Stochastic Jump-Diffusion Process for Computing Medial Axes in Markov Random Fields

ZhuSong-Chun

IF: 23.6

1999-01-01

IEEE Transactions on Pattern Analysis and Machine Intelligence

Abstract:This paper proposes a statistical framework for computing medial axes of 2D shapes. In this paper, the computation of medial axes is posed as a statistical inference problem not as a mathematical t...

What problem does this paper attempt to address?

Stochastic Jump-Diffusion Process for Computing Medial Axes in Markov Random Fields

Song-Chun Zhu

DOI: https://doi.org/10.1109/34.809109

1999-01-01

Abstract:This paper proposes a statistical framework for computing medial axes of 2D shapes. In this paper, the computation of medial axes is posed as a statistical inference problem not as a mathematical transform. This paper contributes to three aspects in computing medial axes. 1) Prior knowledge is adopted for axes and junctions so that axes around junctions are regularized. 2) Multiple interpretations of axes are possible, each being assigned a probability. 3) A novel stochastic jump-diffusion process is proposed for estimating both axes and junctions in Markov random fields. We argue that the stochastic algorithm for computing medial axes is compatible with existing algorithms for image segmentation, such as region growing [31], snake [7], and region competition [26]. Thus, our method provides a new direction for computing medial axes from texture images.Experiments are demonstrated on both synthetic and real 2D shapes. This algorithm has been successfully applied to shape learning and sampling in a companion paper [30].
Stochastic Computation of Medial Axis in Markov Random Fields

SC Zhu

DOI: https://doi.org/10.1109/cvpr.1998.698590

1998-01-01

Abstract:In this paper the computation of medial axis is posed as a statistical inference problem not as a mathematical transform. This method provides answers to two essential problems in computing the medial axis representation. I) Prior knowledge are adopted for axes and junctions so that the axes around junctions become well defined. II) A stochastic jump-diffusion process is proposed for estimating medial axis in a Markov random field. We argue that the stochastic algorithm for computing medial axis is compatible with existing algorithms for image segmentation, such as snake [7] and region competition [18]. Thus it provides a new direction for computing medial axis from real textured images. Experiments are demonstrated on both synthetic and real shapes.
Random medium modeling based on FFT-MA

Yang XiuWei,Zhu PeiMin,Mao NingBo,Xu ZhiWei,Xiao Dan

DOI: https://doi.org/10.6038/cjg2018L0087

2018-01-01

Abstract:Random medium model is an effective tool to characterize the small-scale heterogeneities of the earth. It has been widely used in analysis of the seismic scatter-wave field, reservoir characterization and so forth. In this aspect, a fast and accurate random medium modeling method is the basis for relevant research. This paper introduces the FFT-MA method into random medium modeling, analyzes its advantages, and presents a non-stationary random medium modeling method based on FFT-MA. The modeling experiments indicate that comparing with the traditional random medium modeling method based on spectral decomposition theorem, the method based on FFT-MA can produce the randomness in the spatial domain, which can uncouple the random numbers from the structural parameters. It constructs the link between random numbers and the model in the spatial domain so that it can be used in sub-regional modeling and local modification. It also benefits the non-stationary random medium modeling for changing size of a modeling window along with the change of autocorrelation length. This avoids constructing the whole model each time at every node, thereby improving the modeling efficiency. In one word, FFT-MA random medium modeling can create both stationary and non-stationary random medium models satisfying the expected autocorrelation functions, and is highly efficient, flexible and practical.
Computing Medial Axis Transformations of 2D Point Clouds.

Yanjun Zhong,Falai Chen

DOI: https://doi.org/10.1016/j.gmod.2018.03.004

IF: 1.094

2018-01-01

Graphical Models

Abstract:In this paper, we propose a robust method to compute the medial axis transformation of a 2D point cloud with noise and/or missing data. The basic approach is to first compute the signed distance function of the point cloud by solving the Eikonal equation, and then an approximation of the signed distance function is obtained by sparse optimization technique. The medial axis of the point cloud corresponds to the non-smooth ridge of the distance function which can be extracted by checking the norm of the gradient of the distance function. The medial axis is segmented into branches and a compact spline representation of each branch can be obtained. We perform experiments on various examples and compare our method with state-of-the-art methods. Experimental results demonstrate that our method outperforms previous methods in obtaining accurate and reliable representations for medial axis transformations of noisy 2D point clouds.
Random fields representation over manifolds via isometric feature mapping‐based dimension reduction

