension. We introduce the concept of perfect domain pairs to view comonotonicity on vectors as an extension of inclusion chains on sets. The piecewise multi-linear extension is (slice-)rank preserving, which closely relates to Tao's lemma on diagonal tensors. More discrete-to-continuous equalities are obtained, including a general form involving log-concave polynomials. And by employing these fundamental correspondences, we get further results and applications on tensors, Turán's problem, signed (hyper-)graphs, etc. (2) We derive the mountain pass characterization, linking theorems, nodal domain inequalities, inertia bounds, duality theorems and distribution of eigenvalues for pairs of $p$-homogeneous functions. We establish a new property on the subderivative of a convex function which relates to the Gauss map of the graph of the convex function. Based on these fundamental results, we can analyze the structure of eigenspaces in depth. For example, we show a simple one-to-one correspondence between the nonzero eigenvalues of the vertex p-Laplacian and the edge $p^*$-Laplacian of a graph. We also apply the theory to Cheeger inequalities and $p$-Laplacians on oriented hypergraphs and simplicial complexes. Also, the first nonlinear analog of Huang's approach for hypergraphs is provided.

math.CO↗

arXiv2021

Cell-Level State of Charge Estimation for Battery Packs Under Minimal Sensing

This manuscript presents an algorithm for individual Lithium-ion (Li-ion) battery cell state of charge (SOC) estimation in a large-scale battery pack under minimal sensing, where only pack-level voltage and current are measured. For battery packs consisting of up to thousands of cells in electric vehicle or stationary energy storage applications, it is desirable to estimate individual cell SOCs without cell local measurements in order to reduce sensing costs. Mathematically, pure series connected cells yield dynamics given by ordinary differential equations under classical full voltage sensing. In contrast, parallel--series connected battery packs are evidently more challenging because the dynamics are governed by a nonlinear differential--algebraic equations (DAE) system. The majority of the conventional studies on SOC estimation for battery packs benefit from idealizing the pack as a lumped single cell which ultimately lose track of cell-level conditions and are blind to potential risks of cell-level over-charge and over-discharge. This work explicitly models a battery pack with high fidelity cell-by-cell resolution based on the interconnection of single cell models, and examines the observability of cell-level state with only pack-level measurements. A DAE-based state observer with linear output error injection is formulated, where the individual cell SOC and current can be reconstructed from minimal number of pack sensing. The mathematically guaranteed asymptotic convergence of differential and algebraic state estimates is established by considering local Lipschitz continuity property of system nonlinearities. Simulation results for Graphite/NMC cells illustrate convergence for cell SOCs, currents, and voltages.

eess.SY↗

arXiv2021

Congestion in networks and manifolds, and fair-division problems

Several large scale networks, such as the backbone of the Internet, have been observed to behave like convex Riemannian manifolds of negative curvature. In particular, this paradigm explains the observed existence, for networks of this type, of a "congestion core" through which a surprising large fraction of the traffic transits, while this phenomenon cannot be detected by purely local criteria. In this practical situation, it is important to estimate and predict the size and location of this congestion core. In this article we reverse the point of view and, motivated by the physical problem, we study congestion phenomena in the purely theoretical framework of convex Riemannian manifolds of negative curvature. In particular, we introduce a novel method of fair-division algorithm to estimate the size and impact of the congestion core in this context.

math.DG↗

arXiv2021

Coloring the normalized Laplacian for oriented hypergraphs

The independence number, coloring number and related parameters are investigated in the setting of oriented hypergraphs using the spectrum of the normalized Laplace operator. For the independence number, both an inertia--like bound and a ratio--like bound are shown. A Sandwich Theorem involving the clique number, the vector chromatic number and the coloring number is proved, as well as a lower bound for the vector chromatic number in terms of the smallest and the largest eigenvalue of the normalized Laplacian. In addition, spectral partition numbers are studied in relation to the coloring number.

math.CO↗

arXiv2021

$p$-Laplace Operators for Oriented Hypergraphs

The $p$-Laplacian for graphs, as well as the vertex Laplace operator and the hyperedge Laplace operator for the general setting of oriented hypergraphs, are generalized. In particular, both a vertex $p$-Laplacian and a hyperedge $p$-Laplacian are defined for oriented hypergraphs, for all $p\geq 1$. Several spectral properties of these operators are investigated.

math.CO↗

arXiv2021

Deep image prior for undersampling high-speed photoacoustic microscopy

Photoacoustic microscopy (PAM) is an emerging imaging method combining light and sound. However, limited by the laser's repetition rate, state-of-the-art high-speed PAM technology often sacrifices spatial sampling density (i.e., undersampling) for increased imaging speed over a large field-of-view. Deep learning (DL) methods have recently been used to improve sparsely sampled PAM images; however, these methods often require time-consuming pre-training and large training dataset with ground truth. Here, we propose the use of deep image prior (DIP) to improve the image quality of undersampled PAM images. Unlike other DL approaches, DIP requires neither pre-training nor fully-sampled ground truth, enabling its flexible and fast implementation on various imaging targets. Our results have demonstrated substantial improvement in PAM images with as few as 1.4$\%$ of the fully sampled pixels on high-speed PAM. Our approach outperforms interpolation, is competitive with pre-trained supervised DL method, and is readily translated to other high-speed, undersampling imaging modalities.

