arXiv subjects

Rui Gao

Publications and source records attributed to Rui Gao.

At least 91 records · Page 5Linked to original sources

arXiv2021

Autonomous Tracking and State Estimation with Generalised Group Lasso

We address the problem of autonomous tracking and state estimation for marine vessels, autonomous vehicles, and other dynamic signals under a (structured) sparsity assumption. The aim is to improve the tracking and estimation accuracy with respect to classical Bayesian filters and smoothers. We formulate the estimation problem as a dynamic generalised group Lasso problem and develop a class of smoothing-and-splitting methods to solve it. The Levenberg--Marquardt iterated extended Kalman smoother-based multi-block alternating direction method of multipliers (LM-IEKS-mADMM) algorithms are based on the alternating direction method of multipliers (ADMM) framework. This leads to minimisation subproblems with an inherent structure to which three new augmented recursive smoothers are applied. Our methods can deal with large-scale problems without pre-processing for dimensionality reduction. Moreover, the methods allow one to solve nonsmooth nonconvex optimisation problems. We then prove that under mild conditions, the proposed methods converge to a stationary point of the optimisation problem. By simulated and real-data experiments including multi-sensor range measurement problems, marine vessel tracking, autonomous vehicle tracking, and audio signal restoration, we show the practical effectiveness of the proposed methods.

stat.ME↗

arXiv2021

Robust Hypothesis Testing with Wasserstein Uncertainty Sets

We consider a data-driven robust hypothesis test where the optimal test will minimize the worst-case performance regarding distributions that are close to the empirical distributions with respect to the Wasserstein distance. This leads to a new non-parametric hypothesis testing framework based on distributionally robust optimization, which is more robust when there are limited samples for one or both hypotheses. Such a scenario often arises from applications such as health care, online change-point detection, and anomaly detection. We study the computational and statistical properties of the proposed test by presenting a tractable convex reformulation of the original infinite-dimensional variational problem exploiting Wasserstein's properties and characterizing the radii selection for the uncertainty sets. We also demonstrate the good performance of our method on synthetic and real data.

math.ST↗

arXiv2021

Hierarchical Non-Stationary Temporal Gaussian Processes With $L^1$-Regularization

This paper is concerned with regularized extensions of hierarchical non-stationary temporal Gaussian processes (NSGPs) in which the parameters (e.g., length-scale) are modeled as GPs. In particular, we consider two commonly used NSGP constructions which are based on explicitly constructed non-stationary covariance functions and stochastic differential equations, respectively. We extend these NSGPs by including $L^1$-regularization on the processes in order to induce sparseness. To solve the resulting regularized NSGP (R-NSGP) regression problem we develop a method based on the alternating direction method of multipliers (ADMM) and we also analyze its convergence properties theoretically. We also evaluate the performance of the proposed methods in simulated and real-world datasets.

stat.ME↗

arXiv2021

Generalize Ultrasound Image Segmentation via Instant and Plug & Play Style Transfer

Deep segmentation models that generalize to images with unknown appearance are important for real-world medical image analysis. Retraining models leads to high latency and complex pipelines, which are impractical in clinical settings. The situation becomes more severe for ultrasound image analysis because of their large appearance shifts. In this paper, we propose a novel method for robust segmentation under unknown appearance shifts. Our contribution is three-fold. First, we advance a one-stage plug-and-play solution by embedding hierarchical style transfer units into a segmentation architecture. Our solution can remove appearance shifts and perform segmentation simultaneously. Second, we adopt Dynamic Instance Normalization to conduct precise and dynamic style transfer in a learnable manner, rather than previously fixed style normalization. Third, our solution is fast and lightweight for routine clinical adoption. Given 400*400 image input, our solution only needs an additional 0.2ms and 1.92M FLOPs to handle appearance shifts compared to the baseline pipeline. Extensive experiments are conducted on a large dataset from three vendors demonstrate our proposed method enhances the robustness of deep segmentation models.

eess.IV↗

arXiv2020

Wasserstein Distributionally Robust Optimization and Variation Regularization

Wasserstein distributionally robust optimization (DRO) has recently achieved empirical success for various applications in operations research and machine learning, owing partly to its regularization effect. Although connection between Wasserstein DRO and regularization has been established in several settings, existing results often require restrictive assumptions, such as smoothness or convexity, that are not satisfied for many problems. In this paper, we develop a general theory on the variation regularization effect of the Wasserstein DRO - a new form of regularization that generalizes total-variation regularization, Lipschitz regularization and gradient regularization. Our results cover possibly non-convex and non-smooth losses and losses on non-Euclidean spaces. Examples include multi-item newsvendor, portfolio selection, linear prediction, neural networks, manifold learning, and intensity estimation for Poisson processes, etc. As an application of our theory of variation regularization, we derive new generalization guarantees for adversarial robust learning.

