Omni-Detection of Adversarial Examples with Diverse Magnitudes

Deep neural networks(DNNs)are poten-tially susceptible to adversarial examples that are ma-liciously manipulated by adding imperceptible pertur-bations to legitimate inputs,leading to abnormal be-havior of models.Plenty of methods have been pro-posed to defend against adversarial examples.How-ever,the majority of them are suffering the follow-ing weaknesses:1)lack of generalization and prac-ticality.2)fail to deal with unknown attacks.To ad-dress the above issues,we design the adversarial na-ture eraser(ANE)and feature map detector(FMD)to detect fragile and high-intensity adversarial examples,respectively.Then,we apply the ensemble learning method to compose our detector,dealing with adver-sarial examples with diverse magnitudes in a divide-and-conquer manner.Experimental results show that our approach achieves 99.30%and 99.62%Area un-der Curve(AUC)scores on average when tested with various Lp norm-based attacks on CIFAR-10 and Im-ageNet,respectively.Furthermore,our approach also shows its potential in detecting unknown attacks.

An Empirical Study on the Effectiveness of Adversarial Examples in Malware Detection

Antivirus vendors and the research community employ Machine Learning(ML)or Deep Learning(DL)-based static analysis techniques for efficient identification of new threats,given the continual emergence of novel malware variants.On the other hand,numerous researchers have reported that Adversarial Examples(AEs),generated by manipulating previously detected malware,can successfully evade ML/DL-based classifiers.Commercial antivirus systems,in particular,have been identified as vulnerable to such AEs.This paper firstly focuses on conducting black-box attacks to circumvent ML/DL-based malware classifiers.Our attack method utilizes seven different perturbations,including Overlay Append,Section Append,and Break Checksum,capitalizing on the ambiguities present in the PE format,as previously employed in evasion attack research.By directly applying the perturbation techniques to PE binaries,our attack method eliminates the need to grapple with the problem-feature space dilemma,a persistent challenge in many evasion attack studies.Being a black-box attack,our method can generate AEs that successfully evade both DL-based and ML-based classifiers.Also,AEs generated by the attack method retain their executability and malicious behavior,eliminating the need for functionality verification.Through thorogh evaluations,we confirmed that the attack method achieves an evasion rate of 65.6%against well-known ML-based malware detectors and can reach a remarkable 99%evasion rate against well-known DL-based malware detectors.Furthermore,our AEs demonstrated the capability to bypass detection by 17%of vendors out of the 64 on VirusTotal(VT).In addition,we propose a defensive approach that utilizes Trend Locality Sensitive Hashing(TLSH)to construct a similarity-based defense model.Through several experiments on the approach,we verified that our defense model can effectively counter AEs generated by the perturbation techniques.In conclusion,our defense model alleviates the limitation of the most promising defense method,adversarial training,which is only effective against the AEs that are included in the training classifiers.

微探仁爱版初中英语教材中“Example”部分的教学策略

仁爱版初中英语教材中的“Example”部分是听说课教学的重要内容之一,也是学生在语言实践中迁移和运用知识的集中体现。如何提高“Example”部分教学内容的指导性、针对性和有效性是一线教师关注的焦点。文章分析了“Example”部分内容教学存在的问题,提出以主题意义为导向、深度融合教学与评价、合理分配教学时间和巩固教学成果等教学策略,以期提升“Example”部分的教学成效,逐步培养学生的英语学科核心素养。

Selection of Example Varieties Used in the DUS Test Guideline of Tagetes L.

[Objective] Taking the characteristic of flower diameter of Tagetes L.as an example,this study aimed to select example varieties used in the DUS Test Guideline of Tagetes L.[Method] Two continuous years of measurements of flower diameter of 25 varieties were collected and then analyzed by using the box plot to illustrate the uniformity and stability of flower diameter of each variety.[Result] According to the information of variability,distribution symmetry of measurements and outliers of flower diameter of varieties provided by box plots,variety 16,2 and 4 were selected as the example varieties for the three expression states with respective flower diameter of 3.0-4.4,6.0-7.4 and 9.0-10.4 cm.[Conclusion] The box plot is an efficient method for the general analysis of varieties,which provides information covering the actual and possible expression range,median and outliers of measurements of flower diameter of each variety.It also provides references for selecting example varieties for other quantitative characteristics and evaluating the quality of varieties.

