Nanocellulose-Based Ink for Vertically 3D Printing Micro-Architectures with High-Resolution

Nanocellulose has become an important renewable component for composite inks, owing to its desirable physical properties, reinforcing capabilities, and tunable self-assembly behavior. However, it is difficult to improve the rheological performance of the nanocellulose-based composite to meet the requirement for 3D printing high resolution microarchitectures. Herein, a strategy is proposed that incorporation of amphiphilic molecular surfactant into nanocellulose gel can increase the molecular interaction via hydrophobic bonds and enhance the ink viscoelasticity. Following the design, a composite ink is formulated by adding xylan and Nonaethylene glycol monododecyl ether (C12E9) within nanocellulose gel. A new printing program is designed to achieve vertical writing of the composite ink and obtain free-standing micropillars and microhemispheres with high resolution in dozens of micrometers. The microhemisphere on an atomic force microscope (AFM) cantilever can be used as colloidal probe. This work proves that nanocellulose composite ink is a candidate for 3D printing functional devices with special microstructures.     

Reliable Patterning,Transfer Printing and Post‐Assembly of Multiscale Adhesion‐Free Metallic Structures for Nanogap Device Applications

Patterning of metallic nanogaps with ultrasmall gap size on arbitrary substrates is of great importance for various applications in nanoelectronics, nanoplasmonics, and flexible optoelectronics. Common lithographic approaches suffer from limited resolution in defining ultrasmall nanogaps and restrictive available substrates for flexible and stretchable devices. In this work, a process portfolio to overcome the above limitations is proposed, enabling the fabrication of multiscale metallic nanogaps with reduced gap size on specific substrates for functional devices. The portfolio combines the recently developed sketch and peel lithography strategy, nanotransfer printing, and post‐mechanical assembly. Among the portfolio, the sketch and peel lithography strategy provides the unique capability to rapidly and reliably define multiscale adhesion‐free metallic nanostructures and nanogaps, which significantly facilitates the subsequent transfer printing process. Nanoplasmonic and nanoelectronic devices with ultrasmall nanogaps that are inaccessible with existing patterning approaches are fabricated to demonstrate the applicability of this fabrication strategy. The portfolio could also have potential for a variety of other applications in flexible and stretchable optics, electronics, and optoelectronics.     

基于迁移学习的软件缺陷预测

传统软件缺陷预测方法在解决跨项目缺陷预测过程中适应能力不足,主要是因为源项目和目标项目之间存在不同的特征分布.为了解决这个问题,提出一种新的加权贝叶斯迁移学习算法,算法首先收集训练数据和测试数据的特征信息,然后计算特征差异,将不同项目数据之间差异转化为训练数据权重,最后基于这些权重数据建立预测模型.在8个开源项目数据集上进行实验比较,实验结果表明与其他方法相比本文方法显著提高跨项目缺陷预测性能.     

Highly Customizable All Solution–Processed Polymer Light Emitting Diodes with Inkjet Printed Ag and Transfer Printed Conductive Polymer Electrodes

Inkjet and transfer printing processes are combined to easily form patterned poly(3,4‐ethylenedioxythiophene):poly(styrenesulfonate) (PEDOT:PSS) films as top anodes of all solution–processed inverted polymer light emitting diodes (PLEDs) on rigid glass and flexible plastic substrates. An adhesive PEDOT:PSS ink is formulated and fully customizable patterns are obtained using the inkjet printing process. In order to transfer the patterned PEDOT:PSS films, adhesion properties at interfaces during multistep transfer printing processes are carefully adjusted. The transferred PEDOT:PSS film on the plastic substrates shows not only a sheet resistance of 260.6 Ω/□ and a transmittance of 92.1% at 550 nm wavelength but also excellent mechanical flexibility. The PLEDs with spin‐coated functional layers sandwiched between the transferred PEDOT:PSS top anodes and inkjet‐printed Ag bottom cathodes are fabricated. The fabricated PLEDs on the plastic substrates show a high current efficiency of 10.4 cd A^?1 and high mechanical stability. It is noted that because both Ag and PEDOT:PSS electrodes can be patterned with a high degree of freedom via the inkjet printing process, highly customizable PLEDs with various pattern sizes and shapes are demonstrated on the glass and plastic substrates. Finally, with all solution process, a 5 × 7 passive matrix PLED array is demonstrated.     

Sample Target Substrates with Reduced Spot Size for MALDI‐TOF Mass Spectrometry Based on Patterned Self‐Assembled Monolayers

The wetting properties of structured self‐assembled monolayers are used to fabricate sample target substrates for MALDI‐TOF mass spectrometry. Combining the advantages of a hydrophobic‐hydrophilic surface pattern and the possibility of obtaining micrometer patterns allows an increase in the sensitivity of MALDI‐TOF mass spectrometry analysis and a reduction in the traceable concentration down to fmol µL^?1. This easy, cheap and fast patterning process provides substrates that allow sensitive, high‐resolution mass spectrometry of dilute solutions.     

