Enhancing the Adversarial Transferability with Channel Decomposition

The current adversarial attacks against deep learning models have achieved incredible success in the white-box scenario.However,they often exhibit weak transferability in the black-box scenario,especially when attacking those with defense mechanisms.In this work,we propose a new transfer-based blackbox attack called the channel decomposition attack method(CDAM).It can attack multiple black-box models by enhancing the transferability of the adversarial examples.On the one hand,it tunes the gradient and stabilizes the update direction by decomposing the channels of the input example and calculating the aggregate gradient.On the other hand,it helps to escape from local optima by initializing the data point with random noise.Besides,it could combine with other transfer-based attacks flexibly.Extensive experiments on the standard ImageNet dataset show that our method could significantly improve the transferability of adversarial attacks.Compared with the state-of-the-art method,our approach improves the average success rate from 88.2%to 96.6%when attacking three adversarially trained black-box models,demonstrating the remaining shortcomings of existing deep learning models.

A Low Power,High Sensitivity SiGe HBT Static Frequency Divider up to 90 GHz for Millimeter-Wave Application

A layout and connection optimization for static frequency divider is presented. The layout optimization provides a new circle topology transistors placement and reasonable connection structure, which reduces the parasitic effectively and enables self-oscillation frequency enhancement. Besides, bandwidth enhancement techniques based on a center-tap capacitor in input balun design and inductive peaking in latch design are adopted to improve further high frequency performance with low power consumption. As a proof of concept, design of a divide-by-2 static frequency divider in 0.13 μm SiGe BiCMOS technology is reported. With single-ended input clock signal, the divider is measured to be operated from 40 to 90 GHz. Phase noise measurements of a 90 GHz input clock signal indicate ideal behavior with no measurable noise contribution from the divider. The divider followed by a buffer that can deliver more than-10 dBm output power, which is sufficient to drive succeeding stage. To the author's knowledge, the divider exhibits a competitive power dissipation and the highest FOM among silicon based frequency dividers that operating higher than 70 GHz.

Aero-Engine Surge Fault Diagnosis Using Deep Neural Network

Deep learning techniques have outstanding performance in feature extraction and modelfitting.In thefield of aero-engine fault diagnosis,the intro-duction of deep learning technology is of great significance.The aero-engine is the heart of the aircraft,and its stable operation is the primary guarantee of the aircraft.In order to ensure the normal operation of the aircraft,it is necessary to study and diagnose the faults of the aero-engine.Among the many engine fail-ures,the one that occurs more frequently and is more hazardous is the wheeze,which often poses a great threat toflight safety.On the basis of analyzing the mechanism of aero-engine surge,an aero-engine surge fault diagnosis method based on deep learning technology is proposed.In this paper,key sensor data are obtained by analyzing different engine sensor data.An aero-engine surge data-set acquisition algorithm(ASDA)is proposed to sample the fault and normal points to generate the training set,validation set and test set.Based on neural net-work models such as one-dimensional convolutional neural network(1D-CNN),convolutional neural network(RNN),and long-short memory neural network(LSTM),different neural network optimization algorithms are selected to achieve fault diagnosis and classification.The experimental results show that the deep learning technique has good effect in aero-engine surge fault diagnosis.The aero-engine surge fault diagnosis network(ASFDN)proposed in this paper achieves better results.Through training,the network achieves more than 99%classification accuracy for the test set.

Stabilization of mixed-halide lead perovskites under light by photothermal effects

Mixed-halide lead perovskites(MHLPs) are semiconductor materials with bandgaps that are tunable across the visible spectrum and have seen promising applications in photovoltaics and optoelectronics.However, their segregation into phases with enriched halide components, under resonant light illumination and/or electric field, have hindered their practical applications. Herein, we demonstrate the stabilization of the MHLP photoluminescence(PL) peak as a function of their excitation intensities.This effect is associated with the phase segregation of MHLPs and their subsequent remixing by photothermal heating. We conclude that the balance between these opposing processes dictates the equilibrium PL peak of the MHLPs. The findings in this work could serve as a potential approach to obtain MHLP with stable emission peaks under operating conditions.

Hydrolysis mechanism of YbB_(2)C_(2)and the microstructure of the carbon derived from the hydrolysis reaction

Carbide-derived carbon(CDC)materials have gained great attention due to the excellent properties for various potential applications.Here,graphite crystal is formed during a room-temperature hydrolysis pro-cess of layered compound YbB_(2)C_(2).The formation mechanism can be demonstrated by a YbB_(2)C_(2)molecular cell:Yb^(3+)acts as a cathode where H_(2)O molecule is reduced to H atom and OH^(−)ion,while(B_(2)C_(2))^(3−)acts as an anode where OH−ion is oxidized to O atom.Then,YbB_(2)C_(2)molecular cell begins to disintegrate,i.e.,Yb^(3+)ion,B and C atoms dissociate from the molecular cell.The as-produced C atoms combine to form graphite crystal.The initial graphite crystal is a cabbage-like microsphere,and then it gradually disintegrates and transforms into layered graphite.In addition,YbB_(6),Yb_(3)(OH)_(3)n(BO_(3))_((3-n))sol,hydrogen,hydrocarbons,and carbon oxides form simultaneously.Our method provides a general and inexpensive route to obtain carbide-derived graphite crystal.