基于语料库的学习者例证词“for example”对比研究

基于英美大学生作文语料库(LOCNESS)和中国英语学习者语料库(CLEC)调查中国英语学习者在写作中使用例证词"for example"的特点,研究发现,中国英语学习者无论在使用频数还是在用法上都具有明显区别于本族语者的特征。

Auto-expanded multi query examples technology in content-based image retrieval

In order to narrow the semantic gap existing in content-based image retrieval （CBIR）,a novel retrieval technology called auto-extended multi query examples （AMQE） is proposed.It expands the single one query image used in traditional image retrieval into multi query examples so as to include more image features related with semantics.Retrieving images for each of the multi query examples and integrating the retrieval results,more relevant images can be obtained.The property of the recall-precision curve of a general retrieval algorithm and the K-means clustering method are used to realize the expansion according to the distance of image features of the initially retrieved images.The experimental results demonstrate that the AMQE technology can greatly improve the recall and precision of the original algorithms.

An Intelligent Secure Adversarial Examples Detection Scheme in Heterogeneous Complex Environments

Image-denoising techniques are widely used to defend against Adversarial Examples(AEs).However,denoising alone cannot completely eliminate adversarial perturbations.The remaining perturbations tend to amplify as they propagate through deeper layers of the network,leading to misclassifications.Moreover,image denoising compromises the classification accuracy of original examples.To address these challenges in AE defense through image denoising,this paper proposes a novel AE detection technique.The proposed technique combines multiple traditional image-denoising algorithms and Convolutional Neural Network(CNN)network structures.The used detector model integrates the classification results of different models as the input to the detector and calculates the final output of the detector based on a machine-learning voting algorithm.By analyzing the discrepancy between predictions made by the model on original examples and denoised examples,AEs are detected effectively.This technique reduces computational overhead without modifying the model structure or parameters,effectively avoiding the error amplification caused by denoising.The proposed approach demonstrates excellent detection performance against mainstream AE attacks.Experimental results show outstanding detection performance in well-known AE attacks,including Fast Gradient Sign Method(FGSM),Basic Iteration Method(BIM),DeepFool,and Carlini&Wagner(C&W),achieving a 94%success rate in FGSM detection,while only reducing the accuracy of clean examples by 4%.

A New Example of Retrograde Solubility Model for Carbonate Rocks

Objective The dissolution and precipitation of carbonate during burial diagenetic process controls the reservoir property in deep buried strata. The geological process related with it has become a research focus during recent years. The most important dissolution fluids to carbonates are probably H2S and CO2 as byproducts of sulfate reduction in deep-buried setting with sulfate minerals, but carbonates are more soluble in relatively low temperature, which is the so-called retrograde solubility. Several geological processes can result in the decrease of temperature, including the upward migration of thermal fluids and tectonic uplift.

Digital image inpainting by example-based image synthesis method

A simple and effective image inpainting method is proposed in this paper, which is proved to be suitable for different kinds of target regions with shapes from little scraps to large unseemly objects in a wide range of images. It is an important improvement upon the traditional image inpainting techniques. By introducing a new bijeetive-mapping term into the matching cost function, the artificial repetition problem in the final inpainting image is practically solved. In addition, by adopting an inpainting error map, not only the target pixels are refined gradually during the inpainting process but also the overlapped target patches are combined more seamlessly than previous method. Finally, the inpainting time is dramatically decreased by using a new acceleration method in the matching process.

Characteristics of Authigenic Pyrite and its Sulfur Isotopes Influenced by Methane Seep--Taking the Core A at Site 79 of the Middle Okinawa Trough as an Example

Objective Authigenic pyrite often develops extensively in marine sediments,which is an important product of sulfate reduction in an anoxic environment.It has a specific appearance and complicated sulfur isotopic properties,and acts as important evidence of methane seep in marine sediments.Strong AOM（anaerobic oxidation of methane）activity has developed in the Okinawa Trough.

Patterns of Clay Minerals Transformation in Clay Gouge, with Examples from Revers Fault Rocks in Devonina Niqiuhe Formation in The Dayangshu Basin

The role of authigenic clay growth in clay gouge is increasingly recognized as a key to understanding the mechanics of berittle faulting and fault zone processes,including creep and seismogenesis,and providing new insights into the ongoing debate about the frictional strength of brittle fault(Haines and van der Pluijm,2012).However,neither the conditions nor the processes which

