Visible to the public Biblio

Found 1681 results

Filters: First Letter Of Last Name is Z  [Clear All Filters]
Book Chapter
Li, Bo, Vorobeychik, Yevgeniy.  2014.  Feature Cross-Substitution in Adversarial Classification. Advances in Neural Information Processing Systems 27. :2087–2095.

The success of machine learning, particularly in supervised settings, has led to numerous attempts to apply it in adversarial settings such as spam and malware detection. The core challenge in this class of applications is that adversaries are not static data generators, but make a deliberate effort to evade the classifiers deployed to detect them. We investigate both the problem of modeling the objectives of such adversaries, as well as the algorithmic problem of accounting for rational, objective-driven adversaries. In particular, we demonstrate severe shortcomings of feature reduction in adversarial settings using several natural adversarial objective functions, an observation that is particularly pronounced when the adversary is able to substitute across similar features (for example, replace words with synonyms or replace letters in words). We offer a simple heuristic method for making learning more robust to feature cross-substitution attacks. We then present a more general approach based on mixed-integer linear programming with constraint generation, which implicitly trades off overfitting and feature selection in an adversarial setting using a sparse regularizer along with an evasion model. Our approach is the first method for combining an adversarial classification algorithm with a very general class of models of adversarial classifier evasion. We show that our algorithmic approach significantly outperforms state-of-the-art alternatives.

Conference Paper
Zong, Fang, Yong, Ouyang, Gang, Liu.  2018.  3D Modeling Method Based on Deep Belief Networks (DBNs) and Interactive Evolutionary Algorithm (IEA). Proceedings of the 2018 International Conference on Big Data and Computing. :124-128.

3D modeling usually refers to be the use of 3D software to build production through the virtual 3D space model with 3D data. At present, most 3D modeling software such as 3dmax, FLAC3D and Midas all need adjust models to get a satisfactory model or by coding a precise modeling. There are many matters such as complicated steps, strong profession, the high modeling cost. Aiming at this problem, the paper presents a new 3D modeling methods which is based on Deep Belief Networks (DBN) and Interactive Evolutionary Algorithm (IEA). Following this method, firstly, extract characteristic vectors from vertex, normal, surfaces of the imported model samples. Secondly, use the evolution strategy, to extract feature vector for stochastic evolution by artificial grading control the direction of evolution, and in the process to extract the characteristics of user preferences. Then, use evolution function matrix to establish the fitness approximation evaluation model, and simulate subjective evaluation. Lastly, the user can control the whole machine simulation evaluation process at any time, and get a satisfactory model. The experimental results show that the method in this paper is feasible.

Yang, B., Liu, F., Yuan, L., Zhang, Y..  2020.  6LoWPAN Protocol Based Infrared Sensor Network Human Target Locating System. 2020 15th IEEE Conference on Industrial Electronics and Applications (ICIEA). :1773–1779.
This paper proposes an infrared sensor human target locating system for the Internet of Things. In this design, the wireless sensor network is designed and developed to detect human targets by using 6LoWPAN protocol and pyroelectric infrared (PIR) sensors. Based on the detection data acquired by multiple sensor nodes, K-means++ clustering algorithm combined with cost function is applied to complete human target location in a 10m×10m detection area. The experimental results indicate the human locating system works well and the user can view the location information on the terminal devices.
Dong, C., Liu, Y., Zhang, Y., Shi, P., Shao, X., Ma, C..  2018.  Abnormal Bus Data Detection of Intelligent and Connected Vehicle Based on Neural Network. 2018 IEEE International Conference on Computational Science and Engineering (CSE). :171–176.
In the paper, our research of abnormal bus data analysis of intelligent and connected vehicle aims to detect the abnormal data rapidly and accurately generated by the hackers who send malicious commands to attack vehicles through three patterns, including remote non-contact, short-range non-contact and contact. The research routine is as follows: Take the bus data of 10 different brands of intelligent and connected vehicles through the real vehicle experiments as the research foundation, set up the optimized neural network, collect 1000 sets of the normal bus data of 15 kinds of driving scenarios and the other 300 groups covering the abnormal bus data generated by attacking the three systems which are most common in the intelligent and connected vehicles as the training set. In the end after repeated amendments, with 0.5 seconds per detection, the intrusion detection system has been attained in which for the controlling system the abnormal bus data is detected at the accuracy rate of 96% and the normal data is detected at the accuracy rate of 90%, for the body system the abnormal one is 87% and the normal one is 80%, for the entertainment system the abnormal one is 80% and the normal one is 65%.
Wang, X., Zhou, S..  2017.  Accelerated Stochastic Gradient Method for Support Vector Machines Classification with Additive Kernel. 2017 First International Conference on Electronics Instrumentation Information Systems (EIIS). :1–6.

