A decomposition structure learning algorithm in Bayesian network based on a two-stage combination method

Decomposition hybrid algorithms with the recursive framework which recursively decompose the structural task into structural subtasks to reduce computational complexity are employed to learn Bayesian network (BN) structure. Merging rules are commonly adopted as the combination method in the combination step. The direction determination rule of merging rules has problems in using the idea of keeping v-structures unchanged before and after combination to determine directions of edges in the whole structure. It breaks down in one case due to appearances of wrong v-structures, and is hard to operate in practice. Therefore, we adopt a novel approach for direction determination and propose a two-stage combination method. In the first-stage combination method, we determine nodes, links of edges by merging rules and adopt the idea of permutation and combination to determine directions of contradictory edges. In the second-stage combination method, we restrict edges between nodes that do not satisfy the decomposition property and their parent nodes by determining the target domain according to the decomposition property. Simulation experiments on four networks show that the proposed algorithm can obtain BN structure with higher accuracy compared with other algorithms. Finally, the proposed algorithm is applied to the thickening process of gold hydrometallurgy to solve the practical problem.

The incidence and mortality of COVID-19 related TB infection in Sub-Saharan Africa: A systematic review and meta-analysis

Background Coronavirus disease 2019 (COVID-19) is also associated with other co-morbidities among with previous and current pulmonary tuberculosis (PTB). PTB is a risk factor for COVID-19, both in terms of severity and mortality, regardless of human immunodeficiency virus (HIV) status. However, there is less information available on COVID-19 associated with PTB in point of view incidence and mortality rates in sub-Saharan Africa (SSA) as a high burden TB region. This systematic review served to provide data synthesis of available evidence on COVID-19/PTB incidence and case fatality rates, and mortality rate found in clinical and post-mortem COVID-19/PTB diagnostics in SSA. Methods We conducted a systematic electronic search in the PubMed, Medline, Google Scholar, Medrxix and COVID-19 Global literature on coronavirus disease databases for studies including COVID-19 associated with PTB in sub-Saharan Africa. The main outcomes were the proportion of people with COVID-19 associated to current /or previous PTB and the case fatality associated to COVID-19/PTB. The combination method was based on methodological similarities in the included random effect model studies using Prometa 3 software. We further undertook sensitivity analysis and meta-regression. Results From the 548 references extracted by the literature search, 25 studies were selected and included in the meta-analysis with a total of 191, 250 COVID-19 infected patients and 11, 452 COVID-19 deaths. The pooled COVID-19/PTB incidence was 2% [1%-3%] and mortality of 10% [4%-20%]. The pooled estimates for case fatality rate among COVID-19/PTB were 6% [3%-11%] for clinical PTB diagnostic and 26% [14%-48%] for post-mortem PTB diagnostic. Meta-regression model including the effect sizes and cumulative COVID-19 cases (P= 0.032), HIV prevalence (P= 0.041) and TB incidence (P= 0.002) to explained high heterogeneity between studies. Conclusion As a summary, the incidence of TB associated with COVID-19 and case fatality rates are higher in SSA. However, COVID-19 associated to TB may be underreported in the studies conducted in SSA as the post-mortem TB diagnostic was higher. Large-scale cohort studies that adequately clear tool on previous and/or current TB diagnostic tools are required to confirmed COVID-19/TB incidence and case fatality in SSA.

Multi-GNSS Combined Orbit and Clock Solutions at iGMAS

Global navigation services from the quad-constellation of GPS, GLONASS, BDS, and Galileo are now available. The international GNSS monitoring and assessment system (iGMAS) aims to evaluate the navigation performance of the current quad systems under a unified framework. In order to assess impact of orbit and clock errors on the positioning accuracy, the user range error (URE) is always taken as a metric by comparison with the precise products. Compared with the solutions from a single analysis center, the combined solutions derived from multiple analysis centers are characterized with robustness and reliability and preferred to be used as references to assess the performance of broadcast ephemerides. In this paper, the combination method of iGMAS orbit and clock products is described, and the performance of the combined solutions is evaluated by various means. There are different internal precisions of the combined orbit and clock for different constellations, which indicates that consistent weights should be assigned for individual constellations and analysis centers included in the combination. For BDS-3, Galileo, and GLONASS combined orbits of iGMAS, the root-mean-square error (RMSE) of 5 cm is achieved by satellite laser ranging (SLR) observations. Meanwhile, the SLR residuals are characterized with a linear pattern with respect to the position of the sun, which indicates that the solar radiation pressure (SRP) model adopted in precise orbit determination needs further improvement. The consistency between combined orbit and clock of quad-constellation is validated by precise point positioning (PPP), and the accuracies of simulated kinematic tests are 1.4, 1.2, and 2.9 cm for east, north, and up components, respectively.