De‐Cheng Feng,Yan‐Ping Liang,Xiaodan Ren,Jie Li

DOI: https://doi.org/10.1111/mice.12752

2021-08-10

Abstract:Generating random fields over irregular geometries (e.g., two-dimensional (2D) manifolds embedded in the three-dimensional (3D) Euclidean space) is a great challenge because the geometry structure is complex and the correlation function can hardly be derived, thus the traditional methods, for example, spatial discretization methods or series expansion methods, cannot be directly adopted. To solve this issue, the present paper develops a two-stage strategy to simulate random fields over manifolds. The core idea is to map the manifolds into the 2D Euclidean space through Isometric feature mapping (Isomap), with which the geodesic distance between points in the mapped 2D Euclidean space and the original manifold space is kept as the same. Therefore, the correlation function can be readily derived and the conventional methods can be directly employed to generate random fields. To validate the proposed method, several different types of manifolds are dimensionally reduced to 2D planar domains, and the stochastic harmonic function method is used to generate the random fields. Finally, case studies on the stochastic finite element analysis of two different structures are also performed to demonstrate the applicability and efficiency of the proposed method.

transportation science & technology,construction & building technology,computer science, interdisciplinary applications,engineering, civil
Diffusion bridges for stochastic Hamiltonian systems and shape evolutions

Alexis Arnaudon,Frank van der Meulen,Moritz Schauer,Stefan Sommer

DOI: https://doi.org/10.1137/21M1406283

2021-12-02

Abstract:Stochastically evolving geometric systems are studied in shape analysis and computational anatomy for modelling random evolutions of human organ shapes. The notion of geodesic paths between shapes is central to shape analysis and has a natural generalisation as diffusion bridges in a stochastic setting. Simulation of such bridges is key to solve inference and registration problems in shape analysis. We demonstrate how to apply state-of-the-art diffusion bridge simulation methods to recently introduced stochastic shape deformation models thereby substantially expanding the applicability of such models. We exemplify these methods by estimating template shapes from observed shape configurations while simultaneously learning model parameters.

Numerical Analysis,Computational Engineering, Finance, and Science,Computational Physics
Statistical Depth in Spatial Point Process

Xinyu Zhou,Wei Wu

DOI: https://doi.org/10.3390/math12040595

IF: 2.4

2024-02-17

Mathematics

Abstract:Statistical depth is widely used as a powerful tool to measure the center-outward rank of multivariate and functional data. Recent studies have introduced the notion of depth to the temporal point process, which exhibits randomness in the cardinality as well as distribution in the observed events. The proposed methods can well capture the rank of a point process in a given time interval, where a critical step is to measure the rank by using inter-arrival events. In this paper, we propose to extend the depth concept to multivariate spatial point process. In this case, the observed process is in a multi-dimensional location and there are no conventional inter-arrival events in the temporal process. We adopt the newly developed depth in metric space by defining two different metrics, namely the penalized metric and the smoothing metric, to fully explore the depth in the spatial point process. The mathematical properties and the large sample theory, as well as depth-based hypothesis testings, are thoroughly discussed. We then use several simulations to illustrate the effectiveness of the proposed depth method. Finally, we apply the new method in a real-world dataset and obtain desirable ranking performance.

mathematics
Scaling Riemannian Diffusion Models

Aaron Lou,Minkai Xu,Stefano Ermon

2023-10-31

Abstract:Riemannian diffusion models draw inspiration from standard Euclidean space diffusion models to learn distributions on general manifolds. Unfortunately, the additional geometric complexity renders the diffusion transition term inexpressible in closed form, so prior methods resort to imprecise approximations of the score matching training objective that degrade performance and preclude applications in high dimensions. In this work, we reexamine these approximations and propose several practical improvements. Our key observation is that most relevant manifolds are symmetric spaces, which are much more amenable to computation. By leveraging and combining various ansätze, we can quickly compute relevant quantities to high precision. On low dimensional datasets, our correction produces a noticeable improvement, allowing diffusion to compete with other methods. Additionally, we show that our method enables us to scale to high dimensional tasks on nontrivial manifolds. In particular, we model QCD densities on $SU(n)$ lattices and contrastively learned embeddings on high dimensional hyperspheres.

