Globalized and localized matrix-pattern-oriented classification machine

As most of the key properties of a product are determined by its design concept, customer orientation must be ensured very early in the conceptual design phase. Hierarchical concept development can help to reduce the complexity of this strategically significant task. This paper examines the process of hierarchical concept design by means of the House of Concepts (HoC) approach, which was conceived specifically to support customer-oriented concept development. This approach, which builds upon Quality Function Deployment (QFD), is compact, illustrative, enhances comprehensibility and can be considered both a method and a guiding principle. Supporting the creation of solution concepts by analysing the dependencies between requirements, functions, structure and properties with matrix-based methods allows verifying the suitable product concepts. The HoC method promotes customer orientation, especially during conceptual design, and supports systematic concept development at different levels of abstraction. Thus, it enables step-by-step development of product concepts and helps to reduce the complexity of this important engineering task. The HoC approach can be used to ensure customer orientation in the development of product concepts, which leads to an increase in effectiveness in product development.

In this paper a new matrix-based method is proposed for. identification of radar pulse train. The proposed method, which is called NASH, can be used to identify the PRI for constant, staggered, and jittered signal. Previous matrix-based methods can only identify the first type of signal, i.e. constant PRI. The complexity of computation in NASH is approximately equal to 2.5N/sup 2/ which is lower than histogram-based methods. Also NASH can find missing pulses' position in the pulse train.

Matrix-pattern-oriented Ho–Kashyap classifier with regularization learning

We introduce and test a workflow that integrates petrophysical constraints and geological data in joint geophysical inversion in order to decrease the uncertainty of the results. This workflow uses statistical petrophysical properties to constrain the values retrieved by the geophysical inversion and geological prior information to decrease the effect of non-uniqueness. We integrate the different sources of information in a Bayesian framework, which takes into account the state of information. This permits us to quantify the posterior state of knowledge, the reduction of the uncertainty and to calculate the influence of prior information using quality indicators based on fixed-point statistics. This workflow was first tested using simple synthetic datasets to validate the method and assess the robustness of the workflow. As a result, the use of petrophysical constraints permits us to retrieve sharper boundaries, while prior structural information from geology permits to retrieve the geometry more accurately. Overall, the integration of the different constraints provides a model, with reduced uncertainties and better resolved parameters.

For computer vision based appraoches such as image classification (Krizhevsky et al. 2012), object detection (Ren et al. 2015) or pixel-wise weed classification (Milioto et al. 2017) machine learning is used for both feature extraction and processing (e.g. classification or regression). Historically, feature extraction (e.g. PCA; Ch. 12.1. in Bishop 2006) and processing were sequential and independent tasks (Wöber et al. 2013). Since the rise of convolutional neuronal networks (LeCun et al. 1989), a deep machine learning approach optimized for images, in 2012 (Krizhevsky et al. 2012), feature extraction for image analysis became an automated procedure. A convolutional neuronal net uses a deep architecture of artificial neurons (Goodfellow 2016) for both feature extraction and processing. Based on prior information such as image classes and supervised learning procedures, parameters of the neuronal nets are adjusted. This is known as the learning process. Simultaneously, geometric morphometrics (Tibihika et al. 2018, Cadrin and Friedland 1999) are used in biodiversity research for association analysis. Those approaches use deterministic two-dimensional locations on digital images (landmarks; Mitteroecker et al. 2013), where each position corresponds to biologically relevant regions of interest. Since this methodology is based on scientific results and compresses image content into deterministic landmarks, no uncertainty regarding those landmark positions is taken into account, which leads to information loss (Pearl 1988). Both, the reduction of this loss and novel knowledge detection, can be done using machine learning. Supervised learning methods (e.g., neuronal nets or support vector machines (Ch. 5 and 6. in Bishop 2006)) map data on prior information (e.g. labels). This increases the performance of classification or regression but affects the latent representation of the data itself. Unsupervised learning (e.g. latent variable models) uses assumptions concerning data structures to extract latent representations without prior information. Those representations does not have to be useful for data processing such as classification and due to that, the use of supervised and unsupervised machine learning and combinations of both, needs to be chosen carefully, according to the application and data. In this work, we discuss unsupervised learning algorithms in terms of explainability, performance and theoretical restrictions in context of known deep learning restrictions (Marcus 2018, Szegedy et al. 2014, Su et al. 2017). We analyse extracted features based on multiple image datasets and discuss shortcomings and performance for processing (e.g. reconstruction error or complexity measurement (Pincus 1997)) using the principal component analysis (Wöber et al. 2013), independent component analysis (Stone 2004), deep neuronal nets (auto encoders; Ch. 14 in Goodfellow 2016) and Gaussian process latent variable models (Titsias and Lawrence 2010, Lawrence 2005).