Support vector machines (SVMs) have been widely used for classification in machine learning and data mining. However, SVM faces a huge challenge in large scale classification tasks. Recent progresses have enabled additive kernel version of SVM efficiently solves such large scale problems nearly as fast as a linear classifier. This paper proposes a new accelerated mini-batch stochastic gradient descent algorithm for SVM classification with additive kernel (AK-ASGD). On the one hand, the gradient is approximated by the sum of a scalar polynomial function for each feature dimension; on the other hand, Nesterov's acceleration strategy is used. The experimental results on benchmark large scale classification data sets show that our proposed algorithm can achieve higher testing accuracies and has faster convergence rate.

Perry, David M., Mattavelli, Andrea, Zhang, Xiangyu, Cadar, Cristian.  2017.  Accelerating Array Constraints in Symbolic Execution. Proceedings of the 26th ACM SIGSOFT International Symposium on Software Testing and Analysis. :68–78.

Despite significant recent advances, the effectiveness of symbolic execution is limited when used to test complex, real-world software. One of the main scalability challenges is related to constraint solving: large applications and long exploration paths lead to complex constraints, often involving big arrays indexed by symbolic expressions. In this paper, we propose a set of semantics-preserving transformations for array operations that take advantage of contextual information collected during symbolic execution. Our transformations lead to simpler encodings and hence better performance in constraint solving. The results we obtain are encouraging: we show, through an extensive experimental analysis, that our transformations help to significantly improve the performance of symbolic execution in the presence of arrays. We also show that our transformations enable the analysis of new code, which would be otherwise out of reach for symbolic execution.

Tan, Gaosheng, Zhang, Rui, Ma, Hui, Tao, Yang.  2017.  Access Control Encryption Based on LWE. Proceedings of the 4th ACM International Workshop on ASIA Public-Key Cryptography. :43–50.

Damgard et al. proposed a new primitive called access control encryption (ACE) [6] which not only protects the privacy of the message, but also controls the ability of the sender to send the message. We will give a new construction based on the Learning with Error (LWE) assumption [12], which is one of the two open problems in [6]. Although there are many public key encryption schemes based on LWE and supporting homomorphic operations. We find that not every scheme can be used to build ACE. In order to keep the security and correctness of ACE, the random constant chosen by the sanitizer should satisfy stricter condition. We also give a different security proof of ACE based on LWE from it based on DDH. We will see that although the modulus of LWE should be super-polynomial, the ACE scheme is still as secure as the general public key encryption scheme based on the lattice [5].

Zhang, Xianzhen, Chen, Zhanfang, Gong, Yue, Liu, Wen.  2019.  A Access Control Model of Associated Data Sets Based on Game Theory. 2019 International Conference on Machine Learning, Big Data and Business Intelligence (MLBDBI). :1–4.
With the popularity of Internet applications and rapid development, data using and sharing process may lead to the sensitive information divulgence. To deal with the privacy protection issue more effectively, in this paper, we propose the associated data sets protection model based on game theory from the point of view of realizing benefits from the access of privacy is about happen, quantify the extent to which visitors gain sensitive information, then compares the tolerance of the sensitive information owner and finally decides whether to allow the visitor to make an access request.
Shahid, U., Farooqi, S., Ahmad, R., Shafiq, Z., Srinivasan, P., Zaffar, F..  2017.  Accurate Detection of Automatically Spun Content via Stylometric Analysis. 2017 IEEE International Conference on Data Mining (ICDM). :425–434.