cs.LG↗

arXiv2020

Contrastive Rendering for Ultrasound Image Segmentation

Ultrasound (US) image segmentation embraced its significant improvement in deep learning era. However, the lack of sharp boundaries in US images still remains an inherent challenge for segmentation. Previous methods often resort to global context, multi-scale cues or auxiliary guidance to estimate the boundaries. It is hard for these methods to approach pixel-level learning for fine-grained boundary generating. In this paper, we propose a novel and effective framework to improve boundary estimation in US images. Our work has three highlights. First, we propose to formulate the boundary estimation as a rendering task, which can recognize ambiguous points (pixels/voxels) and calibrate the boundary prediction via enriched feature representation learning. Second, we introduce point-wise contrastive learning to enhance the similarity of points from the same class and contrastively decrease the similarity of points from different classes. Boundary ambiguities are therefore further addressed. Third, both rendering and contrastive learning tasks contribute to consistent improvement while reducing network parameters. As a proof-of-concept, we performed validation experiments on a challenging dataset of 86 ovarian US volumes. Results show that our proposed method outperforms state-of-the-art methods and has the potential to be used in clinical practice.

eess.IV↗

arXiv2020

On fair entropy of the tent family

The notions of fair measure and fair entropy were introduced by Misiurewicz and Rodrigues recently, and discussed in detail for piecewise monotone interval maps. In particular, they showed that the fair entropy $h(a)$ of the tent map $f_a$, as a function of the parameter $a=\exp(h_{top}(f_a))$, is continuous and strictly increasing on $[\sqrt{2},2]$. In this short note, we extend the last result and characterize regularity of the function $h$ precisely. We prove that $h$ is $\frac{1}{2}$-Hölder continuous on $[\sqrt{2},2]$ and identify its best Hölder exponent on each subinterval of $[\sqrt{2},2]$. On the other hand, parallel to a recent result on topological entropy of the quadratic family due to Dobbs and Mihalache, we give a formula of pointwise Hölder exponents of $h$ at parameters chosen in an explicitly constructed set of full measure. This formula particularly implies that the derivative of $h$ vanishes almost everywhere.

math.DS↗

arXiv2020

Variable Splitting Methods for Constrained State Estimation in Partially Observed Markov Processes

In this paper, we propose a class of efficient, accurate, and general methods for solving state-estimation problems with equality and inequality constraints. The methods are based on recent developments in variable splitting and partially observed Markov processes. We first present the generalized framework based on variable splitting, then develop efficient methods to solve the state-estimation subproblems arising in the framework. The solutions to these subproblems can be made efficient by leveraging the Markovian structure of the model as is classically done in so-called Bayesian filtering and smoothing methods. The numerical experiments demonstrate that our methods outperform conventional optimization methods in computation cost as well as the estimation performance.

math.OC↗

arXiv2020

UO2/BeO interfacial thermal resistance and its effect on fuel thermal conductivity

UO2/BeO interfacial thermal resistance (ITR) is calculated by diffuse mismatch model (DMM) and the effects of ITR on UO2-BeO thermal conductivity are investigated. ITR predicted by DMM is on the order of 10-9 m2K/W. Using this ITR, UO2-BeO thermal conductivities are calculated by theoretical models and compared with experimental data. The results indicate that DMM prediction is applicable to the interface between UO2 and dispersed BeO, while not applicable to the interface between UO2 and continuous BeO. If the thermal conductivity of UO2 containing continuous BeO was to be in agreement with experimental data, its ITR should be on the order of 10-6 - 10-5 m2K/W. Therefore, the vibrational mismatch between UO2 and BeO considered by DMM is the major mechanism for attenuating the heat flux through UO2/dispersed-BeO interface, but not for UO2/continuous-BeO interface. Furthermore, it is found that the presence of ITR leads to the dependence of the thermal conductivity of UO2 containing dispersed BeO on BeO size. With the decrease in BeO size, UO2-BeO thermal conductivity decreases. When BeO size is smaller than a critical value, UO2-BeO thermal conductivity becomes even smaller than UO2 thermal conductivity. For UO2 containing continuous BeO, the thermal conductivity decreases with the decrease in the size of UO2 granule surrounded by BeO, but not necessarily smaller than UO2 thermal conductivity. Under a critical temperature, UO2-BeO thermal conductivity is always larger than UO2 thermal conductivity. Above the critical temperature, UO2-BeO thermal conductivity is larger than UO2 thermal conductivity only when UO2 granule size is large enough. The conditions for achieving the targeted enhancement of UO2 thermal conductivity by doping with BeO are derived. These conditions can be used to design and optimize the distribution, content, size of BeO, and the size of UO2 granule.