A Survey on Adversarial Examples in Deep Learning

Adversarial examples are hot topics in the field of security in deep learning.The feature,generation methods,attack and defense methods of the adversarial examples are focuses of the current research on adversarial examples.This article explains the key technologies and theories of adversarial examples from the concept of adversarial examples,the occurrences of the adversarial examples,the attacking methods of adversarial examples.This article lists the possible reasons for the adversarial examples.This article also analyzes several typical generation methods of adversarial examples in detail:Limited-memory BFGS(L-BFGS),Fast Gradient Sign Method(FGSM),Basic Iterative Method(BIM),Iterative Least-likely Class Method(LLC),etc.Furthermore,in the perspective of the attack methods and reasons of the adversarial examples,the main defense techniques for the adversarial examples are listed:preprocessing,regularization and adversarial training method,distillation method,etc.,which application scenarios and deficiencies of different defense measures are pointed out.This article further discusses the application of adversarial examples which currently is mainly used in adversarial evaluation and adversarial training.Finally,the overall research direction of the adversarial examples is prospected to completely solve the adversarial attack problem.There are still a lot of practical and theoretical problems that need to be solved.Finding out the characteristics of the adversarial examples,giving a mathematical description of its practical application prospects,exploring the universal method of adversarial example generation and the generation mechanism of the adversarial examples are the main research directions of the adversarial examples in the future.

for example还是such as

高中学生写英语作文时，常需列举事物，如：

DroidEnemy: Battling adversarial example attacks for Android malware detection

In recent years,we have witnessed a surge in mobile devices such as smartphones,tablets,smart watches,etc.,most of which are based on the Android operating system.However,because these Android-based mobile devices are becoming increasingly popular,they are now the primary target of mobile malware,which could lead to both privacy leakage and property loss.To address the rapidly deteriorating security issues caused by mobile malware,various research efforts have been made to develop novel and effective detection mechanisms to identify and combat them.Nevertheless,in order to avoid being caught by these malware detection mechanisms,malware authors are inclined to initiate adversarial example attacks by tampering with mobile applications.In this paper,several types of adversarial example attacks are investigated and a feasible approach is proposed to fight against them.First,we look at adversarial example attacks on the Android system and prior solutions that have been proposed to address these attacks.Then,we specifically focus on the data poisoning attack and evasion attack models,which may mutate various application features,such as API calls,permissions and the class label,to produce adversarial examples.Then,we propose and design a malware detection approach that is resistant to adversarial examples.To observe and investigate how the malware detection system is influenced by the adversarial example attacks,we conduct experiments on some real Android application datasets which are composed of both malware and benign applications.Experimental results clearly indicate that the performance of Android malware detection is severely degraded when facing adversarial example attacks.

Relic Oceanic Crust at Sub-arc Depth: an Example from UHP Eclogites Enclosed in Serpentinites from the Southwestern Tianshan Mountains, China

Ultra-high pressure(UHP)eclogites that derive from subducted oceanic crust are rarely found at the Earth’s surface because they need to be enclosed in a buoyant host rock such as serpentinites that facilitate exhumation

Relic Oceanic Crust at Sub-Arc Depth: An Example from UHP Eclogites Enclosed in Serpentinites from the Southwestern Tianshan, China

Ultra-high pressure（UHP）eclogites that derive from subducted oceanic crust are rarely found at the Earth’s surface because they need to be enclosed in a buoyant host rock such as serpentinites that facilitate exhumation（Hermann et al.,2000;Guillot et al.,2001）.Under normal subduction geotherms,serpentinites break down just before UHP conditions are reached and therefore most of the exhumed eclogites representing subducted oceanic crust formed under fore-arc conditions.We investigated eclogite blocks enclosed into serpentinites that occur in the southwestern Tianshan oceanic subduction,China.A previous study proved that the serpentinites derive from altered oceanic crust and experienced UHP metamorphism at low temperatures of 510-530°C（Shen et al.,2015）.Three relatively fresh eclogite samples were studied in detail.Sample 129-7 shows the retrograde mineral assemblage of amphibole+biotite+albite+chlorite+minor titanite and peak metamorphic relics of omphacite+garnet±chlorite.Sample C107-23 is mainly composed of amphibole+albite+chlorite+zoisite+muscovite+minor titanite as a retrograde assemblage and garnet+phengite as the peak metamorphic relics with omphacite only found as inclusions in garnet.Similar to sample C107-23,sample C11066 preserves large-grained euhedral to subhedral garnet relics with omphacite inclusions,and epidote,diopside,amphibole,muscovite,chlorite,albite and biotite are in the matrix belong to the retrograde assemblage.These three retrograde eclogite samples were modelled using thermodynamic calculations in the Mn NCKFMSHO（Mn O-Na;O-Ca O-K;O-FeO-Mg O-Al;O;-SiO;-H;O-Fe;O;）system.Based on the peak assemblage of omphacite+garnet and the crossing of the grossular and pyrope isopleths in garnet,peak P-T conditions of;60-470oC,28-29 kbar（129-7）,450-500oC,28-35 kbar（C107-23）,;75-505oC,26-29 kbar（C11066）were calculated.The retrograde assemblages indicate near isothermal decompression resulting in a clockwise P-T evolution of these eclogites.The peak metamorphic pressures at 500°C are well within UHP conditions（coesite stability field）and are within error the same as peak conditions of the host serpentinites（Shen et al.,2015）.This provides evidence that eclogites and serpentinites shared the same evolution.We infer that the subducted low-density serpentinites were assembled with the high-density eclogites during subdution and helped the latter to exhume back to the surface.The studied eclogites thus represent rare examples of relics of oceanic crust that was subducted to sub-arc depth.

