Searcharxiv⌕ Search

arXiv subjects

Ting Yang

Publications and source records attributed to Ting Yang.

49 records · Page 3Linked to original sources

Law of large numbers for supercritical superprocesses with non-local branching

In this paper we establish a weak and a strong law of large numbers for supercritical superprocesses with general non-local branching mechanisms. Our results complement earlier results obtained for superprocesses with only local branching. Several interesting examples are developed, including multitype continuous-state branching processes, multitype superdiffusions and superprocesses with discontinuous spatial motions and non-decomposable branching mechanisms.

math.PR↗

Robust X-ray Sparse-view Phase Tomography via Hierarchical Synthesis Convolutional Neural Networks

Convolutional Neural Networks (CNN) based image reconstruction methods have been intensely used for X-ray computed tomography (CT) reconstruction applications. Despite great success, good performance of this data-based approach critically relies on a representative big training data set and a dense convoluted deep network. The indiscriminating convolution connections over all dense layers could be prone to over-fitting, where sampling biases are wrongly integrated as features for the reconstruction. In this paper, we report a robust hierarchical synthesis reconstruction approach, where training data is pre-processed to separate the information on the domains where sampling biases are suspected. These split bands are then trained separately and combined successively through a hierarchical synthesis network. We apply the hierarchical synthesis reconstruction for two important and classical tomography reconstruction scenarios: the spares-view reconstruction and the phase reconstruction. Our simulated and experimental results show that comparable or improved performances are achieved with a dramatic reduction of network complexity and computational cost. This method can be generalized to a wide range of applications including material characterization, in-vivo monitoring and dynamic 4D imaging.

cs.CV↗

Slice regular functions of several octonionic variables

Octonionic analysis is becoming eminent due to the role of octonions in the theory of G2 manifold. In this article, a new slice theory is introduced as a generalization of the holomorphic theory of several complex variables to the noncommutative or nonassociative realm. The Bochner-Martinelli formula is established for slice functions of several octonionic variables as well as several quaternionic variables. In this setting, we find the Hartogs phenomena for slice regular functions

math.CV↗

Growth theorems in slice analysis of several variables

In this paper, we define a class of slice mappings of several Clifford variables, and the corresponding slice regular mappings. Furthermore, we establish the growth theorem for slice regular starlike or convex mappings on the unit ball of several slice Clifford variables, as well as on the bounded slice domain which is slice starlike and slice circular.

math.CV↗

Local Uncertainty Sampling for Large-Scale Multi-Class Logistic Regression

A major challenge for building statistical models in the big data era is that the available data volume far exceeds the computational capability. A common approach for solving this problem is to employ a subsampled dataset that can be handled by available computational resources. In this paper, we propose a general subsampling scheme for large-scale multi-class logistic regression and examine the variance of the resulting estimator. We show that asymptotically, the proposed method always achieves a smaller variance than that of the uniform random sampling. Moreover, when the classes are conditionally imbalanced, significant improvement over uniform sampling can be achieved. Empirical performance of the proposed method is compared to other methods on both simulated and real-world datasets, and these results match and confirm our theoretical analysis.

stat.CO↗

Bayesian Model Averaging with Exponentiated Least Square Loss

The model averaging problem is to average multiple models to achieve a prediction accuracy not much worse than that of the best single model in terms of mean squared error. It is known that if the models are misspecified, model averaging is superior to model selection. Specifically, let $n$ be the sample size, then the worst case regret of the former decays at a rate of $O(1/n)$ while the worst case regret of the latter decays at a rate of $O(1/\sqrt{n})$. The recently proposed $Q$-aggregation algorithm \citep{DaiRigZhang12} solves the model averaging problem with the optimal regret of $O(1/n)$ both in expectation and in deviation; however it suffers from two limitations: (1) for continuous dictionary, the proposed greedy algorithm for solving $Q$-aggregation is not applicable; (2) the formulation of $Q$-aggregation appears ad hoc without clear intuition. This paper examines a different approach to model averaging by considering a Bayes estimator for deviation optimal model averaging by using exponentiated least squares loss. We establish a primal-dual relationship of this estimator and that of $Q$-aggregation and propose new greedy procedures that satisfactorily resolve the above mentioned limitations of $Q$-aggregation.