physics.app-ph↗

arXiv2020

Remove Appearance Shift for Ultrasound Image Segmentation via Fast and Universal Style Transfer

Deep Neural Networks (DNNs) suffer from the performance degradation when image appearance shift occurs, especially in ultrasound (US) image segmentation. In this paper, we propose a novel and intuitive framework to remove the appearance shift, and hence improve the generalization ability of DNNs. Our work has three highlights. First, we follow the spirit of universal style transfer to remove appearance shifts, which was not explored before for US images. Without sacrificing image structure details, it enables the arbitrary style-content transfer. Second, accelerated with Adaptive Instance Normalization block, our framework achieved real-time speed required in the clinical US scanning. Third, an efficient and effective style image selection strategy is proposed to ensure the target-style US image and testing content US image properly match each other. Experiments on two large US datasets demonstrate that our methods are superior to state-of-the-art methods on making DNNs robust against various appearance shifts.

eess.IV↗

arXiv2020

Thermal behaviors of light scalar resonances at low temperatures

We study the thermal properties of the lowest multiplet of the QCD light-flavor scalar resonances, including the $f_0(500)/σ$, $K_{0}^{*}(700)/κ$, $f_0(980)$ and $a_0(980)$, in the framework of unitarized $U(3)$ chiral perturbation theory. After the successful fits to the meson-meson scattering inputs, such as the phase shifts and inelasticities, we obtain the unknown parameters and further calculate the resonance poles and their residues at zero temperature. By including the finite-temperature effects in the unitarized meson-meson scattering amplitudes, the thermal behaviors of the scalar resonance poles in the complex energy plane are studied. The masses of $σ$ and $κ$ are found to considerably decrease when increasing the temperatures, while their widths turn out to be still large when the temperatures reach around $200$ MeV. In contrast, both the masses and widths of the $f_0(980)$ and $a_0(980)$ are only slightly changed.

hep-ph↗

arXiv2019

A hybrid stochastic differential reinsurance and investment game with bounded memory

This paper investigates a hybrid stochastic differential reinsurance and investment game between one reinsurer and two insurers, including a stochastic Stackelberg differential subgame and a non-zero-sum stochastic differential subgame. The reinsurer, as the leader of the Stackelberg game, can price reinsurance premium and invest its wealth in a financial market that contains a risk-free asset and a risky asset. The two insurers, as the followers of the Stackelberg game, can purchase proportional reinsurance from the reinsurer and invest in the same financial market. The competitive relationship between two insurers is modeled by the non-zero-sum game, and their decision making will consider the relative performance measured by the difference in their terminal wealth. We consider wealth processes with delay to characterize the bounded memory feature. This paper aims to find the equilibrium strategy for the reinsurer and insurers by maximizing the expected utility of the reinsurer's terminal wealth with delay and maximizing the expected utility of the combination of insurers' terminal wealth and the relative performance with delay. By using the idea of backward induction and the dynamic programming approach, we derive the equilibrium strategy and value functions explicitly. Then, we provide the corresponding verification theorem. Finally, some numerical examples and sensitivity analysis are presented to demonstrate the effects of model parameters on the equilibrium strategy. We find the delay factor discourages or stimulates investment depending on the length of delay. Moreover, competitive factors between two insurers make their optimal reinsurance-investment strategy interact, and reduce reinsurance demand and reinsurance premium price.

q-fin.MF↗

arXiv2019

The Numerical Simulation of Quanto Option Prices Using Bayesian Statistical Methods

In the paper, the pricing of Quanto options is studied, where the underlying foreign asset and the exchange rate are correlated with each other. Firstly, we adopt Bayesian methods to estimate unknown parameters entering the pricing formula of Quanto options, including the volatility of stock, the volatility of exchange rate and the correlation. Secondly, we compute and predict prices of different four types of Quanto options based on Bayesian posterior prediction techniques and Monte Carlo methods. Finally, we provide numerical simulations to demonstrate the advantage of Bayesian method used in this paper comparing with some other existing methods. This paper is a new application of the Bayesian methods in the pricing of multi-asset options.

