Explore the vast range of titles available.

Three air target intention recognition methods based on deep learning are proposed to realize the function of recognizing air target intention based on real-time situation information to resolve the problem that pilots cannot effectively observe and analyze observation within a short period of time in complex air battlefield environments and traditional air target intention recognition algorithms have shortcomings such as complex feature filtering and reliance on expert experience. The methods use expert experience to simulate combat in the air on the simulation platform we designed, obtain and filter key posture information of aerial targets and corresponding expert script intention labels during combat, and sends them to a designed full connect network, a convolutional neural network and a recurrent neural network for training. The training results show that all three networks can achieve air target intention recognition, and the recurrent neural network based model can achieve the intention recognition with an accuracy of 80%. Compared with traditional methods, such as D-S inference, the proposed method is more general and robust. Finally, the feasibility and effectiveness of the method we proposed is verified by the simulation and experiments.

Breaking through cell membrane barriers is a crucial step for intracellular drug delivery in antitumor chemotherapy. Hereby, a magnetic nanorobot, capable of exerting mechanical agitation on cellular membrane to promote intracellular drug delivery, was developed. The main body of the nanorobots was composed of nano-scaled gold nanospikes that were deposited with Ni and Ti nanolayers for magnetic activation and biocompatibility, responsively. The nanorobots can be precisely navigated to target cancer cells under external magnetic field control. By virtue of the sharp nanospike structures, the magnetically powered rotation behavior of the nanorobots can impose mechanical agitation on the living cell membrane and thus improve the membrane permeability, leading to promoted transmembrane cargo delivery. Coarse-grained molecular dynamics simulation revealed that the mechanism of mechanical intervention regulated permeability of the bilayer lipid membrane, allowing for enhanced transmembrane diffusion of small cargo molecules. An in vitro study demonstrated that these nanorobots can markedly enhance the efficiency of drug entry into tumor cells, thus improving the effectiveness of tumor therapy under magnetic activation in vivo. This work paves a new way for overcoming cell membrane barriers for intracellular drug delivery by using a magnetic nanorobotic system, which is expected to promote further application of magnetically controlled nanorobot technology in the field of precision medicine.

The Protocol for the Analysis of Land Surface Models (PALS) Land Surface Model Benchmarking Evaluation Project (PLUMBER) illustrated the value of prescribing a priori performance targets in model intercomparisons. It showed that the performance of turbulent energy flux predictions from different land surface models, at a broad range of flux tower sites using common evaluation metrics, was on average worse than relatively simple empirical models. For sensible heat fluxes, all land surface models were outperformed by a linear regression against downward shortwave radiation. For latent heat flux, all land surface models were outperformed by a regression against downward shortwave radiation, surface air temperature, and relative humidity. These results are explored here in greater detail and possible causes are investigated. It is examined whether particular metrics or sites unduly influence the collated results, whether results change according to time-scale aggregation, and whether a lack of energy conservation in flux tower data gives the empirical models an unfair advantage in the intercomparison. It is demonstrated that energy conservation in the observational data is not responsible for these results. It is also shown that the partitioning between sensible and latent heat fluxes in LSMs, rather than the calculation of available energy, is the cause of the original findings. Finally, evidence is presented that suggests that the nature of this partitioning problem is likely shared among all contributing LSMs. While a single candidate explanation for why land surface models perform poorly relative to empirical benchmarks in PLUMBER could not be found, multiple possible explanations are excluded and guidance is provided on where future research should focus.

Due to the superiority of wire rope in strength, cost, ductility and other characteristics, wire rope has become an irreplaceable material in metallurgy, mining, oil and gas drilling, machinery, chemical industry, aerospace and other fields. Therefore, the wire rope flaw detection is very important work, and in the wire rope detection, jitter is very serious interference. In this article, we firstly discuss the quantitative relationship between displacement and direction of shaking and leakage magnetic field, and this study can also be used as a essential basis for subsequent quantitative studies of local flaws to remove the effect of shaking on leakage magnetic signal.

The Protocol for the Analysis of Land Surface Models (PALS) Land Surface Model Benchmarking Evaluation Project (PLUMBER) illustrated the value of prescribing a priori performance targets in model intercomparisons. It showed that the performance of turbulent energy flux predictions from different land surface models, at a broad range of flux tower sites using common evaluation metrics, was on average worse than relatively simple empirical models. For sensible heat fluxes, all land surface models were outperformed by a linear regression against downward shortwave radiation. For latent heat flux, all land surface models were outperformed by a regression against downward shortwave radiation, surface air temperature, and relative humidity. These results are explored here in greater detail and possible causes are investigated. It is examined whether particular metrics or sites unduly influence the collated results, whether results change according to timescale aggregation, and whether a lack of energy conservation in flux tower data gives the empirical models an unfair advantage in the intercomparison. It is demonstrated that energy conservation in the observational data is not responsible for these results. It is also shown that the partitioning between sensible and latent heat fluxes in LSMs, rather than the calculation of available energy, is the cause of the original findings. Finally, evidence is presented that suggests that the nature of this partitioning problem is likely shared among all contributing LSMs. While a single candidate explanation for why land surface models perform poorly relative to empirical benchmarks in PLUMBER could not be found, multiple possible explanations are excluded and guidance is provided on where future research should focus.