Spammers use automated content spinning techniques to evade plagiarism detection by search engines. Text spinners help spammers in evading plagiarism detectors by automatically restructuring sentences and replacing words or phrases with their synonyms. Prior work on spun content detection relies on the knowledge about the dictionary used by the text spinning software. In this work, we propose an approach to detect spun content and its seed without needing the text spinner's dictionary. Our key idea is that text spinners introduce stylometric artifacts that can be leveraged for detecting spun documents. We implement and evaluate our proposed approach on a corpus of spun documents that are generated using a popular text spinning software. The results show that our approach can not only accurately detect whether a document is spun but also identify its source (or seed) document - all without needing the dictionary used by the text spinner.

Copty, Fady, Danos, Matan, Edelstein, Orit, Eisner, Cindy, Murik, Dov, Zeltser, Benjamin.  2018.  Accurate Malware Detection by Extreme Abstraction. Proceedings of the 34th Annual Computer Security Applications Conference. :101–111.

Modern malware applies a rich arsenal of evasion techniques to render dynamic analysis ineffective. In turn, dynamic analysis tools take great pains to hide themselves from malware; typically this entails trying to be as faithful as possible to the behavior of a real run. We present a novel approach to malware analysis that turns this idea on its head, using an extreme abstraction of the operating system that intentionally strays from real behavior. The key insight is that the presence of malicious behavior is sufficient evidence of malicious intent, even if the path taken is not one that could occur during a real run of the sample. By exploring multiple paths in a system that only approximates the behavior of a real system, we can discover behavior that would often be hard to elicit otherwise. We aggregate features from multiple paths and use a funnel-like configuration of machine learning classifiers to achieve high accuracy without incurring too much of a performance penalty. We describe our system, TAMALES (The Abstract Malware Analysis LEarning System), in detail and present machine learning results using a 330K sample set showing an FPR (False Positive Rate) of 0.10% with a TPR (True Positive Rate) of 99.11%, demonstrating that extreme abstraction can be extraordinarily effective in providing data that allows a classifier to accurately detect malware.

Li, Chunhua, He, Jinbiao, Lei, Cheng, Guo, Chan, Zhou, Ke.  2018.  Achieving Privacy-Preserving CP-ABE Access Control with Multi-Cloud. 2018 IEEE Intl Conf on Parallel Distributed Processing with Applications, Ubiquitous Computing Communications, Big Data Cloud Computing, Social Computing Networking, Sustainable Computing Communications (ISPA/IUCC/BDCloud/SocialCom/SustainCom). :801—808.
Cloud storage service makes it very convenient for people to access and share data. At the same time, the confidentiality and privacy of user data is also facing great challenges. Ciphertext-Policy Attribute-Based Encryption (CP-ABE) scheme is widely considered to be the most suitable security access control technology for cloud storage environment. Aiming at the problem of privacy leakage caused by single-cloud CP-ABE which is commonly adopted in the current schemes, this paper proposes a privacy-preserving CP-ABE access control scheme using multi-cloud architecture. By improving the traditional CP-ABE algorithm and introducing a proxy to cut the user's private key, it can ensure that only a part of the user attribute set can be obtained by a single cloud, which effectively protects the privacy of user attributes. Meanwhile, the intermediate logical structure of the access policy tree is stored in proxy, and only the leaf node information is stored in the ciphertext, which effectively protects the privacy of the access policy. Security analysis shows that our scheme is effective against replay and man-in-the-middle attacks, as well as user collusion attack. Experimental results also demonstrates that the multi-cloud CP-ABE does not significantly increase the overhead of storage and encryption compared to the single cloud scheme, but the access control overhead decreases as the number of clouds increases. When the access policy is expressed with a AND gate structure, the decryption overhead is obviously less than that of a single cloud environment.
Wang, Z., Hu, H., Zhang, C..  2017.  On achieving SDN controller diversity for improved network security using coloring algorithm. 2017 3rd IEEE International Conference on Computer and Communications (ICCC). :1270–1275.

