Bathymetry is a key element in the modeling of river systems for flood mapping, geomorphology, or stream habitat characterization. Standard practices rely on the interpolation of in situ depth measurements obtained wi...
详细信息
Bathymetry is a key element in the modeling of river systems for flood mapping, geomorphology, or stream habitat characterization. Standard practices rely on the interpolation of in situ depth measurements obtained with differential GPS or total station surveys, while more advanced techniques involve bathymetric LiDAR or acoustic soundings. However, these high-resolution active techniques are not so easily applied over large areas. Alternative methods using passive optical imagery present an interesting trade-off: they rely on the fact that wavelengths composing solar radiation are not attenuated at the same rates in water. Under certain assumptions, the logarithm of the ratio of radiances in two spectral bands is linearly correlated with depth. In this study, we go beyond these ratio methods in defining a multispectral hue that retains all spectral information. Given n coregistered bands, this spectral invariant lies on the (n-2)-sphere embedded in Rn-1, denoted Sn-2 and tagged 'hue hypersphere'. It can be seen as a generalization of the RGB 'color wheel' (S1) in higher dimensions. We use this mapping to identify a hue-depth relation in a 35 km reach of the Garonne River, using high resolution (0.50 m) airborne imagery in four bands and data from 120 surveyed cross-sections. The distribution of multispectral hue over river pixels is modeled as a mixture of two components: one component represents the distribution of substrate hue, while the other represents the distribution of 'deep water' hue;parameters are fitted such that membership probability for the 'deep' component correlates with depth.
In this paper, we address the problem of combining linear support vector machines (SVMs) for classification of large-scale nonlinear datasets. The motivation is to exploit both the efficiency of linear SVMs (LSVMs) in...
详细信息
In this paper, we address the problem of combining linear support vector machines (SVMs) for classification of large-scale nonlinear datasets. The motivation is to exploit both the efficiency of linear SVMs (LSVMs) in learning and prediction and the power of nonlinear SVMs in classification. To this end, we develop a LSVM mixture model that exploits a divide-and-conquer strategy by partitioning the feature space into subregions of linearly separable datapoints and learning a LSVM for each of these regions. We do this implicitly by deriving a generative model over the joint data and label distributions. Consequently, we can impose priors on the mixing coefficients and do implicit model selection in a top-down manner during the parameter estimation process. This guarantees the sparsity of the learned model. Experimental results show that the proposed method can achieve the efficiency of LSVMs in the prediction phase while still providing a classification performance comparable to nonlinear SVMs.
Breast tumor is one of the causes of women's death in the world after cardiovascular diseases. Recently, the diagnosis and treatment planning of this kind of tumors are based on magnetic resonance imaging techniqu...
详细信息
Breast tumor is one of the causes of women's death in the world after cardiovascular diseases. Recently, the diagnosis and treatment planning of this kind of tumors are based on magnetic resonance imaging techniques, which are the reference imaging modality in breast tumors analysis since it can better differentiate soft tissues (compared to mammography and ultrasound). Segmentation of the breast cancer is a very important task for cancer response prediction in neoadjuvant chemotherapy treatment based either on texture analysis or parametric response maps. In most of the previous works in the literature, the segmentation is generally done with manual annotation of tumor regions, which is time-consuming and error-prone. In this paper, we propose a new strategy for an automatic segmentation of breast tumors in MRI images. We propose first to separate the two breasts, and then, we use the expectation-maximization algorithm to segment and detect the tumor lesion.
Block and Basu bivariate exponential distribution is one of the most popular absolutely continuous bivariate distributions. In this article, we have considered a Block-Basu type bivariate Pareto distribution. Here in ...
详细信息
Block and Basu bivariate exponential distribution is one of the most popular absolutely continuous bivariate distributions. In this article, we have considered a Block-Basu type bivariate Pareto distribution. Here in the standard manner, first Marshall-Olkin type singular bivariate distribution has been constructed, and then by taking away the singular component similar to the Block and Basu model, an absolute continuous BB-BVPA model has been constructed. Further, the location and scale parameters also have been introduced. Therefore, the model has seven parameters. We drive different properties of this absolutely continuous distribution. Since the maximum likelihood estimators of the parameters cannot be expressed in a closed form, we propose to use an EM algorithm to compute the estimators of the model parameters. Some simulation experiments have been performed for illustrative purposes. We fit the model to rainfall data in the context of landslide risk estimation. The methods are implemented in the publicly available R package bvpa.
Structure function, which quantitatively represents the relation between system states and unit states, is essential for system reliability assessment and oftentimes may not be known in advance due to complicated inte...
详细信息
Structure function, which quantitatively represents the relation between system states and unit states, is essential for system reliability assessment and oftentimes may not be known in advance due to complicated interactions among units. In this article, a dynamic Bayesian network (DBN) model is put forth to leverage incomplete observation sequences of hierarchical multi-state systems for structure function learning. To achieve a consistent structure function at different time instants, a customized expectation-maximization (EM) algorithm with parameter modularization is proposed and executed by two steps: (1) filling the missing values in the incomplete observation sequences with their expectations to break the dependencies among nodes;(2) decomposing the graphical network into V-shape structures, and then integrating the identical V-shape structures at different time slices to learn the parameters in the DBN model. Based on the learned DBN model, system state distribution and reliability function over time can be readily assessed. Two illustrative examples are presented and the results demonstrate that the structure function of a hierarchical multi-state system can be accurately learned despite the incompleteness of observation sequences.
This paper addresses the problem of analyzing the performance of 3D face alignment (3DFA), or facial landmark localization. This task is usually supervised, based on annotated datasets. Nevertheless, in the particular...