Machine Learning,Differential Geometry
Quasi-Medial Distance Field (Q-MDF): A Robust Method for Approximating and Discretizing Neural Medial Axis

Jiayi Kong,Chen Zong,Jun Luo,Shiqing Xin,Fei Hou,Hanqing Jiang,Chen Qian,Ying He

2024-10-23

Abstract:The medial axis, a lower-dimensional shape descriptor, plays an important role in the field of digital geometry processing. Despite its importance, robust computation of the medial axis transform from diverse inputs, especially point clouds with defects, remains a significant challenge. In this paper, we tackle the challenge by proposing a new implicit method that diverges from mainstream explicit medial axis computation techniques. Our key technical insight is the difference between the signed distance field (SDF) and the medial field (MF) of a solid shape is the unsigned distance field (UDF) of the shape's medial axis. This allows for formulating medial axis computation as an implicit reconstruction problem. Utilizing a modified double covering method, we extract the medial axis as the zero level-set of the UDF. Extensive experiments show that our method has enhanced accuracy and robustness in learning compact medial axis transform from thorny meshes and point clouds compared to existing methods.

Computer Vision and Pattern Recognition,Graphics
An efficient Wasserstein-distance approach for reconstructing jump-diffusion processes using parameterized neural networks

Mingtao Xia,Xiangting Li,Qijing Shen,Tom Chou

2024-06-04

Abstract:We analyze the Wasserstein distance ($W$-distance) between two probability distributions associated with two multidimensional jump-diffusion processes. Specifically, we analyze a temporally decoupled squared $W_2$-distance, which provides both upper and lower bounds associated with the discrepancies in the drift, diffusion, and jump amplitude functions between the two jump-diffusion processes. Then, we propose a temporally decoupled squared $W_2$-distance method for efficiently reconstructing unknown jump-diffusion processes from data using parameterized neural networks. We further show its performance can be enhanced by utilizing prior information on the drift function of the jump-diffusion process. The effectiveness of our proposed reconstruction method is demonstrated across several examples and applications.

Machine Learning,Probability,Applications,Methodology
Simultaneous Inference for Non-Stationary Random Fields, with Application to Gridded Data Analysis

Yunyi Zhang,Zhou Zhou

2024-09-02

Abstract:Current statistics literature on statistical inference of random fields typically assumes that the fields are stationary or focuses on models of non-stationary Gaussian fields with parametric/semiparametric covariance families, which may not be sufficiently flexible to tackle complex modern-era random field data. This paper performs simultaneous nonparametric statistical inference for a general class of non-stationary and non-Gaussian random fields by modeling the fields as nonlinear systems with location-dependent transformations of an underlying `shift random field'. Asymptotic results, including concentration inequalities and Gaussian approximation theorems for high dimensional sparse linear forms of the random field, are derived. A computationally efficient locally weighted multiplier bootstrap algorithm is proposed and theoretically verified as a unified tool for the simultaneous inference of the aforementioned non-stationary non-Gaussian random field. Simulations and real-life data examples demonstrate good performances and broad applications of the proposed algorithm.

Statistics Theory,Methodology
Fast Computation of Stationary Joint Probability Distribution of Sparse Markov Chains

Weiyang Ding,Michael Ng,Yimin Wei

DOI: https://doi.org/10.1016/j.apnum.2017.10.008

IF: 2.994

2017-01-01

Applied Numerical Mathematics

Abstract:In this paper, we study a fast algorithm for finding stationary joint probability distributions of sparse Markov chains or multilinear PageRank vectors which arise from data mining applications. In these applications, the main computational problem is to calculate and store solutions of many unknowns in joint probability distributions of sparse Markov chains. Our idea is to approximate large-scale solutions of such sparse Markov chains by two components: the sparsity component and the rank-one component. Here the non-zero locations in the sparsity component refer to important associations in the joint probability distribution and the rank-one component refers to a background value of the solution. We propose to determine solutions by formulating and solving sparse and rank-one optimization problems via closed form solutions. The convergence of the truncated power method is established. Numerical examples of multilinear PageRank vector calculation and second-order web-linkage analysis are presented to show the efficiency of the proposed method. It is shown that both computation and storage are significantly reduced by comparing with the traditional power method.
Statistical inference for ergodic diffusion with Markovian switching

Yuzhong Cheng,Hiroki Masuda

2024-10-15

Abstract:This study explores a Gaussian quasi-likelihood approach for estimating parameters of diffusion processes with Markovian regime switching. Assuming the ergodicity under high-frequency sampling, we will show the asymptotic normality of the unknown parameters contained in the drift and diffusion coefficients and present a consistent explicit estimator for the generator of the Markov chain. Simulation experiments are conducted to illustrate the theoretical results obtained.