植物氮形态利用策略及对外来植物入侵性的影响

氮是影响外来植物入侵性的重要因素之一,但相关研究多关注土壤氮水平的效应,较少考虑氮形态的作用。为从土壤氮形态利用的角度阐释外来植物的入侵机制,本文在植物氮形态利用策略分析的基础上,综述了外来植物氮形态利用的偏好性及其对入侵性的影响。植物的氮形态利用策略有偏好性和可塑性两种,这可能与植物对土壤氮形态特性的长期适应有关;植物不仅可以对土壤氮形态做出响应,而且还能改造土壤氮形态,并对改变后的土壤氮形态做出反馈响应。很多外来植物入侵硝态氮占优势的干扰生境,偏好硝态氮的外来植物与本地植物竞争硝态氮;而偏好铵态氮的外来植物通过抑制土壤硝化作用,营造铵态氮环境,促进自身生长,同时抑制偏好硝态氮的本地植物生长。然而,植物氮形态利用策略不是一成不变的,而是受多种生物和非生物因素共同作用影响的复杂过程,今后应加强多因素交互作用对外来入侵植物氮形态利用策略的影响及机制研究,更好地揭示氮形态利用策略,尤其是氮形态利用的可塑性与外来植物入侵性的关系。

5G毫米波大规模MIMO阵列失效特性分析

大规模MIMO阵列是第五代移动通信(5G)的关键技术之一.然而,阵列在经历了长时间工作之后,会出现部分通道失效的情况,从而对大规模MIMO阵列的辐射性能造成不同程度的影响.本文针对5G毫米波大规模MIMO阵列中不同数量通道失效以及失效通道处于不同失效状态时的辐射特性进行了系统的仿真与实验研究,并在此基础上对MIMO阵列的辐射性能进行了统计分析.提出并建立了大规模MIMO阵列部分通道失效状态下的方向图参数概率模型,对概率模型的结果和统计分析的结果进行了对比,验证了概率模型的正确性和有效性,所给出的概率模型可以代替基于大量样本的仿真或实验结果进行统计分析的方法,对有不同比例的通道失效及不同失效状态时阵列的辐射特性进行快速评估,直接得出阵列主要辐射参数的期望值和方差.此外,本文还分析了不同失效比例和状态下主要辐射特性的概率分布.

Electrodeposition and Characterization of Ni/Ti_3Si(Al)C_2 Composite Coatings

Pure nickel and Ni/Ti3Si（AI）C2 composite coatings were prepared by electrodeposition method from an additive-free nickel Watt＇s bath and were characterized by X-ray diffraction, laser confocal microscopy and scanning electron microscopy. The effect of current and Ti3Si（AI）C2 concentration in the solution on the composition, grain size, preferred orientation and surface morphology of the electrodeposited coatings were investigated. （200）-oriented Ni coatings could be deposited at high current （62.5 mA）, while （220）-oriented Ni coatings could be obtained at low current （25 mA）. However, the presence of Ti3Si（Al）C2 particles disturbed the surface texture of Ni crystallites in the composite coatings. Based on the simulated morphology and the observed microstructure, the mechanisms for the change of preferred orientation and crystallite shapes in the presence of Ti3Si（Al）C2 particles were discussed. Moreover, microhardness and friction properties of pure nickel and Ni/Ti3Si（Al）C2 composite coatings were also compared.

Millimeter-wave wireless communications for home network in fiber-to-the-room scenario

Millimeter-wave(mmWave)technology has been well studied for both outdoor long-distance transmission and indoor short-range communication.In the recently emerging fiber-to-the-room(FTTR)architecture in the home network of the fifth generation fixed networks(F5G),mmWave technology can be cascaded well to a new optical network terminal in the room to enable extremely high data rate communication(i.e.,>10 Gb/s).In the FTTR+mmWave scenario,the rapid degradation of the mmWave signal in long-distance transmission and the significant loss against wall penetration are no longer the bottlenecks for real application.Moreover,the surrounding walls of every room provide excellent isolation to avoid interference and guarantee security.This paper provides insights and analysis for the new FTTR+mmWave architecture to improve the customer experience in future broadband services such as immersive audiovisual videos.

Microstructure and mechanical properties of h-BN/Yb_4Si_2O_7N_2 composites

A series of h-BN based composites with Yb_4Si_2O_7N_2 as a secondary phase were successfully synthesized by an in situ reaction hot pressing method. It was found that the relative density and room-temperature mechanical properties monotonically increased with increasing the content of Yb_4Si_2O_7N_2 from 20 to 50 vol%. When 50 vol% Yb_4Si_2O_7N_2 was introduced, the relative density of the composite reached 98.75%, and its flexural strength, compressive strength, fracture toughness, and hardness reached 338±10 MPa, 803±49 MPa, 2.06±0.06 MPa·m1/2, and 2.69±0.10 GPa, respectively. The strengthening effect of Yb_4Si_2O_7N_2 was mainly attributed to its high modulus and high hardness. Fine microstructure was also advantageous to strength and could lead to more tortuous crack propagation paths and then improve the fracture toughness of the composites simultaneously. Meanwhile, the composites maintained good machinability.

Raman Spectroscopy of Layered Compound YbB_(2)C_(2)

REB_(2)C_(2)(RE=Y and lanthanides)compounds have gained attention for their unique layered crystal structure.However,there have been few reports about Raman spectroscopy of REB_(2)C_(2) compounds up to now.Here,the Raman spectrum of YbB_(2)C_(2) is obtained by a micro-Raman spectroscope and the first-principles calculations.Raman active vibrational modes of YbB_(2)C_(2) are confirmed as A_(1g)(627 and 1311 cm^(-1)),B_(1g)(944 and 1172 cm^(-1)),B_(2g)(330 and 885 cm^(-1))and E_(g)(357 and 530 cm^(-1)).Atomic displacements of these modes are different,they can be divided into two groups:A_(1g)’,B_(1g) and B_(2g) correspond to ring breathing(δ_(in),in the plane)of B_(2)C_(2) layer;E_(g) is due to ring deformation(δ_(oop),out of the plane)of B_(2)C_(2) layer.These results are helpful to understand the individual structure of REB_(2)C_(2).