Exploring cell tower data dumps for supervised learning-based point-of-interest prediction (industrial paper)

Exploring massive mobile data for location-based services becomes one of the key challenges in mobile data mining. In this paper, we investigate a problem of finding a correlation between the collective behavior of mobile users and the distribution of points of interest (POIs) in a city. Specifically, we use large-scale cell tower data dumps collected from cell towers and POIs extracted from a popular social network service, Weibo. Our objective is to make use of the data from these two different types of sources to build a model for predicting the POI densities of different regions in the covered area. An application domain that may benefit from our research is a business recommendation application, where a prediction result can be used as a recommendation for opening a new store/branch. The crux of our contribution is the method of representing the collective behavior of mobile users as a histogram of connection counts over a period of time in each region. This representation ultimately enables us to apply a supervised learning algorithm to our problem in order to train a POI prediction model using the POI data set as the ground truth. We studied 12 state-of-the-art classification and regression algorithms; experimental results demonstrate the feasibility and effectiveness of the proposed method.     

GM2NAS: multitask multiview graph neural architecture search

Knowledge and Information Systems - Graph neural network-based multitask learning models on multiview graphs have achieved acceptable results in different real-world applications. However,...     

A weighted ADI scheme with variable time steps for diffusion-wave equations

Calcolo - We present a local projection stabilization for the virtual element discretization of the nonlinear convection–diffusion–reaction equation. We consider a...     

A Self-Assembled Vertical-Gradient and Well-Dispersed MXene Structure for Flexible Large-Area Perovskite Modules

Advancing hole transport layers (HTL) to realize large-area, flexible, and high-performance perovskite solar cells (PSCs) is one of the most challenging issues for its commercialization. Here, a self-assembled gradient Ti₃C₂T_x MXene incorporated PEDOT:PSS HTL is demonstrated to achieve high-performance large-area PSCs by establishing half-caramelization-based glucose-induced MXene redistribution. Through this process, the Ti₃C₂T_x MXene nanosheets are spontaneously dispersed and redistributed at the top region of HTL to form the unique gradient distribution structure composed of MXene:Glucose:PEDOT:PSS (MG-PEDOT). These results show that the MG-PEDOT HTL not only offers favorable energy level alignment and efficient charge extraction, but also improves the film quality of perovskite layer featuring enlarged grain size, lower trap density, and longer carrier lifetime. Consequently, the power conversion efficiency (PCE) of the flexible device based on MG-PEDOT HTL is increased by 36% compared to that of pristine PEDOT:PSS HTL. Meanwhile, the flexible perovskite solar minimodule (15 cm² area) using MG-PEDOT HTL achieve a PCE of 17.06%. The encapsulated modules show remarkable long-term storage stability at 85 °C in ambient air (≈90% efficiency retention after 1200 h) and enhanced operational lifetime (≈90% efficiency retention after 200 h). This new approach shows a promising future of the self-assembled HTLs for developing optoelectronic devices.     

Pre-Activation of CO2 at Cobalt Phthalocyanine-Mg(OH)2 Interface for Enhanced Turnover Rate

Cobalt phthalocyanine (CoPc) anchored on heterogeneous scaffold has drawn great attention as promising electrocatalyst for carbon dioxide reduction reaction (CO₂RR), but the molecule/substrate interaction is still pending for clarification and optimization to maximize the reaction kinetics. Herein, a CO₂RR catalyst is fabricated by affixing CoPc onto the Mg(OH)₂ substrate primed with conductive carbon, demonstrating an ultra-low overpotential of 0.31 ± 0.03 V at 100 mA cm⁻² and high faradaic efficiency of >95% at a wide current density range for CO production, as well as a heavy-duty operation at 100 mA cm⁻² for more than 50 h in a membrane electrode assembly. Mechanistic investigations employing in situ Raman and attenuated total reflection surface-enhanced infrared absorption spectroscopy unravel that Mg(OH)₂ plays a pivotal role to enhance the CO₂RR kinetics by facilitating the first-step electron transfer to form anionic *CO₂⁻ intermediates. DFT calculations further elucidate that introducing Lewis acid sites help to polarize CO₂ molecules absorbed at the metal centers of CoPc and consequently lower the activation barrier. This work signifies the tailoring of catalyst-support interface at molecular level for enhancing the turnover rate of CO₂RR.     