Statistics Theory
Diffusion Probabilistic Models for 3D Point Cloud Generation

Shitong Luo,Wei Hu

DOI: https://doi.org/10.1109/cvpr46437.2021.00286

2021-06-01

Abstract:We present a probabilistic model for point cloud generation, which is fundamental for various 3D vision tasks such as shape completion, upsampling, synthesis and data augmentation. Inspired by the diffusion process in non-equilibrium thermodynamics, we view points in point clouds as particles in a thermodynamic system in contact with a heat bath, which diffuse from the original distribution to a noise distribution. Point cloud generation thus amounts to learning the reverse diffusion process that transforms the noise distribution to the distribution of a desired shape. Specifically, we propose to model the reverse diffusion process for point clouds as a Markov chain conditioned on certain shape latent. We derive the variational bound in closed form for training and provide implementations of the model. Experimental results demonstrate that our model achieves competitive performance in point cloud generation and auto-encoding. The code is available at https://github.com/luost26/diffusion-point-cloud.
Data augmentation-based statistical inference of diffusion processes

Yasen Wang,Cheng Cheng,Hongwei Sun,Junyang Jin,Huazhen Fang

DOI: https://doi.org/10.1063/5.0124763

2023-03-10

Abstract:The identification of diffusion processes is challenging for many real-world systems with sparsely sampled observation data. In this work, we propose a data augmentation-based sparse Bayesian learning method to identify a class of diffusion processes from sparsely sampled data. We impute latent unsampled diffusion paths between adjacent observations and construct a candidate model for the diffusion processes with the sparsity-inducing prior on model parameters. Given the augmented data and candidate model, we investigate the full joint posterior distribution of all the parameters and latent diffusion paths under a Bayesian learning framework. We then design a Markov chain Monte Carlo sampler with non-degenerate acceptance probability on system dimension to draw samples from the posterior distribution to estimate the parameters and latent diffusion paths. Particularly, the proposed method can handle sparse data that are regularly or irregularly sampled in time. Simulations on the well-known Langevin equation, homogeneous diffusion in a symmetric double-well potential, and stochastic Lotka–Volterra equation demonstrate the effectiveness and considerable accuracy of the proposed method.
Equation-free, multiscale computation for unsteady random diffusion

Dongbin Xiu,Ioannis Kevrekidis

DOI: https://doi.org/10.48550/arXiv.math/0504273

2005-04-13

Abstract:We present an ``equation-free'' multiscale approach to the simulation of unsteady diffusion in a random medium. The diffusivity of the medium is modeled as a random field with short correlation length, and the governing equations are cast in the form of stochastic differential equations. A detailed fine-scale computation of such a problem requires discretization and solution of a large system of equations, and can be prohibitively time-consuming. To circumvent this difficulty, we propose an equation-free approach, where the fine-scale computation is conducted only for a (small) fraction of the overall time. The evolution of a set of appropriately defined coarse-grained variables (observables) is evaluated during the fine-scale computation, and ``projective integration'' is used to accelerate the integration. The choice of these coarse variables is an important part of the approach: they are the coefficients of pointwise polynomial expansions of the random solutions. Such a choice of coarse variables allows us to reconstruct representative ensembles of fine-scale solutions with "correct" correlation structures, which is a key to algorithm efficiency. Numerical examples demonstrating accuracy and efficiency of the approach are presented.

Numerical Analysis
Stochastic force inference via density estimation

Victor Chardès,Suryanarayana Maddu,Michael J. Shelley

2023-10-04

Abstract:Inferring dynamical models from low-resolution temporal data continues to be a significant challenge in biophysics, especially within transcriptomics, where separating molecular programs from noise remains an important open problem. We explore a common scenario in which we have access to an adequate amount of cross-sectional samples at a few time-points, and assume that our samples are generated from a latent diffusion process. We propose an approach that relies on the probability flow associated with an underlying diffusion process to infer an autonomous, nonlinear force field interpolating between the distributions. Given a prior on the noise model, we employ score-matching to differentiate the force field from the intrinsic noise. Using relevant biophysical examples, we demonstrate that our approach can extract non-conservative forces from non-stationary data, that it learns equilibrium dynamics when applied to steady-state data, and that it can do so with both additive and multiplicative noise models.

