K-means algorithm is widely used in spatial clustering. It takes the mean value of each cluster centroid as Heuristic information, so it has some disadvantages: sensitive to initial and instability. The improved clustering referred best centriod which searched during optimization centroid. That increased searching probability around stability algorithm. experiment on two groups representative dataset proved that performs better global less

Naturalistic driving action recognition plays an important role in understanding drivers' distracted behaviors the traffic environment. The main challenge of this task is accurate localization temporal boundary for each behavior video. Although many methods can identify categories, it difficult to predict boundaries since actions same category usually present large intra-class variation. In paper, we introduce a Coarse-to-Fine Boundary Localization method called CFBL, which obtains...

Normalized random measures with independent increments (NRMIs) represent a large class of Bayesian nonparametric priors and are widely used in the framework. In this paper, we provide posterior consistency analysis for these NRMIs through their characterizing Lévy intensities. Assumptions introduced on intensities to analyse verified multiple interesting examples. Another focus paper is Bernstein-von Mises theorem particular subclass NRMIs, namely normalized generalized gamma processes...

Homogeneous normalized random measures with independent increments (hNRMIs) represent a broad class of Bayesian nonparametric priors and thus are widely used. In this paper, we obtain the strong law large numbers, central limit theorem functional hNRMIs when concentration parameter $a$ approaches infinity. To quantify convergence rate obtained theorem, further study Berry-Esseen bound, which turns out to be form $O \left( \frac{1}{\sqrt{a}}\right)$. As an application present delta method,...

In this paper, we extend our prior research named DKIC [1] and propose the perceptual-oriented learned image compression method, PO-DKIC, which is shown in Figure 1 . Specifically, adopts a dynamic kernel-based residual block group to enhance transform coding an asymmetric space-channel context entropy model facilitate estimation of Gaussian parameters. Based on DKIC, PO-DKIC introduces PatchGAN LPIPS loss visual quality. Furthermore, maximize overall perceptual quality under rate...

In this paper, we extend our prior research named DKIC and propose the perceptual-oriented learned image compression method, PO-DKIC. Specifically, adopts a dynamic kernel-based residual block group to enhance transform coding an asymmetric space-channel context entropy model facilitate estimation of gaussian parameters. Based on DKIC, PO-DKIC introduces PatchGAN LPIPS loss visual quality. Furthermore, maximize overall perceptual quality under rate constraint, formulate challenge into...

In this paper, we introduce our hybrid image and video compression scheme enhanced by CNN-optimized in-loop filter. Specifically, a Structure Preserving in-Loop Filter (SPiLF) is incorporated in the codec Enhanced Compression Model (ECM), where two branches, i.e., gradient branch pixel branch, are developed based on dense residual unit (DRU). To provide pleasant visual quality, Generative adversarial networks (GAN) loss LPIPS further considered. Therefore, proposal mainly focusing...

We obtain the strong law of large numbers, Glivenko-Cantelli theorem, central limit functional theorem for various Bayesian nonparametric priors which include stick-breaking process with general weights, two-parameter Poisson-Dirichlet process, normalized inverse Gaussian generalized gamma and Dirichlet process. For we introduce two conditions such that hold. Except in case since finite dimensional distributions these processes are either hard to or complicated use even they available,...

Bayesian nonparametric models have been extensively developed and widely used in statistics, machine learning other areas since the ground breaking work of Ferguson. The fundamental is a special class random probability measures: Dirichlet processes. This paper introduces constructions, properties some recent developments processes as well their applications to estimation problems. We are also concerned with two-parameter Poisson-Dirichlet processes, Beta more general stick-breaking properties.

We obtain the empirical strong law of large numbers, Glivenko-Cantelli theorem, central limit functional theorem for various nonparametric Bayesian priors which include Dirichlet process with general stick-breaking weights, Poisson-Dirichlet process, normalized inverse Gaussian generalized gamma and process. For we introduce two conditions such that hold. Except in case since finite dimensional distributions these processes are either hard to or complicated use even they available, method...

Normalized random measures with independent increments represent a large class of Bayesian nonaprametric priors and are widely used in the nonparametric framework. In this paper, we provide posterior consistency analysis for normalized (NRMIs) through corresponding Levy intensities to characterize completely construction NRMIs. Assumptions introduced on analyze NRMIs verified multiple interesting examples. A focus paper is Bernstein-von Mises theorem generalized gamma process (NGGP) when...

A default assumption in reinforcement learning (RL) and optimal control is that observations arrive at discrete time points on a fixed clock cycle. Yet, many applications involve continuous-time systems where the discretization, principle, can be managed. The impact of discretization RL methods has not been fully characterized existing theory, but more detailed analysis its effect could reveal opportunities for improving data-efficiency. We address this gap by analyzing Monte-Carlo policy...

In order to solve the target search problem of multiple Unmanned Aerial Vehicles (UAVs) under condition implicit communication, this paper firstly establishes a multi-objective optimization model formation structure. The objective is maximize invulnerability information transmission topology and efficiency. constraints such as connectivity safety distance are considered. Then, proposes evolutionary algorithm CLUinD&O-MOEA based on NSGA-II, which combines clustering method improved crowding...