Search

Contemporarily, ultraviolet photodetectors have already been successfully applied in many advanced areas, including astronomy, disaster forecasting, fire alarm, marine oil pollution monitoring, biomedical and more. Nowadays, many researchers are focusing on developing the wide bandgap UV photodetectors which possess advantages in fast response speed, high sensitivity, self-power, etc. This paper will discuss principle, types, and component of ultraviolet photodetectors based on information retrieval and literature analysis. Specifically, vacuum UV detectors, solid-state UV detectors and wide bandgap UV photodetectors are demonstrated. According to the analysis, wide bandgap UV photodetectors have better performance in multiple properties but still be limited with several drawbacks, included material problems and design issues. Once the material problems and design issues could be overcome, they could replace the traditional UV photodetectors rapidly, which are able to benefit more relevant areas in both military and civilization. These results shed light on guiding further exploration in UV photodetectors current.

The hydrolysis-based post-polymerization modification method was introduced into the self-assembly process and a modification-induced self-assembly (MISA) technique was presented.

As an alternative to traditional riveting and welding materials, high-temperature-resistant adhesives, with their unique advantages, have been widely used in aviation, aerospace, and other fields. Among them, polyimide (PI) adhesives have been one of the most studied species both from basic and practical application aspects. However, in the main applications of solvent-type PI adhesives, pinholes or bubbles often exist in the cured PI adhesive layers due to the solvent volatilization and dehydration reaction, which directly affect the adhesive performance. To address this issue, electrospun PI nanofibrous membranes (NFMs) were employed as solvent-free or solvent-less adhesives for stainless steel in the current work. To enhance the adhesion of PI adhesives to the metal substrates, phenolphthalein groups and flexible ether bonds were introduced into the main chain of PIs via the monomers of 4,4′-oxydiphthalic anhydride (ODPA) and 3,3-bis[4-(4-aminophenoxy)phenyl] phthalide (BAPPT). At the same time, the methylethynyl group was used as the end-capping component, and the crosslinking reaction of the alkynyl group at high temperature further increased the adhesive strength of the PI adhesives. Three kinds of methylethynyl-terminated PI (METI) NFMs with the set molecular weights of 5000, 10,000, and 20,000 g/mol were first prepared via the one-step high-temperature polycondensation procedure. Then, the PI NFMs were fabricated via the standard electrospinning procedure from the soluble METI solutions. The afforded METI NFMs showed excellent melt-flowing behaviors at high temperature. Incorporation of the methylethynyl end-capping achieved a crosslinking reaction at 280−310 °C for the NFMs, which was about 70 °C lower than those of the phenylacetylene end-capping counterparts. Using the METI NFMs as adhesive, stainless steel adherends were successfully bonded, and the single-lap shear strength (LSS) was higher than 20.0 MPa at both room temperature (25 °C) and high temperature (200 °C).

Filter pruning is a significant feature selection technique to shrink the existing feature fusion schemes (especially on convolution calculation and model size), which helps to develop more efficient feature fusion models while maintaining state-of-the-art performance. In addition, it reduces the storage and computation requirements of deep neural networks (DNNs) and accelerates the inference process dramatically. Existing methods mainly rely on manual constraints such as normalization to select the filters. A typical pipeline comprises two stages: first pruning the original neural network and then fine-tuning the pruned model. However, choosing a manual criterion can be somehow tricky and stochastic. Moreover, directly regularizing and modifying filters in the pipeline suffer from being sensitive to the choice of hyperparameters, thus making the pruning procedure less robust. To address these challenges, we propose to handle the filter pruning issue through one stage: using an attention-based architecture that adaptively fuses the filter selection with filter learning in a unified network. Specifically, we present a pruning method named adding before pruning (ABP) to make the model focus on the filters of higher significance by training instead of man-made criteria such as norm, rank, etc. First, we add an auxiliary attention layer into the original model and set the significance scores in this layer to be binary. Furthermore, to propagate the gradients in the auxiliary attention layer, we design a specific gradient estimator and prove its effectiveness for convergence in the graph flow through mathematical derivation. In the end, to relieve the dependence on the complicated prior knowledge for designing the thresholding criterion, we simultaneously prune and train the filters to automatically eliminate network redundancy with recoverability. Extensive experimental results on the two typical image classification benchmarks, CIFAR-10 and ILSVRC-2012, illustrate that the proposed approach performs favorably against previous state-of-the-art filter pruning algorithms.

This paper proposes a topological mapping based failure recovery algorithm in multi-domain quantum key distribution networks. Simulation results show that the algorithm can achieve good performance in terms of success rate of service recovery under multi-concurrent failures.

