In the above article <xref ref-type="bibr" rid="ref1">1 , a difference in the definitions of our simulated reflectance (with respect to instantaneous TOA radiance) and the operational MERSI-II L1 ...reflectance (with respect to solar constant) causes errors in their direct comparisons.
Sanz and colleagues emphasized in their report that the denisty-scaling component for the silicone oil DC704 is state-point dependent within the reported statistical accuracy. After repeating the ...measurements at 218 K, results were obtained in agreement with other findings. The error in the original measurement orginated from insufficient equilibration at the lowest temperature due to extremelv Iong flow times in the narrow capacitor gap.
Deep neural networks with millions of parameters may suffer from poor generalization due to overfitting. To mitigate the issue, we propose a new regularization method that penalizes the predictive ...distribution between similar samples. In particular, we distill the predictive distribution between different samples of the same label during training. This results in regularizing the dark knowledge (i.e., the knowledge on wrong predictions) of a single network (i.e., a self-knowledge distillation) by forcing it to produce more meaningful and consistent predictions in a class-wise manner. Consequently, it mitigates overconfident predictions and reduces intra-class variations. Our experimental results on various image classification tasks demonstrate that the simple yet powerful method can significantly improve not only the generalization ability but also the calibration performance of modern convolutional neural networks.
We demonstrate laser-driven two-qubit and single-qubit logic gates with respective fidelities 99.9(1)% and 99.9934(3)%, significantly above the ≈99% minimum threshold level required for ...fault-tolerant quantum computation, using qubits stored in hyperfine ground states of calcium-43 ions held in a room-temperature trap. We study the speed-fidelity trade-off for the two-qubit gate, for gate times between 3.8 μs and 520 μs, and develop a theoretical error model which is consistent with the data and which allows us to identify the principal technical sources of infidelity.
Near-surface air temperature is an important indicator of climate change and extreme events. ERA5-Land reanalysis products feature finer spatial and temporal resolutions, and have been widely adopted ...in global climate-related research. However, the performance of ERA5-Land air temperature data in coastal urban agglomerations has received little attention. In this study, a comprehensive evaluation is conducted in the Guangdong−Hong Kong−Macau Greater Bay Area (GBA) using the observations of 1080 automatic weather stations in 2018 as reference. Generally, ERA5-Land underestimates temperature (an average bias of 0.90 °C), and performs better at low temperatures than at high temperatures. At the station level, it is observed that the correlation shows a strong positive linear relationship with the distance to the coastline in summer, and that the bias increases with increasing altitude throughout the year. With respect to different land cover types, data accuracy over urban and built-up lands is the lowest. The spatial pattern of ERA5-Land is generally consistent with that of stations but relatively poor in urban areas. In addition, ERA5-Land properly captures daily and monthly variations, as well as intraday temperature fluctuations. These conclusions provide a reference for the implementation of ERA5-Land in coastal urban agglomerations.
Display omitted
•ERA5-Land performs better at low temperatures than at high temperatures.•Distance to the coastline and altitude affect the performance of ERA5-Land.•Data accuracy on croplands is highest, followed by forests, urban and built-up lands.•ERA5-Land is capable of capturing intraday fluctuations and seasonal variations.
Abstract Measuring the flatness error of large precision workpieces quickly and accurately is a difficult problem. A new method for preprocessing flatness measurement data based on MSE (mean squared ...error) is proposed. A mathematical model of a new data preprocessing method was established, and the mathematical formula for model solving was derived in detail. The data were measured by digital level on the plane of the granite base with dimensions of 2340 m×1540 mm. The new method and SmartLevel (basic measurement system of the level computer) were used to calculate and process the data. The flatness errors after diagonal evaluation were 4.07 μm and 3.90 μm, respectively. The relative error of the two was 4.36%, which confirmed the reliability and accuracy of the new method. The data results show that this method can be effectively used for the engineering measurement of the flatness of large precision workpieces.
A serious error exists in the paper: Alharbi KAM, Riaz A, Sikandar S. An entropy model for Carreau nanofluid ciliary flow with electroosmosis and thermal radiations: a numerical study. ...Electrophoresis. 2024;45:1112–1129.
Our work focuses on tackling the challenging but natural visual recognition task of long-tailed data distribution (i.e., a few classes occupy most of the data, while most classes have rarely few ...samples). In the literature, class re-balancing strategies (e.g., re-weighting and re-sampling) are the prominent and effective methods proposed to alleviate the extreme imbalance for dealing with long-tailed problems. In this paper, we firstly discover that these re-balancing methods achieving satisfactory recognition accuracy owe to that they could significantly promote the classifier learning of deep networks. However, at the same time, they will unexpectedly damage the representative ability of the learned deep features to some extent. Therefore, we propose a unified Bilateral-Branch Network (BBN) to take care of both representation learning and classifier learning simultaneously, where each branch does perform its own duty separately. In particular, our BBN model is further equipped with a novel cumulative learning strategy, which is designed to first learn the universal patterns and then pay attention to the tail data gradually. Extensive experiments on four benchmark datasets, including the large-scale iNaturalist ones, justify that the proposed BBN can significantly outperform state-of-the-art methods. Furthermore, validation experiments can demonstrate both our preliminary discovery and effectiveness of tailored designs in BBN for long-tailed problems. Our method won the first place in the iNaturalist 2019 large scale species classification competition, and our code is open-source and available at https://github.com/Megvii-Nanjing/BBN.