q-fin.CP↗

arXiv2019

Iterated Extended Kalman Smoother-based Variable Splitting for $L_1$-Regularized State Estimation

In this paper, we propose a new framework for solving state estimation problems with an additional sparsity-promoting $L_1$-regularizer term. We first formulate such problems as minimization of the sum of linear or nonlinear quadratic error terms and an extra regularizer, and then present novel algorithms which solve the linear and nonlinear cases. The methods are based on a combination of the iterated extended Kalman smoother and variable splitting techniques such as alternating direction method of multipliers (ADMM). We present a general algorithmic framework for variable splitting methods, where the iterative steps involving minimization of the nonlinear quadratic terms can be computed efficiently by iterated smoothing. Due to the use of state estimation algorithms, the proposed framework has a low per-iteration time complexity, which makes it suitable for solving a large-scale or high-dimensional state estimation problem. We also provide convergence results for the proposed algorithms. The experiments show the promising performance and speed-ups provided by the methods.

cs.IT↗

arXiv2019

Effective-range-expansion study of near threshold heavy-flavor resonances

In this work we study the resonances near the thresholds of the open heavy-flavor hadrons using the effective-range-expansion method. The unitarity, analyticity and compositeness coefficient are also taken into account in our theoretical formalism. We consider the $Z_c(3900)$, $X(4020)$, $χ_{c1}(4140)$, $ψ(4260)$ and $ψ(4660)$. The scattering lengths and effective ranges from the relevant elastic $S$-wave scattering amplitudes are determined. Tentative discussions on the inner structures of the aforementioned resonances are given.

hep-ph↗

arXiv2019

Beam tests of a large-scale TORCH time-of-flight demonstrator

The TORCH time-of-flight detector is designed to provide particle identification in the momentum range 2-10 GeV/c over large areas. The detector exploits prompt Cherenkov light produced by charged particles traversing a 10 mm thick quartz plate. The photons propagate via total internal reflection and are focused onto a detector plane comprising position-sensitive Micro-Channel Plate Photo-Multiplier Tubes (MCP-PMT) detectors. The goal is to achieve a single-photon timing resolution of 70 ps, giving a timing precision of 15 ps per charged particle by combining the information from around 30 detected photons. The MCP-PMT detectors have been developed with a commercial partner (Photek Ltd, UK), leading to the delivery of a square tube of active area 53 $\times$ 53mm$^2$ with a granularity of 8 $\times$ 128 pixels equivalent. A large-scale demonstrator of TORCH, having a quartz plate of dimensions 660 $\times$ 1250 $\times$ 10 mm$^3$ and read out by a pair of MCP-PMTs with custom readout electronics, has been verified in a test beam campaign at the CERN PS. Preliminary results indicate that the required performance is close to being achieved. The anticipated performance of a full-scale TORCH detector at the LHCb experiment is presented.

physics.ins-det↗

arXiv2018

Status of the TORCH time-of-flight project

TORCH is a time-of-flight detector, designed to provide charged pi/K particle identification up to a momentum of 10 GeV/c for a 10 m flight path. To achieve this level of performance, a time resolution of 15 ps per incident particle is required. TORCH uses a plane of quartz of 1 cm thickness as a source of Cherenkov photons, which are then focussed onto square Micro-Channel Plate Photomultipliers (MCP-PMTs) of active area 53 x 53 mm^2, segmented into 8 x 128 pixels equivalent. A small-scale TORCH demonstrator with a customised MCP-PMT and associated readout electronics has been successfully operated in a 5 GeV/c mixed pion/proton beam at the CERN PS facility. Preliminary results indicate that a single-photon resolution better than 100 ps can be achieved. The expected performance of a full-scale TORCH detector for the Upgrade II of the LHCb experiment is also discussed.

physics.ins-det↗

arXiv2018

TORCH: a large area time-of-flight detector for particle identification

TORCH is a time-of-flight detector that is being developed for the Upgrade II of the LHCb experiment, with the aim of providing charged particle identification over the momentum range 2-10 GeV/c. A small-scale TORCH demonstrator with customised readout electronics has been operated successfully in beam tests at the CERN PS. Preliminary results indicate that a single-photon resolution better than 100 ps can be achieved.

physics.ins-det↗