Group Relationship Mining of College Students Based on Predictive Social Network

With the rapid development of the Internet, social networks have shown an unprecedented development trend among college students. Closer social activities among college students have led to the emergence of college students with new social characteristics. The traditional method of college students’ group classification can no longer meet the current demand. Therefore, this paper proposes a social network link prediction method-combination algorithm, which combines neighbor information and a random block. By mining the social networks of college students’ group relationships, the classification of college students’ groups can be realized. Firstly, on the basis of complex network theory, the essential relationship of college student groups under a complex network is analyzed. Secondly, a new combination algorithm is proposed by using the simplest linear combination method to combine the proximity link prediction based on neighbor information and the likelihood analysis link prediction based on a random block. Finally, the proposed combination algorithm is verified by using the social data of college students’ networks. Experimental results show that, compared with the traditional link prediction algorithm, the proposed combination algorithm can effectively dig out the group characteristics of social networks and improve the accuracy of college students’ association classification.

An Enhanced Multi-frequency Distorted Born Iterative Method for Ultrasound Tomography based on Fundamental Tone and Overtones

Conventional Distorted Born Iterative Method (DBIM) using single frequency has low resolution and is prone to creating images with high-contrast subjects. We propose a productive frequency combination method to better result in tomographic ultrasound imaging based on the multi-frequency technique. This study uses the natural mechanism of emitting oscillators' frequencies and uses these frequencies for imaging in iterations. We use a fundamental tone (i.e., the starting frequency f0) for the first iteration in DBIM, then consecutively use its overtones for the next ones. The digital simulation scenarios are tested with other multi-frequency approaches to prove our method's feasibility. We performed 57 different simulation scenarios on the use of multi-frequency information for the DBIM method. As a result, the proposed method for the smallest normalization error (RRE = 0.757). The proposed method's imaging time is not significantly longer than the way of using single frequency information.

BiFDANet: Unsupervised Bidirectional Domain Adaptation for Semantic Segmentation of Remote Sensing Images

When segmenting massive amounts of remote sensing images collected from different satellites or geographic locations (cities), the pre-trained deep learning models cannot always output satisfactory predictions. To deal with this issue, domain adaptation has been widely utilized to enhance the generalization abilities of the segmentation models. Most of the existing domain adaptation methods, which based on image-to-image translation, firstly transfer the source images to the pseudo-target images, adapt the classifier from the source domain to the target domain. However, these unidirectional methods suffer from the following two limitations: (1) they do not consider the inverse procedure and they cannot fully take advantage of the information from the other domain, which is also beneficial, as confirmed by our experiments; (2) these methods may fail in the cases where transferring the source images to the pseudo-target images is difficult. In this paper, in order to solve these problems, we propose a novel framework BiFDANet for unsupervised bidirectional domain adaptation in the semantic segmentation of remote sensing images. It optimizes the segmentation models in two opposite directions. In the source-to-target direction, BiFDANet learns to transfer the source images to the pseudo-target images and adapts the classifier to the target domain. In the opposite direction, BiFDANet transfers the target images to the pseudo-source images and optimizes the source classifier. At test stage, we make the best of the source classifier and the target classifier, which complement each other with a simple linear combination method, further improving the performance of our BiFDANet. Furthermore, we propose a new bidirectional semantic consistency loss for our BiFDANet to maintain the semantic consistency during the bidirectional image-to-image translation process. The experiments on two datasets including satellite images and aerial images demonstrate the superiority of our method against existing unidirectional methods.

A Precision Grinding Technology for Zirconium Alloy Tubes Based on Ultrasonic Wall Thickness Automatic Measurement System

To ensure the safety and long-term performance of nuclear fuel cladding zirconium tubes, the wall thickness uniformity of each cross section is strictly needed. Therefore, this paper presents comprehensive investigations on development of an automatic ultrasonic wall thickness measurement system for detecting the nuclear zirconium tubes. Based on the determination of overall scheme, optimization of key mechanical structures and design of control system, a series of performance testing analyses of this developed auto-measuring system were performed from aspects of measuring accuracy, measuring efficiency, stability and practicability. The results revealed that it could accurately obtain the wall thickness distribution and effectively guide the subsequent grinding process by automatically generated deviation correcting procedures to achieve the requirement of the wall thickness uniformity. The new combination method of ultrasonic auto-measuring and numerical control grinding proposed in this work would have a great significance for the development and application of nuclear reaction zirconium alloy container.