Phase-Separated Polyzwitterionic Hydrogels with Tunable Sponge-Like Structures for Stable Solar Steam Generation

Solar steam generation (SSG) through hydrogel-based evaporators has shown great promise for freshwater production. However, developing hydrogel-based evaporators with stable SSG performance in high-salinity brines remains challenging. Herein, phase-separated polyzwitterionic hydrogel-based evaporators are presented with sponge-like structures comprising interconnected pores for stable SSG performance, which are fabricated by photopolymerization of sulfobetaine methacrylate (SBMA) in water-dimethyl sulfoxide (DMSO) mixed solvents. It is shown that driven by competitive adsorption, the structures of the resulting poly(sulfobetaine methacrylate) (PSBMA) hydrogels can be readily tuned by the volume ratio of DMSO to achieve phase separation. The optimized phase-separated PSBMA hydrogels, combining the unique anti-polyelectrolyte effects of polyzwitterionic hydrogels, demonstrate a rapid water transport capability in brines. After introducing photothermal polypyrrole particles on the surface of the phase-separated PSBMA hydrogel evaporators, a stable water evaporation rate of ≈2.024 kg m⁻² h⁻¹ and high solar-to-vapor efficiency of ≈97.5% in a 3.5 wt.% brine are obtained under simulated solar light irradiation (1.0 kW m⁻²). Surprisingly, the evaporation rates remain stable even under high-intensity solar irradiation (2.0 kW m⁻²). It is anticipated that the polyzwitterionic hydrogel evaporators with sponge-like porous structures will contribute to developing SSG technology for high-salinity seawater applications.     

多层自适应聚合的自监督小样本图像分类

目的 在图像分类领域,小样本学习旨在利用从大规模数据集中训练到的知识来处理仅包含少量有标记训练样本的下游分类任务。通常情况下,下游任务只涉及新类样本,由于元训练阶段会构造大量任务随机抽取训练集中不同类别的样本且训练集与测试集类别间存在领域间隙,因此模型训练周期长且可能对训练集过拟合,以致元知识无法迁移到测试集,进而导致模型泛化性差。针对以上问题,提出一种多层自适应聚合的自监督小样本图像分类模型。方法 首先使用分组卷积对残差块进行改进,减少神经网络参数量,降低训练难度,缩短训练时间;然后采用多层自适应聚合的方法改进骨干网络,对网络各层语义信息加以提炼聚合,自适应分配各层权重,将聚合后的特征图作为后续分类的依据;最后加入自监督对比学习结合有监督学习挖掘样本自身潜在的信息,从而提升样本特征表达能力。结果 在mini-ImageNet数据集和CUB(Caltech-UCSD birds-200-2011)数据集上与当前主流模型进行分类效果对比实验,与baseline相比,所提模型的准确率在mini-ImageNet数据集的5-way 1-shot与5-way 5-shot实验上分别提升了6.3...     

基于平面电容的油水多界面检测方法研究

含水量是衡量原油质量的重要指标,在原油的生产和储运过程中油水混合液分界面不断变化,因此全过程都要用到高精度传感器对其进行界面检测。基于电容边缘效应设计了一种侵入式平面电容传感器,其主要结构由基板和平面电极阵列两部分组成。运用有限元软件建立8电极阵列传感器模型,对不同电极工作时的电场分布进行研究,分析了平面电容传感器的检测灵敏度和成像精度。并且,研究了电极宽度、长度和相邻间距对传感器灵敏场分布的影响。对介电分布进行图像重建,使设计的平面电容阵列传感器可以检测3个分界面的高度,且经过尺寸参数优化,提高了传感器的成像精度。实验证明,运用平面电容阵列检测油水界面的方法具有可行性和有效性。