Hierarchical Decomposition Research Articles

A Dynamic Stochastic Integrated Climate–Economic Spatiotemporal Model for Agricultural Insurance Products

We propose a new methodology for area yield distribution modeling. We explore a variety of new hybrid data emulators for spatialtemporal statistical modeling of agricultural crop yields. The regression models explored leverage from a combination of model-generated and observed features incorporating climate/weather variables relating to temperature and precipitation over space and time and agricultural variables for farms such as crop allocations, crop types, land use, and crop rotation. This provides the modeling framework to achieve a hierarchical decomposition of the yield distribution at multiple spatial scales over time, allowing us to study both county- and individual farm-level information with suitably selected weighting functions that take into account farm size and crop type. A core component of our model framework is the climate model component that involves a spatialtemporal local approximate seasonal autoregressive integrated moving average Gaussian process (La-SARIMA-GP) model that is suitable for accurately studying local monthly temperatures and rainfalls. Upon construction of the crop yield model, we demonstrate that for practitioners, there is a clear incentive to consider such a model because it accommodates apportioning of the county yield information to the farms level. This is of significance for both individual and index-based agricultural crop insurance product design and for farm risk management. We demonstrate an application of our model in the insurance context of crop insurance risk pooling and insurance policy rating where we investigate the impact of different temporal and spatial interpolation methods on insurance loss ratios using a rating game.

Pose-Guided Hierarchical Semantic Decomposition and Composition for Human Parsing.

Human parsing is a fine-grained semantic segmentation task, which needs to understand human semantic parts. Most existing methods model human parsing as a general semantic segmentation, which ignores the inherent relationship among hierarchical human parts. In this work, we propose a pose-guided hierarchical semantic decomposition and composition framework for human parsing. Specifically, our method includes a semantic maintained decomposition and composition (SMDC) module and a pose distillation (PC) module. SMDC progressively disassembles the human body to focus on the more concise regions of interest in the decomposition stage and then gradually assembles human parts under the guidance of pose information in the composition stage. Notably, SMDC maintains the atomic semantic labels during both stages to avoid the error propagation issue of the hierarchical structure. To further take advantage of the relationship of human parts, we introduce pose information as explicit guidance for the composition. However, the discrete structure prediction in pose estimation is against the requirement of the continuous region in human parsing. To this end, we design a PC module to broadcast the maximum responses of pose estimation to form the continuous structure in the way of knowledge distillation. The experimental results on the look-into-person (LIP) and PASCAL-Person-Part datasets demonstrate the superiority of our method compared with the state-of-the-art methods, that is, 55.21% mean Intersection of Union (mIoU) on LIP and 69.88% mIoU on PASCAL-Person-Part.

Fast Finite-Time Tracking Consensus With Applications on Multiple Servo Motors

This article focuses on the fast finite-time tracking consensus for uncertain nonlinear multiagent systems. By gracefully combining the hierarchical decomposition, adding a barrier power integrator and disturbance compensation techniques, a new adaptive fast finite-time controller is proposed. Compared with the existing works, the proposed method has several distinguishing features: 1) It can achieve fast finite-time convergence with full state constraints; 2) it can deal with some fully unknown nonlinearities and state-dependent disturbance. Moreover, the nonlinearities can be related with the states of all the followers; 3) the input of the leader is not required to be zero, meaning a broad class of references signals can be generated; and 4) the controller is computationally simply and ready to be implemented. No fuzzy logic/neural networks are needed. Applications of the proposed method on multiple servo motors are also studied.