math.ST↗

Skeleton decomposition and law of large numbers for supercritical superprocesses

The goal of this paper has two-folds. First, we establish skeleton and spine decompositions for superprocesses whose underlying processes are general symmetric Hunt processes. Second, we use these decompositions to obtain weak and strong law of large numbers for supercritical superprocesses where the spatial motion is a symmetric Hunt process on a locally compact metric space $E$ and the branching mechanism takes the form $$ψ_β(x,λ)=-β(x)λ+α(x)λ^{2}+\int_{(0,{\infty})}(e^{-λy}-1+λy)π(x,dy)$$ with $β\in\mathcal{B}_{b}(E)$, $α\in \mathcal{B}^{+}_{b}(E)$ and $π$ being a kernel from $E$ to $(0,{\infty})$ satisfying $\sup_{x\in E}\int_{(0,{\infty})} (y\wedge y^{2}) π(x,dy)<{\infty}$. The limit theorems are established under the assumption that an associated Schrödinger operator has a spectral gap. Our results cover many interesting examples of superprocesses, including super Ornstein-Uhlenbeck process and super stable-like process. The strong law of large numbers for supercritical superprocesses are obtained under the assumption that the strong law of large numbers for an associated supercritical branching Markov process holds along a discrete sequence of times, extending an earlier result of Eckhoff, Kyprianou and Winkel [Ann. Probab.,43(5), 2594--2659, 2015] for superdiffusions to a large class of superprocesses. The key for such a result is due to the skeleton decomposition of superprocess, which represents a superprocess as an immigration process along a supercritical branching Markov process.

math.PR↗

Boundary Harnack principle and gradient estimates for fractional Laplacian perturbed by non-local operators

Suppose $d\ge 2$ and $0<β<α<2$. We consider the non-local operator $\mathcal{L}^{b}=Δ^{α/2}+\mathcal{S}^{b}$, where $$\mathcal{S}^{b}f(x):=\lim_{\varepsilon\to 0}\mathcal{A}(d,-β)\int_{|z|>\varepsilon}\left(f(x+z)-f(x)\right)\frac{b(x,z)}{|z|^{d+β}}\,dy.$$ Here $b(x,z)$ is a bounded measurable function on $\mathbb{R}^{d}\times\mathbb{R}^{d}$ that is symmetric in $z$, and $\mathcal{A}(d,-β)$ is a normalizing constant so that when $b(x, z)\equiv 1$, $\mathcal{S}^{b}$ becomes the fractional Laplacian $Δ^{β/2}:=-(-Δ)^{β/2}$. In other words, $$\mathcal{L}^{b}f(x):=\lim_{\varepsilon\to 0}\mathcal{A}(d,-β)\int_{|z|>\varepsilon}\left(f(x+z)-f(x)\right) j^b(x, z)\,dz,$$ where $j^b(x, z):= \mathcal{A}(d,-α) |z|^{-(d+α)}+ \mathcal{A}(d,-β) b(x, z)|z|^{-(d+β)}$. It is recently established in Chen and Wang [arXiv:1312.7594 [math.PR]] that, when $j^b(x, z)\geq 0$ on $\mathbb{R}^d\times \mathbb{R}^d$, there is a conservative Feller process $X^{b}$ having $\mathcal{L}^b$ as its infinitesimal generator. In this paper we establish, under certain conditions on $b$, a uniform boundary Harnack principle for harmonic functions of $X^b$ (or equivalently, of $\mathcal{L}^b$) in any $κ$-fat open set. We further establish uniform gradient estimates for non-negative harmonic functions of $X^{b}$ in open sets.

math.PR↗

Law of large numbers for branching symmetric Hunt processes with measure-valued branching rates

We establish weak and strong law of large numbers for a class of branching symmetric Hunt processes with the branching rate being a smooth measure with respect to the underlying Hunt process, and the branching mechanism being general and state-dependent. Our work is motivated by recent work on strong law of large numbers for branching symmetric Markov processes by Chen-Shiozawa [J. Funct. Anal., 250, 374--399, 2007] and for branching diffusions by Engländer-Harris-Kyprianou [Ann. Inst. Henri Poincaré Probab. Stat., 46, 279--298, 2010]. Our results can be applied to some interesting examples that are covered by neither of these papers.