Identifying the Activity Origin of a Cobalt Single-Atom Catalyst for Hydrogen Evolution Using Supervised Learning

Single-atom catalysts (SACs) have become the forefront of energy conversion studies, but unfortunately, the origin of their activity and the interpretation of the synchrotron spectrograms of these materials remain ambiguous. Here, systematic density functional theory computations reveal that the edge sites—zigzag and armchair—are responsible for the activity of the graphene-based Co (cobalt) SACs toward hydrogen evolution reaction (HER). Then, edge-rich (E)-Co single atoms (SAs) were rationally synthesized guided by theoretical results. Supervised learning techniques are applied to interpret the measured synchrotron spectrum of E-Co SAs. The obtained local environments of Co SAs, 65.49% of Co-4N-plane, 13.64% in Co-2N-armchair, and 20.86% in Co-2N-zigzag, are consistent with Athena fitting. Remarkably, E-Co SAs show even better HER electrocatalytic performance than commercial Pt/C at high current density. Using the joint effort of theoretical modeling, thorough characterization of the catalysts aided by supervised learning, and catalytic performance evaluations, this study not only uncovers the activity origin of Co SACs for HER but also lays the cornerstone for the rational design and structural analysis of nanocatalysts.     

Compatible Stealth Metasurface for Laser and Infrared with Radiative Thermal Engineering Enabled by Machine Learning

Metasurface-based mid-infrared stealth compatible with visual or laser provide a promising way to increase survivability of military installations. However, current designs of metasurfaces following traditional paradigm suffer from low efficiency on calculating global structural parameters for multispectral requirements and limited thermal radiation engineering. Here, a metasurface with high-performance compatible stealth and effect thermal management is proposed, based on a machine-learning-enabled inverse design approach. The approach can rapidly generate multiple non-unique solutions in global to match the desired spectra in multi-wavebands, utilizing neural networks with physical-based data dimensionality. The finally generated metasurface has low specular reflectance (<0.01) at near-infrared laser wavelength and enhanced broad absorptance peak at 5–8 µm, which is attributed to the reflection splitting effect and infrared plasmonic resonance, respectively. Furthermore, a low-cost fabrication method is developed to produce the metasurface by colloidal lithography. The metasurface is demonstrated to have excellent capability of radiative thermal control and significantly decrease the apparent temperature under thermal imager (>50 °C). This study reveals an opportunity to inversely generate multiple solutions for photonic structures targeting on multispectral responses, in a systematic and efficient manner.     

基于弱监督学习的去噪受限玻尔兹曼机特征提取算法

针对现有特征提取方法难以实现从含有复杂背景的图像中提取有用目标特征的瓶颈问题,提出了基于弱监督学习的去噪受限玻尔兹曼机特征提取算法.首先,利用训练样本,通过无监督学习方式训练一个标准受限玻尔兹曼机模型,从而获得一个包含可视单元层和隐藏单元层的层次结构模型;然后,对可视层的每个单元引入二值转换单元,对隐藏层,根据各节点的激活值大小和激活频率将其分为两组:前景特征隐层单元和背景特征隐层单元,得到一个二元混合式去噪玻尔兹曼机的模型;最后,通过多模交互方式,利用有限数量的样本标签信息对输入样本逐像素地进行采样训练,以此来提取目标特征.实验表明,本文的特征提取算法能够有效地从复杂的干扰背景中提取目标特征,提高了目标识别精度.     

结合分类与迁移学习的薄云覆盖遥感图像地物信息恢复

利用多源多时相遥感图像,给出一种结合分类与迁移学习的薄云覆盖遥感图像地物信息恢复算法.首先利用多方向非抽样对偶树复小波变换对多源多时相遥感图像进行多分辨率分解,对分解后的薄云图像的高频系数利用贝叶斯方法进行地物初分类;再对每类地物的低频系数通过迁移最小方差支持向量回归模型进行域自适应学习,获取模型参数;最后利用所获的迁移回归模型,用无云参考图像的低频系数预测薄云覆盖图像的低频系数,去除薄云,恢复薄云覆盖图像的地物信息.实验结果表明,本文算法恢复的地物细节清楚,光谱失真较小.特别对地物季节性变化的薄云覆盖遥感图像,本文算法能有效恢复薄云覆盖区域的地物信息.     