Algorithms for the extension two-dimensional fuzzy signal via decomposition and reconstruction in fuzzy wavelets

he decomposition of an image can be done in the following way: The image is split into a low-resolution part, which can be described by a smaller number of samples than the original image, and a signal difference, which describes the difference between the low-resolution image and the real coded image. Therefore, this low-resolution image is also decomposed into a low-resolution image and a difference image, making more efficient coding possible. This decomposition is repeated several times, so that a hierarchical image decomposition is created. Thus, the low-resolution image is only half the size of the original image. This reduced image is enlarged to the size of the original image. The result is a detailed image that is the same size as the original image. Our problem is: "Can we build algorithms allowing the decomposition and reconstruction of a signal in a fuzzy environment? We will answer in the affirmative. This construction is first made possible by one-dimensional fuzzy multiresolution analysis, which will later be extended to two dimensions. In the first part, we recall some definitions and main results obtained on the existence of fuzzy multiresolution analysis in the one-dimensional case, for the decomposition of a fuzzy image. The second part, based on this multi-resolution analysis, presents a fast construction algorithm for the analysis and synthesis of a fuzzy signal. Finally, the third part is nothing else than an extension of this algorithm to 2 dimensions, by the fuzzy multiresolution two-dimensional analysis

Hierarchical Tensor Decomposition of Module Partition for the Mechanical Simulation

The mechanical simulation experiment can provide guidelines for the structural design of materials, but the module partition of mechanical simulation experiments is still in its infancy. A mechanical simulation contour, e.g., strain and stress contour, has hierarchical characteristics. By analyzing the contour at different layers, the physical properties of the structure material can be improved. Current state-of-the-art methods cannot distinguish between simulation strain contours, as well as sparsely distributed spots of strain (stress concentrations) from simulation strain contour images, resulting in simulation data that does not accurately reflect real strain contours. In this paper, a Hierarchical Tensor Decomposition (HTD) method is proposed to extract hierarchical contours and stress concentrations from the simulation strain contours and then improve the mechanical simulation. HTD decomposes a tensor into three classes of components: the multi-smooth layers, the sparse spots layer, and the noise layer. The number of multismooth layers is determined by the scree plot, which is the difference between the smooth layers and the sparse spots layer. The proposed method is validated by several numerical examples, which demonstrate its effectiveness and efficiency. A further benefit of the module partition is the improvement of the mechanical structural properties.

Parallel wave sound analysis based on hierarchical domain decomposition method

Parallel wave sound analysis based on hierarchical domain decomposition method

Parallel full-wave electromagnetic field analysis based on hierarchical domain decomposition method

Parallel full-wave electromagnetic field analysis based on hierarchical domain decomposition method

Evaluation of complex systems: multicriteria approach

The quality of object functioning is evaluated by a set of properties that have a quantitative expression (quality criteria). The selection of object properties is carried out by the decision maker (DM). This selected object properties is a decomposition, resulting in a hierarchical structure. In decision theory the case when a multicriteria task is represented by a two-level hierarchical system is developed most thoroughly in great detail. Here, the problem of evaluation (composition of criteria) of a simple object is usually solved using the mechanism of a single scalar convolution of a vector criterion. The numerical value of the convolution is an assessment of the quality of the functioning of this object as a whole. But even with a three-level hierarchy, the object is considered as complex and its evaluation requires other approaches. It is shown that any problem of vector evaluation of an object can be represented by a hierarchical system of criteria obtained as a result of decomposition of the objectʼs properties. With hierarchical decomposition of object properties, the number of levels depends on the required depth of decomposition. Usually they brought to such properties that have a quantitative expression (called criteria). The difficulty lies in the fact that for each initial property, the depth of decomposition can be different. At the lower level of the hierarchy, the object (alternative) is evaluated by individual properties using the initial vector of criteria, and at the upper level, the object as a whole is evaluated by means of the composition mechanism. The central problem here is the problem of composing criteria by hierarchy levels, which is solved by the method of nested scalar convolutions. The necessary and sufficient conditions for the vector estimation of an object are considered.

Hierarchical causal variance decomposition for institution and provider comparisons in healthcare

Disease-specific quality indicators are used to compare institutions and health care providers in terms of processes or outcomes relevant to treatment of a particular condition. In the context of surgical cancer treatments, the performance variations can be due to hospital and/or surgeon level differences, creating a hierarchical clustering. We consider how the observed variation in care received at patient level can be decomposed into that causally explained by the hospital performance, surgeon performance within hospital, patient case-mix, and unexplained (residual) variation. For this purpose, we derive a four-way variance decomposition, with particular attention to the causal interpretation of the components. For estimation, we use inputs from a mixed-effect model with nested random hospital/surgeon-specific effects, and a multinomial logistic model for the hospital/surgeon-specific patient populations. We investigate the performance of our methods in a simulation study and demonstrate them through analysis of administrative data on kidney cancer care in Ontario.