eess.IV↗

arXiv2021

Spectral Theory of Laplace Operators on Oriented Hypergraphs

Several new spectral properties of the normalized Laplacian defined for oriented hypergraphs are shown. The eigenvalue $1$ and the case of duplicate vertices are discussed; two Courant nodal domain theorems are established; new quantities that bound the eigenvalues are introduced. In particular, the Cheeger constant is generalized and it is shown that the classical Cheeger bounds can be generalized for some classes of hypergraphs; it is shown that a geometric quantity used to study zonotopes bounds the largest eigenvalue from below, and that the notion of coloring number can be generalized and used for proving a Hoffman-like bound. Finally, the spectrum of the unnormalized Laplacian for Cartesian products of hypergraphs is discussed.

math.CO↗

arXiv2021

State Estimation for a Zero-Dimensional Electrochemical Model of Lithium-Sulfur Batteries

Lithium-sulfur (Li-S) batteries have become one of the most attractive alternatives over conventional Li-ion batteries due to their high theoretical specific energy density (2500 Wh/kg for Li-S vs. $\sim$250 Wh/kg for Li-ion). Accurate state estimation in Li-S batteries is urgently needed for safe and efficient operation. To the best of the authors' knowledge, electrochemical model-based observers have not been reported for Li-S batteries, primarily due to the complex dynamics that make state observer design a challenging problem. In this work, we demonstrate a state estimation scheme based on a zero-dimensional electrochemical model for Li-S batteries. The nonlinear differential-algebraic equation (DAE) model is incorporated into an extend Kalman filter. This observer design estimates both differential and algebraic states that represent the dynamic behavior inside the cell, from voltage and current measurements only. The effectiveness of the proposed estimation algorithm is illustrated by numerical simulation results. Our study unlocks how an electrochemical model can be utilized for practical state estimation of Li-S batteries.

eess.SY↗

arXiv2021

Machine Learning based Malicious Payload Identification in Software-Defined Networking

Deep packet inspection (DPI) has been extensively investigated in software-defined networking (SDN) as complicated attacks may intractably inject malicious payloads in the packets. Existing proprietary pattern-based or port-based third-party DPI tools can suffer from limitations in efficiently processing a large volume of data traffic. In this paper, a novel OpenFlow-enabled deep packet inspection (OFDPI) approach is proposed based on the SDN paradigm to provide adaptive and efficient packet inspection. First, OFDPI prescribes an early detection at the flow-level granularity by checking the IP addresses of each new flow via OpenFlow protocols. Then, OFDPI allows for deep packet inspection at the packet-level granularity: (i) for unencrypted packets, OFDPI extracts the features of accessible payloads, including tri-gram frequency based on Term Frequency and Inverted Document Frequency (TF-IDF) and linguistic features. These features are concatenated into a sparse matrix representation and are then applied to train a binary classifier with logistic regression rather than matching with specific pattern combinations. In order to balance the detection accuracy and performance bottleneck of the SDN controller, OFDPI introduces an adaptive packet sampling window based on the linear prediction; and (ii) for encrypted packets, OFDPI extracts notable features of packets and then trains a binary classifier with a decision tree, instead of decrypting the encrypted traffic to weaken user privacy. A prototype of OFDPI is implemented on the Ryu SDN controller and the Mininet platform. The performance and the overhead of the proposed sulotion are assessed using the real-world datasets through experiments. The numerical results indicate that OFDPI can provide a significant improvement in detection accuracy with acceptable overheads.

cs.NI↗

arXiv2020

Four-terminal perovskite/silicon tandem solar cell with integrated Mie-resonant spectral splitter metagrating

A spectral splitting, light trapping dielectric metasurface is designed, fabricated and integrated into a four-terminal perovskite/silicon hybrid tandem solar cell to increase the absorption of light close to the bandgap of the perovskite top cell, and enhance transmission of the near-infrared spectral band towards the bottom cell. The metagrating is composed of a hexagonal array of unit cells of 150-nm-tall hydrogenated amorphous silicon trimer nanostructures with dielectric Mie resonances in the 600-800 nm perovskite near-gap region, made using substrate-conformal imprint lithography. By tailoring the metasurface resonant scattering modes and their interference with the direct reflection paths we minimize specular reflection and obtain high diffraction efficiency that leads to improved light trapping in the perovskite top cell. The measured short-circuit current increase in the perovskite top cell is 0.5 mA/cm2 corresponding to an estimated efficiency gain of 0.26% (absolute) for the metasurface-integrated 4T perovskite/silicon tandem cell. Simulations for a further optimized metasurface spectrum splitter geometry predict a short-circuit current gain in the perovskite top cell of 1.4 mA/cm2 and an efficiency gain for the 4T tandem cell of 0.4% (absolute). The metagrating approach for simultaneous spectral splitting, light trapping and reflectance reduction provides a flexible platform that can be applied to many tandem cell geometries.

physics.optics↗

arXiv2020

Exciton vortices in two-dimensional hybrid perovskite monolayers

We study theoretically the exciton Bose-Einstein condensation and exciton vortices in a two-dimensional (2D) perovskite (PEA)2PbI4 monolayer. Combining the first-principles calculations and the Keldysh model, the exciton binding energy of (PEA)2PbI4 in a monolayer can approach hundreds meV, which make it possible to observe the excitonic effect at room temperature. Due to the large exciton binding energy, and hence the high density of excitons, we find that the critical temperature of the exciton condensation could approach the liquid nitrogen regime. In presence of perpendicular electric fields, the dipole-dipole interaction between excitons is found to drive the condensed excitons into patterned vortices, as the evolution time of vortex patterns is comparable to the exciton lifetime.

cond-mat.mes-hall↗