The SDN (Software Defined Networking) paradigm rings flexibility to the network management and is an enabler to offer huge opportunities for network programmability. And, to solve the scalability issue raised by the centralized architecture of SDN, multi-controllers deployment (or distributed controllers system) is envisioned. In this paper, we focus on increasing the diversity of SDN control plane so as to enhance the network security. Our goal is to limit the ability of a malicious controller to compromise its neighboring controllers, and by extension, the rest of the controllers. We investigate a heterogeneous Susceptible-Infectious-Susceptible (SIS) epidemic model to evaluate the security performance and propose a coloring algorithm to increase the diversity based on community detection. And the simulation results demonstrate that our algorithm can reduce infection rate in control plane and our work shows that diversity must be introduced in network design for network security.

Zhang, H., Ma, J., Wang, Y., Pei, Q..  2009.  An Active Defense Model and Framework of Insider Threats Detection and Sense. 2009 Fifth International Conference on Information Assurance and Security. 1:258—261.
Insider attacks is a well-known problem acknowledged as a threat as early as 1980s. The threat is attributed to legitimate users who take advantage of familiarity with the computational environment and abuse their privileges, can easily cause significant damage or losses. In this paper, we present an active defense model and framework of insider threat detection and sense. Firstly, we describe the hierarchical framework which deal with insider threat from several aspects, and subsequently, show a hierarchy-mapping based insider threats model, the kernel of the threats detection, sense and prediction. The experiments show that the model and framework could sense the insider threat in real-time effectively.
Hui Lin, University of Illinois at Urbana-Champaign, Adam Slagell, University of Illinois at Urbana-Champaign, Catello Di Marino, University of Illinois at Urbana-Champaugn, Zbigniew Kalbarczyk, University of Illinois at Urbana-Champaign, Ravishankar K. Iyer, University of Illinois at Urbana-Champaign.  2013.  Adapting Bro into SCADA: Building a Specification-based Instrusion Detection System for the DNP3 Protocol. Eighth Annual Security and Information Intelligence Research Workshop (CSIRRW 2013).

When SCADA systems are exposed to public networks, attackers can more easily penetrate the control systems that operate electrical power grids, water plants, and other critical infrastructures. To detect such attacks, SCADA systems require an intrusion detection technique that can understand the information carried by their usually proprietary network protocols.

To achieve that goal, we propose to attach to SCADA systems a specification-based intrusion detection framework based on Bro [7][8], a runtime network traffic analyzer. We have built a parser in Bro to support DNP3, a network protocol widely used in SCADA systems that operate electrical power grids. This built-in parser provides a clear view of all network events related to SCADA systems. Consequently, security policies to analyze SCADA-specific semantics related to the network events can be accurately defined. As a proof of concept, we specify a protocol validation policy to verify that the semantics of the data extracted from network packets conform to protocol definitions. We performed an experimental evaluation to study the processing capabilities of the proposed intrusion detection framework.

Ding, Q., Peng, X., Zhang, X., Hu, X., Zhong, X..  2017.  Adaptive observer-based fault diagnosis for sensor in a class of MIMO nonlinear system. 2017 36th Chinese Control Conference (CCC). :7051–7058.

This paper presents a novel sensor parameter fault diagnosis method for generally multiple-input multiple-output (MIMO) affine nonlinear systems based on adaptive observer. Firstly, the affine nonlinear systems are transformed into the particular systems via diffeomorphic transformation using Lie derivative. Then, based on the techniques of high-gain observer and adaptive estimation, an adaptive observer structure is designed with simple method for jointly estimating the states and the unknown parameters in the output equation of the nonlinear systems. And an algorithm of the fault estimation is derived. The global exponential convergence of the proposed observer is proved succinctly. Also the proposed method can be applied to the fault diagnosis of generally affine nonlinear systems directly by the reversibility of aforementioned coordinate transformation. Finally, a numerical example is presented to illustrate the efficiency of the proposed fault diagnosis scheme.