详细信息
This paper addresses the problem of analyzing the performance of 3D face alignment (3DFA), or facial landmark localization. This task is usually supervised, based on annotated datasets. Nevertheless, in the particular case of 3DFA, the annotation process is rarely error-free, which strongly biases the results. Alternatively, unsupervised performance analysis (UPA) is investigated. The core ingredient of the proposed methodology is the robust estimation of the rigid transformation between predicted landmarks and model landmarks. It is shown that the rigid mapping thus computed is affected neither by non-rigid facial deformations, due to variabilities in expression and in identity, nor by landmark localization errors, due to various perturbations. The guiding idea is to apply the estimated rotation, translation and scale to a set of predicted landmarks in order to map them onto a mathematical home for the shape embedded in these landmarks (including possible errors). UPA proceeds as follows: (i) 3D landmarks are extracted from a 2D face using the 3DFA method under investigation;(ii) these landmarks are rigidly mapped onto a canonical (frontal) pose, and (iii) a statistically-robust confidence score is computed for each landmark. This allows to assess whether the mapped landmarks lie inside (inliers) or outside (outliers) a confidence volume. An experimental evaluation protocol, that uses publicly available datasets and several 3DFA software packages associated with published articles, is described in detail. The results show that the proposed analysis is consistent with supervised metrics and that it can be used to measure the accuracy of both predicted landmarks and of automatically annotated 3DFA datasets, to detect errors and to eliminate them. Source code and supplemental materials for this paper are publicly available at https://***/robotlearn/upa3dfa/.
In recent years, the online blogging community is growing bigger as the social network service. When it is growing, the blog posts are increasing day by day. Generally speaking, people were using the blog search engin...
详细信息
In recent years, the online blogging community is growing bigger as the social network service. When it is growing, the blog posts are increasing day by day. Generally speaking, people were using the blog search engines to search and recommend potentially interesting blog posts. When people search from the blog search engines, they were faced with two major problems: synonymy (two different terms with the same meaning) and polysemy (a term with different meanings). In this paper, we use two semantic analysis methods, Latent Semantic Indexing (LSI) and Probabilistic LSI (PLSI), to solve these two problems. LSI uses singular value decomposition as the fundamental method to capture the synonymous relationship between terms. PLSI uses the expectation-maximization algorithm for parameter estimation to additionally deal with the problem of polysemy. Although PLSI can gracefully deal with these two semantic problems, it needs a huge computing time. To solve the problem of computing time, in this paper, we propose a novel termination mechanism to dynamically determine the required number of iterations for PLSI. According to the experiment results, the result derived from our mechanism can not only deal with these two semantic problems but also reach a cost-effective solution.
Two-phase flow is a complex phenomenon present in several industrial applications such as chemical reactors, power generation, and in the exploration, production, and transport of oil and natural gas. The classificati...
详细信息
Two-phase flow is a complex phenomenon present in several industrial applications such as chemical reactors, power generation, and in the exploration, production, and transport of oil and natural gas. The classification of the flow pattern is a fundamental step in such applications, as it influences several derived parameters and sub-processes such as flow rate, void fraction, and pressure drop estimation. In this paper, we propose an objective approach for classifying flow patterns using time series of void fraction (from a wire-mesh sensor), signal processing and machine learning. As novel approach, the time series is modeled as a stochastic process of independent and identically distributed samples with probability density function described by a Gaussian mixture model. The estimated parameters of the mixture are then fed into a Support Vector Machine (SVM), yielding the flow pattern classification. Tests were performed with a vertical liquid-gas flow database from a 52.3-mm-diameter pipe and the results indicate a great potential for application in real systems. The average accuracy and F-score obtained was higher than 0.94 for different test sets, with standard deviation lower than 0.08 for accuracy a lower than 0.11 for F-Score, demonstrating the efficiency and generalization of the proposed method.
Bayesian methods have been extended for the linear system identification problem in the past ten years. The traditional Bayesian identification selects a Gaussian prior and considers the tuning of kernels, i.e., the c...
详细信息
Bayesian methods have been extended for the linear system identification problem in the past ten years. The traditional Bayesian identification selects a Gaussian prior and considers the tuning of kernels, i.e., the covariance matrix of a Gaussian prior. However, Gaussian priors cannot express the system information appropriately for identifying a positive finite impulse response (FIR) model. This paper exploits the truncated Gaussian prior and develops Bayesian identification procedures for positive FIR models. The proposed parameterizations in the truncated Gaussian prior can reflect the decay rate and the correlation of the impulse response of the system to be identified. The expectation-maximization (EM) algorithm is tailored to the hyperparameter estimation problem of positive system identification with the truncated Gaussian prior. Numerical experiments compare the truncated Gaussian prior to the traditional Gaussian prior for positive FIR system identification. The simulation results demonstrate that the truncated Gaussian prior outperforms the Gaussian prior. (C) 2020 Elsevier B.V. All rights reserved.
This paper introduces a robust identification solution for the linear parameter varying Autoregressive Exogenous systems with outlier-contaminated outputs. The Laplace distribution with heavy tails and the expectation...
详细信息
This paper introduces a robust identification solution for the linear parameter varying Autoregressive Exogenous systems with outlier-contaminated outputs. The Laplace distribution with heavy tails and the expectationmaximizationalgorithm are combined to build the robust system identification framework. To overcome the obstacles brought by the outliers, the Laplace distribution which can be decomposed into infinite Gaussian components, is applied to mathematically model the system noise. The problem of parameter estimation is solved using the expectationmaximizationalgorithm, and the equations to infer the system model and noise parameters are simultaneously provided in the developed identification method. Finally, the verification tests performed on a numerical example and a mechanical unit are used to prove the validity of the developed identification method.
暂无评论