Semantic Embedding Model (SEM), a neural network-based Siamese architecture, is gaining momentum in information retrieval and natural language processing. In order to train SEM in a supervised fashion for Web search, the search engine query log is typically utilized to automatically formulate pairwise judgments as training data. Despite the growing application of semantic embeddings in the search engine industry, little work has been done on formulating effective pairwise judgments for training SEM. In this paper, we make the first in-depth investigation of a wide range of strategies for generating pairwise judgments for SEM. An interesting (perhaps surprising) discovery reveals that the conventional pairwise judgment formulation strategy wildly used in the field of pairwise Learning-to-Rank (LTR) is not necessarily effective for training SEM. Through a large-scale empirical study based on query logs and click-through activities from a major commercial search engine, we demonstrate the effective strategies for SEM and highlight the advantages of a hybrid heuristic (i.e., Clicked > Non-Clicked) in comparison to the atomic heuristics (e.g., Clicked > Skipped) in LTR. We conclude with best practices for training SEM and offer promising insights for future research.

The PALS Land sUrface Model Benchmarking Evaluation pRoject (PLUMBER) illustrated the value of prescribing a priori performance targets in model intercomparisons. It showed that the performance of turbulent energy flux predictions from different land surface models, at a broad range of flux tower sites using common evaluation metrics, was on average worse than relatively simple empirical models. For sensible heat fluxes, all land surface models were outperformed by a linear regression against downward shortwave radiation. For latent heat flux, all land surface models were outperformed by a regression against downward shortwave, surface air temperature and relative humidity. These results are explored here in greater detail and possible causes are investigated. We examine whether particular metrics or sites unduly influence the collated results, whether results change according to time-scale aggregation and whether a lack of energy conservation in fluxtower data gives the empirical models an unfair advantage in the intercomparison. We demonstrate that energy conservation in the observational data is not responsible for these results. We also show that the partitioning between sensible and latent heat fluxes in LSMs, rather than the calculation of available energy, is the cause of the original findings. Finally, we present evidence suggesting that the nature of this partitioning problem is likely shared among all contributing LSMs. While we do not find a single candidate explanation forwhy land surface models perform poorly relative to empirical benchmarks in PLUMBER, we do exclude multiple possible explanations and provide guidance on where future research should focus.

It is important for China to break the “low-end lock” of the manufacturing value chain worldwide by revealing how digital trade promotes and reallocates the export technology complexity of the manufacturing industry. Panel data for 30 provinces in China from 2011 to 2020 were employed to measure the digital trade development and export technology complexity of the manufacturing industry. Benchmark regression, intermediary effect regression, panel threshold and other models were used to test the promotion and reallocation of digital trade on the export technology complexity of the manufacturing industry. The findings are as follows: (1) Digital trade promotes the export technology complexity of the manufacturing industry, with significant regional heterogeneity (eastern, central and western regions), and the most obvious promotion in technology-intensive manufacturing. (2) Technological innovation and human capital play a reallocation role in the process of digital trade, affecting the technological complexity of manufacturing exports, with mediating effects of 14.19% and 8.61%, respectively. (3) Digital trade promotes and reallocates the export technology complexity of the manufacturing industry through industrial structure upgrading, and a nonlinear relationship was found. These results provide empirical support and a decision-making basis for digital trade in promoting the export technology complexity of the manufacturing industry. The development of digital trade should be encouraged; the differential development of digital trade in the eastern, central, and western regions should be boosted; importance should be attached to the intermediary incentive role of technological innovation and human capital; and the upgrading of the industrial structure should be promoted scientifically.

In this paper, we demonstrate the generation of high-performance entangled photon-pairs in different degrees of freedom from a single piece of fiber pigtailed periodically poled LiNbO3 (PPLN) waveguide. We utilize cascaded second-order nonlinear optical processes, i.e., second-harmonic generation (SHG) and spontaneous parametric downconversion (SPDC), to generate photon-pairs. Previously, the performance of the photon-pairs is contaminated by Raman noise photons. Here by fiber-integrating the PPLN waveguide with noise-rejecting filters, we obtain a coincidence-to-accidental ratio (CAR) higher than 52,600 with photon-pair generation and detection rate of 52.36 kHz and 3.51 kHz, respectively. Energy-time, frequency-bin, and time-bin entanglement is prepared by coherently superposing correlated two-photon states in these degrees of freedom, respectively. The energy-time entangled two-photon states achieve the maximum value of CHSH-Bell inequality of S = 2.71 ± 0.02 with two-photon interference visibility of 95.74 ± 0.86%. The frequency-bin entangled two-photon states achieve fidelity of 97.56 ± 1.79% with a spatial quantum beating visibility of 96.85 ± 2.46%. The time-bin entangled two-photon states achieve the maximum value of CHSH-Bell inequality of S = 2.60 ± 0.04 and quantum tomographic fidelity of 89.07 ± 4.35%. Our results provide a potential candidate for the quantum light source in quantum photonics.

Multi-label pedestrian attribute recognition in surveillance is inherently a challenging task due to poor imaging quality, large pose variations, and so on. In this paper, we improve its performance from the following two aspects: (1) We propose a cascaded Split-and-Aggregate Learning (SAL) to capture both the individuality and commonality for all attributes, with one at the feature map level and the other at the feature vector level. For the former, we split the features of each attribute by using a designed attribute-specific attention module (ASAM). For the later, the split features for each attribute are learned by using constrained losses. In both modules, the split features are aggregated by using several convolutional or fully connected layers. (2) We propose a Feature Recombination (FR) that conducts a random shuffle based on the split features over a batch of samples to synthesize more training samples, which spans the potential samples’ variability. To the end, we formulate a unified framework, named CAScaded Split-and-Aggregate Learning with Feature Recombination (CAS-SAL-FR), to learn the above modules jointly and concurrently. Experiments on five popular benchmarks, including RAP, PA-100K, PETA, Market-1501 and Duke attribute datasets, show the proposed CAS-SAL-FR achieves new state-of-the-art performance.