math.PR↗

Dirichlet heat kernel estimates for fractional Laplacian under non-local perturbation

For $d\ge 2$ and $0<β<α<2$, consider a family of non-local operators $\mathcal{L}^{b}=Δ^{α/2}+\mathcal{S}^{b}$ on $\mathbb{R}^d$, where $$ \mathcal{S}^{b}f(x):=\lim_{\varepsilon\to 0}\mathcal{A}(d,-β)\int_{ \{z\in \mathbb{R}^d: |z|>\varepsilon\}} (f(x+z)-f(x))\frac{b(x,z)}{|z|^{d+β}}\,dz, $$ and $b(x,z)$ is a bounded measurable function on $\mathbb{R}^{d}\times\mathbb{R}^{d}$ with $b(x,z)=b(x,-z)$ for every $x,z\in\mathbb{R}^{d}$. Here ${\cal A}(d, -β)$ is a normalizing constant so that $\mathcal{S}^b=-(-Δ)^{β/2}$ when $b(x, z)\equiv 1$. It was recently shown in Chen and Wang [arXiv:1312.7594 [math.PR]] that when $b(x, z) \geq -\frac{\mathcal{A}(d, -α)} {\mathcal{A}(d, -β)}\, |z|^{β-α}$, then $\mathcal{L}^b$ admits a unique fundamental solution $p^b(t, x, y)$ which is strictly positive and continuous. The kernel $p^b(t, x, y)$ uniquely determines a conservative Feller process $X^b$, which has strong Feller property. The Feller process $X^b$ is also the unique solution to the martingale problem of $(\mathcal{L}^b, \mathcal{S}(\mathbb{R}^d))$, where $\mathcal{S}(\mathbb{R}^d)$ denotes the space of tempered functions on $\mathbb{R}^d$. In this paper, we are concerned with the subprocess $X^{b,D}$ of $X^{b}$ killed upon leaving a bounded $C^{1,1}$ open set $D\subset \mathbb{R}^d$. We establish explicit sharp two-sided estimates for the transition density function of $X^{b, D}$.

math.PR↗

Photonic analog computing with integrated silicon waveguides

The spectra of silicon integrated waveguides are tailored to process analog computing (i.e.,differential and integral) in optical domain with huge bandwidth.With the theory of signal and system, we design some silicon integrated devices to implement photonic differentiator and optical differential equation solver. The basic principle is to tailor the spectra of silicon integrated waveguides to meet the requirements of analog computing circuits. These analog photonic integrated circuits are very promising in future computing systems with high speed, low cost, and compact size. We also plan to employ these basic computing units in more complex computing modules.

physics.optics↗

Conditional limit theorems for critical continuous-state branching processes

In this paper we study the conditional limit theorems for critical continuous-state branching processes with branching mechanism $ψ(λ)=λ^{1+α}L(1/λ)$ where $α\in [0,1]$ and $L$ is slowly varying at $\infty$. We prove that if $α\in (0,1]$, there are norming constants $Q_{t}\to 0$ (as $t\uparrow +\infty$) such that for every $x>0$, $P_{x}\left(Q_{t}X_{t}\in\cdot|X_{t}>0\right)$ converges weakly to a non-degenerate limit. The converse assertion is also true provided the regularity of $ψ$ at 0. We give a conditional limit theorem for the case $α=0$. The limit theorems we obtain in this paper allow infinite variance of the branching process.

math.PR↗

Experimental observation of optical differentiation and optical Hilbert transformation using a single SOI microdisk chip

Optical differentiation and optical Hilbert transformation play important roles in communications, computing, information processing and signal analysis in optical domain which offering huge bandwidth. Meanwhile, silicon-based photonic integrated circuits are preferable in all-optical signal processing due to their intrinsic advantages of low power consumption, compact footprint and ultra-high speed. In this study, we analyze the interrelation between first-order optical differentiation and optical Hilbert transformation and then experimentally demonstrate a feasible integrated scheme which can simultaneously function as first-order optical differentiation and optical Hilbert transformation based on a single microdisk resonator. This finding may motivate the development of integrated optical signal processors.

physics.optics↗