Supervised learning and frequency domain averaging-based adaptive channel estimation scheme for filterbank multicarrier with offset quadrature amplitude modulation

Filterbank multicarrier with offset quadrature amplitude modulation (FBMC-OQAM) is an attractive alternative to the orthogonal frequency division multiplexing (OFDM) modulation technique. In comparison with OFDM, the FBMC-OQAM signal has better spectral confinement and higher spectral efficiency and tolerance to synchronization errors, primarily due to per-subcarrier filtering using a frequency-time localized prototype filter. However, the filtering process introduces intrinsic interference among the symbols and complicates channel estimation (CE). An efficient way to improve the CE in FBMC-OQAM is using a technique known as windowed frequency domain averaging (FDA); however, it requires a priori knowledge of the window length parameter which is set based on the channel's frequency selectivity (FS). As the channel's FS is not fixed and not a priori known, we propose a k-nearest neighbor-based machine learning algorithm to classify the FS and decide on the FDA's window length. A comparative theoretical analysis of the mean-squared error (MSE) is performed to prove the proposed CE scheme's effectiveness, validated through extensive simulations. The adaptive CE scheme is shown to yield a reduction in CE-MSE and improved bit error rates compared with the popular preamble-based CE schemes for FBMC-OQAM, without a priori knowledge of channel's frequency selectivity.     

Fusing Stretchable Sensing Technology with Machine Learning for Human–Machine Interfaces

Sensors and algorithms are two fundamental elements to construct intelligent systems. The recent progress in machine learning (ML) has produced great advancements in intelligent systems, owing to the powerful data analysis capability of ML algorithms. However, the performance of most systems is still hindered by sensing techniques that typically rely on rigid and bulky sensor devices, which cannot conform to irregularly curved and dynamic surfaces for high-quality data acquisition. Skin-like stretchable sensing technology with unique characteristics, such as high conformability, low modulus, and light weight, has been recently developed to solve this issue. Here, the recent progress in the fusion of emerging stretchable electronics and ML technology, for bioelectrical signal recognition, tactile perception, and multimodal integration is summarized, and the challenges and future developments are further discussed. These efforts aim to accelerate various perception and reasoning tasks for advanced intelligent applications, such as human–machine interfaces, healthcare, and robotics.     

Computational Simulation for Breakdown and Energy Storage Performances with Optimization in Polymer Dielectrics

The breakthrough of energy storage technology will enable energy distribution and adaptation across space-time, which is revolutionary for the generation of energy. Optimizing the energy storage performance of polymer dielectrics remains challenging via the physical process of electrical breakdown in solid dielectrics is hard to be intuitively obtained. In this review article, the application of computational simulation technologies is summarized in energy-storage polymer dielectrics and the effect of control variables and design structures on the material properties with an emphasis on dielectric breakdown and energy storage performance are highlighted. The prediction and evaluation of material properties by combining various data analysis methods are reviewed. Finally, the outlook and challenges are discussed based on their current developments. This article covers not only an overview of the state-of-the-art advances of breakdown modeling in energy-storage polymer dielectrics but also the prospects that provide a new knob to synthesize high energy-storage polymer dielectrics via computational simulation and a new research paradigm.     

多种群纵横双向学习和信息互换的鲸鱼优化算法

鲸鱼优化算法(WOA)相较于传统的群体智能优化算法,具有较好的寻优能力和鲁棒性,但仍存在全局寻优能力有限、局部极值难以跳出等问题。针对上述不平衡问题,该文提出一种多种群纵横双向学习的种群划分思路,子群相互独立,子群内个体受到来自横向和纵向两个方向的最优值影响,从而规避局部最优,在探索和开发之间取得均衡。对纵向种群的所有个体,该文提出一种线性下降概率的个体置换策略,促进不同子群的信息流动,加快算法收敛。基于不同个体的历史进化信息,来进行策略算子选择,从而区别于现有基于随机数的策略算子选择方法。利用基准函数进行跨文献对比,数值结果表明该文算法具有很好的优越性和稳定性,在大多数问题上都获得了全局极值,具有较好的问题适用性。     

聚类与信息共享的多智能体深度强化学习协同控制交通灯

该文提出一种适用于多路口交通灯实时控制的多智能体深度循环Q-网络(MADRQN),目的是提高多个路口的联合控制效果。该方法将交通灯控制建模成马尔可夫决策过程,将每个路口的控制器作为智能体,根据位置和观测信息对智能体聚类,然后在聚类内部进行信息共享和中心化训练,并在每个训练过程结束时将评价值最高的值函数网络参数分享给其它智能体。在城市交通仿真软件(SUMO)下的仿真实验结果表明,所提方法能够减少通信的数据量,使得智能体之间的信息共享和中心化训练更加可行和高效,车辆平均等待时长少于当前最优的基于多智能体深度强化学习的交通灯控制方法,能够有效地缓解交通拥堵。