Machine Learning,Biological Physics,Quantitative Methods
Interacting Jump Processes Preserve Semi-Global Markov Random Fields on Path Space

Ankan Ganguly,Kavita Ramanan

DOI: https://doi.org/10.48550/arXiv.2210.09253

2022-10-18

Abstract:Consider a system of interacting particles indexed by the nodes of a graph whose vertices are equipped with marks representing parameters of the model such as the environment or initial data. Each particle takes values in a countable state space and evolves according to a (possibly non-Markovian) continuous-time pure jump process whose jump intensities depend only on its own state (or history) and marks as well as the states (or histories) and marks of particles and edges in its neighborhood in the graph. Under mild conditions on the jump intensities, it is shown that the trajectories of the interacting particle system exhibit a certain local or semi-global Markov random field property whenever the initial condition satisfies the same property. Our results complement recent works that establish the preservation of a local second-order Markov random field property for interacting diffusions. Our proof methodology in the context of jump processes is different, and works directly on infinite graphs, thereby bypassing any limiting arguments. Our results apply to models arising in diverse fields including statistical physics, neuroscience, epidemiology and opinion dynamics, and have direct applications to the study of marginal distributions of interacting particle systems on Cayley trees.

Probability
Stochastic Gradient Geodesic MCMC Methods.

Chang Liu,Jun Zhu,Yang Song

2016-01-01

Abstract:We propose two stochastic gradient MCMC methods for sampling from Bayesian posterior distributions defined on Riemann manifolds with a known geodesic flow, e.g. hyperspheres. Our methods are the first scalable sampling methods on these manifolds, with the aid of stochastic gradients. Novel dynamics are conceived and 2nd-order integrators are developed. By adopting embedding techniques and the geodesic integrator, the methods do not require a global coordinate system of the manifold and do not involve inner iterations. Synthetic experiments show the validity of the method, and its application to the challenging inference for spherical topic models indicate practical usability and efficiency.
Unraveling the Smoothness Properties of Diffusion Models: A Gaussian Mixture Perspective

Yingyu Liang,Zhenmei Shi,Zhao Song,Yufa Zhou

2024-10-14

Abstract:Diffusion models have made rapid progress in generating high-quality samples across various domains. However, a theoretical understanding of the Lipschitz continuity and second momentum properties of the diffusion process is still lacking. In this paper, we bridge this gap by providing a detailed examination of these smoothness properties for the case where the target data distribution is a mixture of Gaussians, which serves as a universal approximator for smooth densities such as image data. We prove that if the target distribution is a $k$-mixture of Gaussians, the density of the entire diffusion process will also be a $k$-mixture of Gaussians. We then derive tight upper bounds on the Lipschitz constant and second momentum that are independent of the number of mixture components $k$. Finally, we apply our analysis to various diffusion solvers, both SDE and ODE based, to establish concrete error guarantees in terms of the total variation distance and KL divergence between the target and learned distributions. Our results provide deeper theoretical insights into the dynamics of the diffusion process under common data distributions.

Machine Learning,Artificial Intelligence,Computer Vision and Pattern Recognition

Stochastic Jump-Diffusion Process for Computing Medial Axes in Markov Random Fields

Stochastic Jump-Diffusion Process for Computing Medial Axes in Markov Random Fields

Stochastic Computation of Medial Axis in Markov Random Fields

Random medium modeling based on FFT-MA

Computing Medial Axis Transformations of 2D Point Clouds.

Random fields representation over manifolds via isometric feature mapping‐based dimension reduction

Diffusion bridges for stochastic Hamiltonian systems and shape evolutions

Statistical Depth in Spatial Point Process

Scaling Riemannian Diffusion Models

Quasi-Medial Distance Field (Q-MDF): A Robust Method for Approximating and Discretizing Neural Medial Axis

An efficient Wasserstein-distance approach for reconstructing jump-diffusion processes using parameterized neural networks

Simultaneous Inference for Non-Stationary Random Fields, with Application to Gridded Data Analysis

Fast Computation of Stationary Joint Probability Distribution of Sparse Markov Chains

Statistical inference for ergodic diffusion with Markovian switching

Diffusion Probabilistic Models for 3D Point Cloud Generation

Data augmentation-based statistical inference of diffusion processes

Equation-free, multiscale computation for unsteady random diffusion

Stochastic force inference via density estimation

Interacting Jump Processes Preserve Semi-Global Markov Random Fields on Path Space

Stochastic Gradient Geodesic MCMC Methods.

Unraveling the Smoothness Properties of Diffusion Models: A Gaussian Mixture Perspective