Multiclass skin lesion localization and classification using deep learning based features fusion and selection framework for smart healthcare

X-ray luminescence computed tomography (XLCT) is a novel molecular imaging modality that reconstructs the optical distribution of x-ray-excited phosphor particles with prior informational of anatomical CT image. The prior information improves the accuracy of image reconstruction. The system can also present anatomical CT image. The optical system based on a high sensitive charge coupled device (CCD) is perpendicular with a CT system. In the XLCT system, the xray was adopted to excite the phosphor of the sample and CCD camera was utilized to acquire luminescence emitted from the sample in 360 degrees projection free-space. In this study, the fluorescence diffuse optical tomography (FDOT)-like algorithm was used for image reconstruction, the structural prior information was incorporated in the reconstruction by adding a penalty term to the minimization function. The phosphor used in this study is Gd2O2S:Tb. For the simulation and experiments, the data was collected from 16 projections. The cylinder phantom was 40 mm in diameter and contains 8 mm diameter inclusion; the phosphor in the in vivo study was 5 mm in diameter at a depth of 3 mm. Both the errors were no more than 5%. Based on the results from these simulation and experimental studies, the novel XLCT method has demonstrated the feasibility for in vivo animal model studies.

The limited degrees of freedom (DOF) and movement form of the compliant vibration-assisted processing device are inherent constraints of the polishing technique. In this paper, a concept of a 3-DOF rotary vibration-assisted micropolishing system (3D RVMS) is proposed and demonstrated. The 3-DOF means the proposed vibration-assisted polishing device (VPD) is driven by three piezo-electric (PZT) actuators. Compared with the current vibration-assisted polishing technology which generates a trajectory with orthogonal actuators or parallel actuators, a novel 3-DOF piezoelectrically actuated VPD was designed to enable the workpiece to move along the rotational direction. Meanwhile, the proposed VPD can deliver large processing stoke in mrad scale and can be operated at a flexible non-resonant mode. A matrix-based compliance modeling method was adopted for calculating the compliance and amplification ratio of the VPD. Additionally, the dynamic and static properties of the developed VPD were verified using finite element analysis. Then, the VPD was manufactured and experimentally tested to investigate its practical performance. Finally, various polished surfaces which used silicon carbide (SiC) ceramic as workpiece material were uniformly generated by the high-performance 3D RVMS. Compared with a nonvibration polishing system, surface roughness was clearly improved by introducing rotary vibration-assisted processing. Both the analysis and experiments verified the effectiveness of the present 3D RVMS for micro-machining surfaces.

Structural regularized projection twin support vector machine for data classification

Preoperative diagnosis of gastric cancer and primary gastric lymphoma is challenging and has important clinical significance. Inspired by the inductive reasoning learning of the human brain, transfer learning can improve diagnosis performance of target task by utilizing the knowledge learned from the other domains (source domain). However, most studies focus on single-source transfer learning and may lead to model performance degradation when a large domain shift exists between the single-source domain and target domain. By simulating the multi-modal information learning and transfer mechanism of human brain, this study designed a multisource transfer learning feature extraction and classification framework, which can enhance the prediction performance of the target model by using multisource medical data (domain). First, this manuscript designs a feature extraction network that takes the maximum mean difference based on the Wasserstein distance as an adaptive measure of probability distribution and extracts the domain-specific invariant representations between source and target domain data. Then, aiming at the random generation of parameters bringing uncertainties to prediction accuracy and generalization ability of extreme learning machine network, the 1-norm regularization is used to implement sparse constraints of the output weight matrix and improve the robustness of the model. Finally, some experiments are carried out on the data of two medical centers. The experimental results show that the area under curves (AUCs) of the method are 0.958 and 0.929 in the two validation cohorts, respectively. The method in this manuscript can provide doctors with a better diagnostic reference, which has certain practical significance.

