Recently, there has been a growing interest in automating the process of neural architecture design, and the Differentiable Architecture Search (DARTS) method makes the process available within a few GPU days. In particular, a hypernetwork called one-shot model is introduced, over which the architecture can be searched continuously with gradient descent. However, the performance of DARTS is often observed to collapse when the number of search epochs becomes large. Meanwhile, lots of "skip-connects" are found in the selected architectures. In this paper, we claim that the cause of the collapse is that there exist cooperation and competition in the bi-level optimization in DARTS, where the architecture parameters and model weights are updated alternatively. Therefore, we propose a simple and effective algorithm, named "DARTS+", to avoid the collapse and improve the original DARTS, by "early stopping" the search procedure when meeting a certain criterion. We demonstrate that the proposed early stopping criterion is effective in avoiding the collapse issue. We also conduct experiments on benchmark datasets and show the effectiveness of our DARTS+ algorithm, where DARTS+ achieves 2.32% test error on CIFAR10, 14.87% on CIFAR100, and 23.7% on ImageNet. We further remark that the idea of "early stopping" is implicitly included in some existing DARTS variants by manually setting a small number of search epochs, while we give an explicit criterion for "early stopping".

No abstract

Abstract-Space shift keying (SSK) modulation and its extension, the generalized SSK (GSSK), present an attractive framework for the emerging large-scale MIMO systems in reducing hardware costs. In SSK, the maximum likelihood (ML) detector incurs considerable computational complexities. We propose a compressed sensing based detector, NCS, by formulating the SSK-type detection criterion as a convex optimization problem. The proposed NCS requires only O(ntNrNt) complexity, outperforming the O(NrN nt t ) complexity in the ML detector, at the cost of slight fidelity degradation. Simulations are conducted to substantiate the analytical derivation and the detection accuracy.

a b s t r a c tLandfill mining is an environmentally-friendly technology that combines the concepts of material recycling and sustainable waste management, and it has received a great deal of worldwide attention because of its significant environmental and economic potential in material recycling, energy recovery, land reclamation and pollution prevention. This work applied a cost-benefit analysis model for assessing the economic feasibility, which is important for promoting landfill mining. The model includes eight indicators of costs and nine indicators of benefits. Four landfill mining scenarios were designed and analyzed based on field data. The economic feasibility of landfill mining was then evaluated by the indicator of net present value (NPV). According to our case study of a typical old landfill mining project in China (Yingchun landfill), rental of excavation and hauling equipment, waste processing and material transportation were the top three costs of landfill mining, accounting for 88.2% of the total cost, and the average cost per unit of stored waste was 12.7 USD ton À1 . The top three benefits of landfill mining were electricity generation by incineration, land reclamation and recycling soil-like materials. The NPV analysis of the four different scenarios indicated that the Yingchun landfill mining project could obtain a net positive benefit varying from 1.92 million USD to 16.63 million USD. However, the NPV was sensitive to the mode of land reuse, the availability of energy recovery facilities and the possibility of obtaining financial support by avoiding post-closure care.

No abstract

Objective Guidelines from different areas on the use of non-invasive ventilation in COVID-19 have generally been inconsistent. The goals were to appraise the quality and availability of guidelines stated and whether non-invasive ventilation in the early stage of the pandemic is of importance. Design and Method Databases including PubMed, Web of Science, Cochrane Library, and websites of international organizations and gray databases were searched up to June 23, 2020. We also hand-searched the reference lists of eligible papers. Results A total of 26 guidelines met the inclusion criteria. Regarding the appraisal by the Appraisal of Guidelines for Research and Evaluation (AGREE) II instrument, the guidelines’ methodological quality was low. Among six domains, Rigour. of Development and Editorial Independence were of the lowest quality. Given the lack of evidence from randomized clinical trials and the great differences between different regions, non-invasive ventilation’s recommendations generated a considerable debate at the early stage of COVID-19. Conclusions Improving the methodological quality of the guidelines should be a goal in future pandemics. Additionally, more well-designed randomized clinical trials are needed to solve the controversy on the impact of non-invasive ventilation.

No abstract

SUMMARYNowadays, hydropneumatic transformer (short for HP transformer) is widely used as the main part of the power system of air-powered vehicles to pump pressurized hydraulic oil from compressed air with a lower pressure. In this paper, the power system of a new kind of air-powered vehicle is built using HP transformer as the pneumatic pressure boosting structure and hydraulic motor as the power unit. To set a foundation for the optimization of the power system, the dynamic characteristics are analyzed by building the mathematic model and setting up the experimental station. What is more, the affection of several key structure parameters is researched. Through experimental and simulation study, it can be concluded that, firstly, the mathematical model is proved to be effective; secondly, increasing the area ratio of the pistons would improve the output power as well as reduce the system efficiency, and we choose 6 for a balanced choice; thirdly, the output power and the system efficiency will rise with a larger oil discharge orifice; lastly, the increasing pressure of the input air will raise the power and meanwhile decrease the efficiency. In this system, the efficiency will remain above 30%. This research can be referred to in the design of the power system of air-powered vehicle and the study on the optimization of the dynamic characteristics.