Fluidizate-Explosive Occurrences in Ophiolites as Indicator of the Subduction Zone Activity: The Urals Example

It is known that the formation of oceanic crust occurs in different geodynamic settings,accompanying by the emergence of mantle-magmatic ophiolite complexes having a distinctive properties.In the process of mantle-crustal evolution of the ophiolites are undergoing significant changes with the formation of peculiar(on structure and composition)rocks,sometimes with unusual mineral paragenesis.The presence of such rocks in mélange tectonic zones greatly complicates to determine their origin.In the Ural folded belt(length more than 2,000 km)separating the East European Platform and the West Siberian sedimentary basin,ophiolites are widespread forming a chain of mafic-ultramafic massifs(Fig.1)located in the allochthonous position with mélange at the bottom(Puchkov,2013).With the Urals ophiolites are associated occurrences of eclogites,jadeites,ruby and other rocks of unclear nature,sometimes regarded as potentially diamondiferous.Such formations of unclear genesis include the associating with ophiolites metabasites of higher alkalinity composing the body in the mantle peridotites of the mélange Main Uralian Fault(MUF)zone(Shmelev,2005).By this time they are determined in different parts of the fault zone,but most completely are known in the Sub Polar Urals,where are distinguished under the name of Sertynya alkaline-ultramafic complex,which is located just 25 km east of Hartes kimberlitic complex(Fig.1).Formally,its affiliation to diamond-bearing associations is confirmed by finding of grains and fragments of natural diamond in the weathering crust.A detailed study of the rock complexes shows that in reality they have a polygenic nature,combine theelements of proper magmatic and fluidizate-explosive formations,the appearance of which was interfaced with the processes at the slab-mantle wedge boundary in subduction zones.Polygenic nature of the rocks is reflected in the existence of three interrelated structural-geological units:(1)bodies and dikes of uniformmetadiabasesanddensefine-grainedmetadolerites(lamprophyres),(2)fluidal-brecciated dolerites('tuff breccias')and(3)structural weathering crust with angular or rounded fragments(blocks)of metadolerites and serpentinites.The rocks have experienced rodingitization and permeated with net of veins a vesuvianite composition.The host peridotite matrix(harzburgites and dunites)has undergone serpentinizationandchloritization.Structural relationships give grounds for distinguishing in the history of the complex formation the magmatic proper(dolerite dyke and lamprophyre intrusion)and infiltration fluidizate-explosive(metasomatic transformation of dolerite)stages.Peculiarities of petrography and mineralogy of rock complexes does not allow to compare them with lamproites and kimberlites.Metadiabases demonstrate relics of ophitic structure,as primary paragenesis is completely replaced by aggregate of chlorite,zoisite and leucoxene.Dolerites(lamprophyres)have a uniform fine-grained or porphyry structure with phenocrysts of clinopyroxene,brown amphibole and leucoxene(sphene),which are immersed in a fine-scaly aggregate of light green mica.In the rocks amphibole,garnet and vesuvian are present.Clinopyroxene corresponds to augite with moderate content of titanium and alumina(up to 3.5wt.%),showing a normal magmatic zonation in composition.Mica previously wrongly called as phlogopite,actually has an extremely ferrous composition and corresponds to biotite(annite).Amphibole is presented by magmatic titaniferous tschermakite hornblende and metamorphic(bluish)variety of sodium-calcium composition(taramite).Garnet is presented by exceptionally grossular of rodingite type.Mineralogy of weathering crust reveals similar features,but in the samples it is marked the presence of muscovite,orthoclase and weakly ferrous diopside.An important feature of the weathering crust is the presence of shear surfaces on minerals,resulting in fracturing due to internal stress,confirming the explosive nature of protolith.The bulk chemical composition of rocks is characterized by significant variations in the content of silica(30-46 wt.%)and alkalis(0-6.5 wt.%).These metabasites have consistently a low magnesia number and high titanium oxide content(1.5-3.0 wt.