Unsupervised Domain Adaptation (UDA) can tackle the challenge that convolutional neural network(CNN)-based approaches for semantic segmentation heavily rely on the pixel-level annotated data, which is labor-intensive. However, existing UDA approaches in this regard inevitably require the full access to source datasets to reduce the gap between the source and target domains during model adaptation, which are impractical in the real scenarios where the source datasets are private, and thus cannot be released along with the well-trained source models. To cope with this issue, we propose a source-free domain adaptation framework for semantic segmentation, namely SFDA, in which only a well-trained source model and an unlabeled target domain dataset are available for adaptation. SFDA not only enables to recover and preserve the source domain knowledge from the source model via knowledge transfer during model adaptation, but also distills valuable information from the target domain for self-supervised learning. The pixel- and patch-level optimization objectives tailored for semantic segmentation are seamlessly integrated in the framework. The extensive experimental results on numerous benchmark datasets highlight the effectiveness of our framework against the existing UDA approaches relying on source data., Comment: CVPR 2021, 10 pages

The Environmental Kuznets Curve was introduced by Grossman and Krueger in 1991, which implies a dynamic relationship between different indicators of environmental degradation and per capita income. In general, the curve suggests that a country’s economic growth initially leads to environmental degradation. After per capita income of the country reaches certain stages, the relationship reverses (economic growth leads to progressions of environmental quality). This curve has been a dominant modeling for economists investigating relationship between ambient pollution and social developments, and it seems guarantee future reduction in pollutant-emissions of developing countries. This study will focus on air pollution problems in developing countries and investigate whether the trend described by EKZ Curve holds constant with different environmental and social indicators like HDI, AQI and GDP.

An improved method of Fission Track technique combined with Thermal Ionization Mass Spectrometry (FT-TIMS) was established in order to determine isotope ratio of uranium-bearing particle. Working standard of uranium oxide particles with a defined diameter and isotopic composition were prepared and used to review the method. Results showed an excellent agreement with certified values. The developed method was used to analyze isotope ratio of single uranium-bearing particle in swipe samples successfully. The analysis results of uranium-bearing particles in swipe samples accorded with the operation history of the origin.

In semiconductor manufacturing, the monitoring system has been developed very excellently and can be used for comprehensively collecting the historical data of process information and quality characteristics of equipment. However, due to the high turnover rate of personnel and the great variance in manufacturing process, the previous control technique by using intuition and experience of engineers for manufacturing process parameter settings to achieve good product quality is no longer appropriate. Therefore, this research establishes a quality predictor for analyzing the relationship between manufacturing process parameter setting and final product quality in the plasma-enhanced chemical vapor deposition (PECVD) of semiconductor manufacturing by applying the back-propagation neural network (BPNN) algorithm and Taguchi method. The experimental data are categorized into 500 pieces of training data and 150 pieces of verifying data. The proposed analysis method for using in the PECVD process of semiconductor manufacturing is verified by comparing the predicted film thickness of SiO"2 and the predicted refractive index of silicon dioxide films with the measured data. According to the comparison result, the proposed model has an excellent prediction capability of final product quality and can be applied in process control for related manufacturing fields.

The capacity of data management and the query performance are two main metrics for today's cloud storage system. In this paper, we design a two-layer indexing scheme of distributed secondary index on switch-centric data center network (DCN) topologies. We take advantage of the desirable features of switch-centric topologies, such as stability, scalability, and fault tolerance, and successfully improve the query efficiency for cloud storage system. We choose B+-tree as the local layer index to locate data in each local host, while implement segment tree as the global layer index to manage a portion of metadata published from local hosts. We also optimize the query processing protocols to reduce false positives and network cost. In addition, we propose a top-down index selection method for maintenance. We then analyze the efficiency of query processing theoretically and calculate the expected number of routing cost for each query precisely. Finally we validate the efficiency of our design by experiments and comparison with a previous work.

The indices in the distributed storage systems manage the stored data and support diverse queries efficiently. Secondary index, the index built on the attributes other than the primary key, facilitates a variety of queries for different purposes. In this paper, we propose U2-Tree, a universal distributed secondary indexing scheme built on cloud storage systems with tree-like topologies. U2-Tree is composed of two layers, the global index and the local index. We build the local index according to the local data features, and then assign the potential indexing range of the global index for each host. After that, we use several techniques to publish the meta-data about local index to the global index host. The global index is then constructed based on the collected intervals. We take advantage of the topological properties of tree-like topologies, introduce and compare the detailed optimization techniques in the construction of two-layer indexing scheme. Furthermore, we discuss the index updating, index tuning, and the fault tolerance of U2-Tree. Finally, we propose numerical experiments to evaluate the performance of U2-Tree. The universal indexing scheme provides a general approach for secondary index on data centers with tree-like topologies. Moreover, many techniques and conclusions can be applied to other DCN topologies.