Canonisation and Definability for Graphs of Bounded Rank Width

We prove that the combinatorial Weisfeiler-Leman algorithm of dimension (3 k +4) is a complete isomorphism test for the class of all graphs of rank width at most k . Rank width is a graph invariant that, similarly to tree width, measures the width of a certain style of hierarchical decomposition of graphs; it is equivalent to clique width. It was known that isomorphism of graphs of rank width k is decidable in polynomial time (Grohe and Schweitzer, FOCS 2015), but the best previously known algorithm has a running time n f(k) for a non-elementary function f . Our result yields an isomorphism test for graphs of rank width k running in time n O(k) . Another consequence of our result is the first polynomial-time canonisation algorithm for graphs of bounded rank width. Our second main result is that fixed-point logic with counting captures polynomial time on all graph classes of bounded rank width.

Deeply Explain CNN Via Hierarchical Decomposition

In computer vision, some attribution methods for explaining CNNs attempt to study how the intermediate features affect network prediction. However, they usually ignore the feature hierarchies among the intermediate features. This paper introduces a hierarchical decomposition framework to explain CNN’s decision-making process in a top-down manner. Specifically, we propose a gradient-based activation propagation (gAP) module that can decompose any intermediate CNN decision to its lower layers and find the supporting features. Then we utilize the gAP module to iteratively decompose the network decision to the supporting evidence from different CNN layers. The proposed framework can generate a deep hierarchy of strongly associated supporting evidence for the network decision, which provides insight into the decision-making process. Moreover, gAP is effort-free for understanding CNN-based models without network architecture modification and extra training processes. Experiments show the effectiveness of the proposed method. The data and source code will be publicly available at https://mmcheng.net/hdecomp/ .

Feature Extraction Based on Hierarchical Improved Envelope Spectrum Entropy for Rolling Bearing Fault Diagnosis

Bearing is the key part of mechanical equipment, which can support the rotating machinery running. It is crucial to diagnose bearing fault in time to ensure mechanical equipment working well. Effective feature extraction is an essential step in bearing fault diagnosis. However, the bearing vibration signals collected from the equipment usually contain interference, such as heavy noise. It is difficult to extract effective feature from bearing vibration signals due to the interference. To overcome this issue, a new feature extraction method for rolling bearing faults diagnosis is proposed based on hierarchical improved envelope spectrum entropy (HIESE). First, hierarchical decomposition is used to divide the bearing vibration signal into several hierarchical components. Second, the original feature set is obtained by calculating the improved envelope spectrum entropy (IESE) of each hierarchical component. Then, joint approximate diagonalization of eigenmatrices (JADE) is introduced to fuse the original features into a new one. Finally, support vector machines (SVM) is taken to identify the bearing status. Two cases are used to test the proposed method. In case 1, the bearing vibration signals composed of different fault degrees and different fault types under two speeds are used to verify the proposed method. The recognition rate achieves 100%. In case 2, the bearing vibration signals, which consist of different fault types under three operation conditions, are analyzed. The recognition rate also achieves 100% in this case. Experimental results illustrated that the proposed method has good performance in feature extraction, which can provide a new method for feature extraction of rolling bearing.

Integration of forestry management tasks at the forestry level

The article is devoted to the problems of forestry management. Integration of control tasks is proposed on the basis of a single database containing geometric, cartographic and taxation information. The integrated information system is focused on meeting the needs of forestry management at the forestry level. The system must have a user interface with tools for using forest resources. The functional model of the integrated system project is represented by diagrams of different levels and made it possible to identify the main approaches to creating the system. Functional modeling also showed the need to develop and include editors of forest plot boundaries into the proposed integrated system with the ability to visualize them; a layout of the proposed integrated system. To edit the boundary vertices of forest taxation units, a hierarchical decomposition has been proposed, which consists of sequential fragmentation of large areas into smaller areas.