%).Side by side with these are been established the uniform slope REE distribution trends similar to the trend of oceanic basalts N-MORB type(Fig.2).The level of trace element compositions does not depend on variations in the alkalinity of the rocks,but clearly correlates with the titanium content.Unlike them the Hartes kimberlites demonstrate the distribution with deficit of HREE,andthe level of the elements content is correlated with the alkalinity of rocks(Mahotkin et al.,1998).Another important geochemical feature of the Sertynya complex rocks is a regular behavior of the mobile LILE elements(Cs,Rb,Ba,K).In the varieties of rocks with mica enriched by alkalis,it is recorded extremely high level of LILE,exceeding the level of contents in N-MORB basalts at 10-10000 times!In the metabasites varieties with low level of alkalinity,LILE content is sharply(except Cs)reduced to minimum values(Fig.2).The observed pattern of the element distribution is undoubtedly the result of postmagmatic fluid-metasomatic alteration of the original rocks.Tectonic position and the primary composition characteristics of the metadolerites give reason to consider them as fragments of the ophiolite sheeted dike complex(Shmelev,2005).The famous dike complexes in the ophiolite massifs of the MUF zone(east of mélange)belong to suprasubduction formations of Paleozoic age.However the obtained mainly ancient U-Pb zircon dating(up to Archean inclusive)for metadolerites of the Sertynya complex,make it possible to assume its Vendian-Early Cambrian(530-617 Ma)age.It permits to compare the Sertynya metabasites with the Vendian metaophiolites of the MUF zone in the Middle Urals(Petrov et al.,2010).It is noteworthy that similar age datings(520-550 Ma)are also established for kimberlites of the Hartes complex located to the west of ophiolites.Therefore,thepresenceofthe Vendian-Cambrian ophiolite of MOR-type in the MUF mélange zone,'changing'to the east of Ordovician ophiolites SSZ-type,seems quite possible.The obtained data allow to suggest an original interpretation of nature of the Urals fluidizate-explosive formations considering the process specifics in the subduction zones(Bebout and Barton,2002).Accordingto this model,the pre-Ordovician(?)oceanic crust has undergone transformations and deformations on the slab-mantle wedge boundary during the subduction.As a result of slab dehydration it occurred a flow of aqueous fluids,which were enriched with the extracted from sedimentary rocks the LILE elements and percolated through the mantle substrate with dolerite dyke complex.Interaction with them led to the formation of chlorite-zoisite and/or mica(biotite-bearing)fluidizates and in the presence of a gas phase-fluidizate-explosive breccias with subsequent development of weathering crust.In the surrounding peridotites an explosive process is marked by the formation of pseudokimberlite breccias.Fluidized-explosive occurrences in mantle peridotites of mélange zones should be considered as indicators of the subduction slab-mantle interaction at relatively shallow levels involving enriched LILE fluids(without melts participation),rising as the front from the subduction zone.In this interpretation,there is no need toappealtothealkaline-ultramaficor lamproit-kimberlite hypothesis of the genesis of these formations,however,the question of their potential diamondiferous remains to be open.The proposed interpretation of the fluidizate-explosive occurrences makes it possible to comprehendthat in reality the mélange is a complex formation with signs of not onlycollisional(as usually is considered),but also of earlier subduction events.

An Example of the Scientists—Congratulate Professor SHI Changxu on His Seventyth Birthday

November 15 is the seventyth anniver-sary of Prof.SHI Changxu(C.H.Shih),honorary director of the Institute of MetalResearch,Academia Sinica.As a memberof Chinese Academy of Sciences and direc-tor of the First Division of Technical Sci-ence,Academia Sinica,vice president ofthe National Natural Science Foundation ofChina,and editor-in-chief of ActaMetallurgica Sinica,Prof.Shi is learnedand distinguished in his study,and showsenergy and initiative in his work.And he isextremely loyal and devoted to his countryand his people,so that he commands ourdeep respect and sets an example to the