In this study, the previous work, present status, benefits, and limitations of the state-of-the-art technologies used in smart spraying technologies in precision weed management are reviewed. A total of 116 articles were identified from Google Scholar and Scopus to study the research work in the field of smart sprayers and precision weed management. The articles were examined based on the relevance, research focus, novelties, measured parameters, used technologies, and field of applications. Smart sprayers based on machine vision (MV) and artificial intelligence (AI) are keys to improving crop productivity and meeting the food demands of the future by reducing the yield losses due to weeds and working towards a sustainable future in agriculture. Many papers published in recent years have focused more on the machine vision, weed detection, and classification aspects of the weeding robot. Very few studies have attempted to discuss the components of a smart weeding machine, non-chemical-based weeders, the components of spraying systems, their controls, underlying fluid mechanics, and the field trials of these weeding robots. This article reviews conventional weeding techniques, machine-vision-based weeding robots, and spraying systems proposed or constructed in the last twenty-five years. Key technologies such as non-chemical-based weeding machines, image preprocessing, feature extraction, and weed detection based on machine learning (ML) and deep learning (DL) for smart sprayers are discussed. The fundamental components of a smart spraying system are also discussed, and previous works are compared to highlight the key components, the spraying accuracy, and the major advantages and disadvantages. The fluid mechanics of the spraying system and its associated controls involved are also presented. There are still many bottlenecks in weed detection systems and smart spraying systems. The results of the systematic review provide an understanding of the progress made in the field of robotic weed detection, herbicide and non-herbicide-based weed management, the use of machine vision, and the limitations of the current spraying systems.

Noise covariance estimation via autocovariance least-squares with deadbeat filters

We describe a general approach to several RNA sequence analysis problems using probabilistic models that flexibly describe the secondary structure and primary sequence consensus of an RNA sequence family. We call these models 'covariance models'. A covariance model of tRNA sequences is an extremely sensitive and discriminative tool for searching for additional tRNAs and tRNA-related sequences in sequence databases. A model can be built automatically from an existing sequence alignment. We also describe an algorithm for learning a model and hence a consensus secondary structure from initially unaligned example sequences and no prior structural information. Models trained on unaligned tRNA examples correctly predict tRNA secondary structure and produce high-quality multiple alignments. The approach may be applied to any family of small RNA sequences.

Given a source image and a driving video depicting the same object type, the motion transfer task aims to generate a video by learning the motion from the driving video while preserving the appearance from the source image. In this paper, we propose a novel structure-aware motion modeling approach, the deformable anchor model (DAM), which can automatically discover the motion structure of arbitrary objects without leveraging their prior structure information. Specifically, inspired by the known deformable part model (DPM), our DAM introduces two types of anchors or key-points: i) a number of motion anchors that capture both appearance and motion information from the source image and driving video; ii) a latent root anchor, which is linked to the motion anchors to facilitate better learning of the representations of the object structure information. More-over, DAM can be further extended to a hierarchical version through the introduction of additional latent anchors to model more complicated structures. By regularizing motion anchors with latent anchor(s), DAM enforces the corre-spondences between them to ensure the structural information is well captured and preserved. Moreover, DAM can be learned effectively in an unsupervised manner. We validate our proposed DAM for motion transfer on different bench-mark datasets. Extensive experiments clearly demonstrate that DAM achieves superior performance relative to existing state-of-the-art methods.

Dominant Negative Mutants of Guanylyl Cyclase: Probes for Global Functions and Intramolecular Mechanisms