A Strategy-Oriented Bayesian Soft Actor-Critic Model

Adopting reasonable strategies is challenging but crucial for an intelligent agent with limited resources working in hazardous, unstructured, and dynamic environments to improve the system's utility, decrease the overall cost, and increase mission success probability. This paper proposes a novel hierarchical strategy decomposition approach based on the Bayesian chain rule to separate an intricate policy into several simple sub-policies and organize their relationships as Bayesian strategy networks (BSN). We integrate this approach into the state-of-the-art DRL method – soft actor-critic (SAC) and build the corresponding Bayesian soft actor-critic (BSAC) model by organizing several sub-policies as a joint policy. We compare the proposed BSAC method with the SAC and other state-of-the-art approaches such as TD3, DDPG, and PPO on the standard continuous control benchmarks – Hopper-v2, Walker2d-v2, and Humanoid-v2 – in MuJoCo with the OpenAI Gym environment. The results demonstrate that the promising potential of the BSAC method significantly improves training efficiency.

STD-NET: Search of Image Steganalytic Deep-Learning Architecture Via Hierarchical Tensor Decomposition

Steganalysis aims to reveal covert communication established via steganography. In the arm race with steganography, steganalysis has evolved from the old-style hand-crafted features set to deep-learning architectures. However, recent studies show that the majority of existing deep steganalysis models have a large amount of redundancy, which leads to a huge waste of storage and computing resources. The existing model compression method cannot flexibly compress the convolutional layer in residual shortcut block so that a satisfactory shrinking rate cannot be obtained. In this paper, we propose STD-NET, an unsupervised deep-learning architecture search approach via hierarchical tensor decomposition for image steganalysis. Our proposed strategy will not be restricted by various residual connections, since this strategy does not change the number of input and output channels of the convolution block. We propose a normalized distortion threshold to evaluate the sensitivity of each involved convolutional layer of the base model to guide STD-NET to compress target network in an efficient and unsupervised approach, and obtain two network structures of different shapes with low computation cost and similar performance compared with the original one. Extensive experiments have confirmed that, on one hand, our model can achieve comparable or even better detection performance in various steganalytic scenarios due to the great adaptivity of the obtained network architecture. On the other hand, the experimental results also demonstrate that our proposed strategy is more efficient and can remove more redundancy compared with previous steganalytic network compression methods.

An approach to the selection of design parameters based on conflict with constraints

The Axiomatic Design (AD) supports decision-making based on the axioms of independence and information. In the engineering design activity, the designer must make early decisions that impact the success of the project. These decisions are of paramount importance in the design of one-off products. In these products, one must make decisions in the early design phases, with scarce data about the product that will undeniably affect the solution. Applying the information axiom requires data from the alternative solutions under consideration that for one-off products are most often non-existent in the conceptual phase of the design.Additionally, the design process is highly conditioned by existing constraints, which significantly influence the outcome. This work suggests selecting design parameters (DP) for one-off products from different sets of proposals that comply with the independence of functional requirements (FR). AD design follows a hierarchical decomposition process from a conceptual to a detailed design. The proposed method allows selecting the DPs among the set of alternatives by quantifying the degree of conflict each solution presents to the constraints. It shows the first step to define the DPs, at each level of decomposition of the AD process. It is the first step to establish a zig in AD or, otherwise, a way to find a solution in the conceptual phase. It applies to one-off products – any product with one chance to be right. The paper applies this method to the DP selection of a manipulation device.

A Hybrid Fault Diagnosis Method Using Translation Invariant Wavelet Denoising, Hierarchical Entropy, and Support Vector Machine with PSO Algorithm

The rolling element bearing is used in various machinery and produces vibration due to imperfections, surface irregularities during manufacture, damaged bearings, and inaccuracies in the allied element. Also, the rolling element bearing vibration generally shows non-linear dynamic characteristics and is masked with heavy background noise. This noble investigation advances a hybrid technique for removing background noise from the vibration signal and detecting bearing defects. Translation invariant wavelet denoising is the initial stage in this hybrid method for noise removal from the signal. The second phase uses Hierarchical Entropy (HE) for defect feature frequency extraction. Hierarchical entropy at scale four and SampEns of eight hierarchical decomposition nodes was utilized to determine the defect feature vector. In particular, low-frequency components are investigated through multi-scale entropy (MSE), but hierarchical entropy (HE) incorporates low-frequency and high-frequency components and can extract more defective information. Implemented a multi-class support vector machine (SVM) for extracting Hierarchical entropy as feature vectors. These feature vectors are trained by utilizing particle swarm optimization (PSO). To accomplish a prediction model, examine the optimal SVM parameters and then various bearing conditions with the variation of type, size, speed, and load severity identified by SVM. The investigation results show that hierarchical entropy can adequately and more precisely express the features of bearing vibration signals. It is beyond MSE, and the proposed Nobel hybrid Translation invariant wavelet denoising and Hierarchical entropy-based method will effectively remove the noisy background signal. Also, it distinguishes different bearings successfully, indicates the bearing conditions correctly, and is more prominent than those found on MSE.

An Improved High-Resistance Fault Detection Method in DC Microgrid Based on Orthogonal Wavelet Decomposition

High-resistance faults in direct current (DC) microgrids are small and thus difficult to detect. Such faults may be “invisible” in that grid operation continues for a considerable time, which damages the grid. It is essential to detect and remove high-resistance faults; we present a detection method herein. First, the transient DC current during the fault is subjected to hierarchical wavelet decomposition to identify high-resistance faults accurately and sensitively; the wavelet coefficients are detected using the singular value decomposition (SVD) method. The SVD valve can denoise the dc microgrid fault current, which eliminates the influence of converter switching frequency and background noise effectively. Power system computer-aided design (PSCAD)/electromagnetic transients including direct current (EMTDC)-based simulations showed that our method successfully identified high-resistance faults.

Navigation of Multiple Disk-Shaped Robots with Independent Goals within Obstacle-Cluttered Environments

In this work, we propose a hybrid control scheme to address the navigation problem for a team of disk-shaped robotic platforms operating within an obstacle-cluttered planar workspace. Given an initial and a desired configuration of the system, we devise a hierarchical cell decomposition methodology which is able to determine which regions of the configuration space need to be further subdivided at each iteration, thus avoiding redundant cell expansions. Furthermore, given a sequence of free configuration space cells with an arbitrary connectedness and shape, we employ harmonic transformations and harmonic potential fields to accomplish safe transitions between adjacent cells, thus ensuring almost-global convergence to the desired configuration. Finally, we present the comparative simulation results that demonstrate the efficacy of the proposed control scheme and its superiority in terms of complexity while yielding a satisfactory performance without incorporating optimization in the selection of the paths.

A new non-invasive tagging method for leopard coral grouper (Plectropomus leopardus) using deep convolutional neural networks with PDE-based image decomposition

External tagging methods can aid in the research of leopard coral grouper (Plectropomus leopardus) in terms of its spatio-temporal behavior at population and individual scales. However, due to the strong exclusion ability and the damage to the body wall of P. leopardus, the retention rate of traditional invasive tagging methods is low. To develop a non-invasive identification method for P. leopardus, we adopted a multiscale image processing method based on matched filters with Gaussian kernels and partial differential equation (PDE) multiscale hierarchical decomposition with the deep convolutional neural network (CNN) models VGG19 and ResNet50 to extract shape and texture image features of individuals. Then based on image features, we used three classifiers Random forest (RF), support vector machine (SVM), and multilayer perceptron (MLP)) for individual recognition on sequential images of P. leopardus captured for 50 days. The PDE, ResNet50 and MLP combination obtained a maximum accuracy of 0.985 ± 0.045 on the test set. For individual temporal tracking recognition, feature extraction and model training were performed using images taken in 1-20 days. The classifier could achieve an accuracy of 0.960 ± 0.049 on the test set consisting of images collected in the periods of 20-50 days. The results show that CNNs with the PDE decomposition can effectively and accurately identify P. leopardus.