Lv, Chengcheng, Zhang, Long, Zeng, Fanping, Zhang, Jian.  2019.  Adaptive Random Testing for XSS Vulnerability. 2019 26th Asia-Pacific Software Engineering Conference (APSEC). :63–69.
XSS is one of the common vulnerabilities in web applications. Many black-box testing tools may collect a large number of payloads and traverse them to find a payload that can be successfully injected, but they are not very efficient. And previous research has paid less attention to how to improve the efficiency of black-box testing to detect XSS vulnerability. To improve the efficiency of testing, we develop an XSS testing tool. It collects 6128 payloads and uses a headless browser to detect XSS vulnerability. The tool can discover XSS vulnerability quickly with the ART(Adaptive Random Testing) method. We conduct an experiment using 3 extensively adopted open source vulnerable benchmarks and 2 actual websites to evaluate the ART method. The experimental results indicate that the ART method can effectively improve the fuzzing method by more than 27.1% in reducing the number of attempts before accomplishing a successful injection.
Fan, Renshi, Du, Gaoming, Xu, Pengfei, Li, Zhenmin, Song, Yukun, Zhang, Duoli.  2019.  An Adaptive Routing Scheme Based on Q-learning and Real-time Traffic Monitoring for Network-on-Chip. 2019 IEEE 13th International Conference on Anti-counterfeiting, Security, and Identification (ASID). :244—248.
In the Network on Chip (NoC), performance optimization has always been a research focus. Compared with the static routing scheme, dynamical routing schemes can better reduce the data of packet transmission latency under network congestion. In this paper, we propose a dynamical Q-learning routing approach with real-time monitoring of NoC. Firstly, we design a real-time monitoring scheme and the corresponding circuits to record the status of traffic congestion for NoC. Secondly, we propose a novel method of Q-learning. This method finds an optimal path based on the lowest traffic congestion. Finally, we dynamically redistribute network tasks to increase the packet transmission speed and balance the traffic load. Compared with the C-XY routing and DyXY routing, our method achieved improvement in terms of 25.6%-49.5% and 22.9%-43.8%.
Li, F., Jiang, M., Zhang, Z..  2017.  An adaptive sparse representation model by block dictionary and swarm intelligence. 2017 2nd IEEE International Conference on Computational Intelligence and Applications (ICCIA). :200–203.

The pattern recognition in the sparse representation (SR) framework has been very successful. In this model, the test sample can be represented as a sparse linear combination of training samples by solving a norm-regularized least squares problem. However, the value of regularization parameter is always indiscriminating for the whole dictionary. To enhance the group concentration of the coefficients and also to improve the sparsity, we propose a new SR model called adaptive sparse representation classifier(ASRC). In ASRC, a sparse coefficient strengthened item is added in the objective function. The model is solved by the artificial bee colony (ABC) algorithm with variable step to speed up the convergence. Also, a partition strategy for large scale dictionary is adopted to lighten bee's load and removes the irrelevant groups. Through different data sets, we empirically demonstrate the property of the new model and its recognition performance.

Zhang, Xiaoqiang, Wang, Xuesong, Wang, Qingming.  2018.  Additive Spread Spectrum Image Hiding Algorithm Based on Host Signal. Proceedings of the 2018 7th International Conference on Software and Computer Applications. :164-168.

Image hiding is the important tools to protect the ownership rights of digital multimedia contents. To reduce the interference effect of the host signal in the popular Spread Spectrum (SS) image hiding algorithm, this paper proposes an Improved Additive Spread Spectrum (IASS) image hiding algorithm. The proposed IASS image hiding algorithm maintains the simple decoder of the Additive Spread Spectrum (ASS) image hiding algorithm. This paper makes the comparative experiments with the ASS image hiding algorithm and Correlation-and-bit-Aware Spread Spectrum (CASS) image hiding algorithm. For the noise-free scenario, the proposed IASS image hiding algorithm could yield error-free decoding performance in theory. For the noise scenario, the experimental results show that the proposed IASS image hiding algorithm could significantly reduce the host effect in data hiding and improve the watermark decoding performance remarkably.

Zhao, Pu, Liu, Sijia, Wang, Yanzhi, Lin, Xue.  2018.  An ADMM-Based Universal Framework for Adversarial Attacks on Deep Neural Networks. Proceedings of the 26th ACM International Conference on Multimedia. :1065-1073.

Deep neural networks (DNNs) are known vulnerable to adversarial attacks. That is, adversarial examples, obtained by adding delicately crafted distortions onto original legal inputs, can mislead a DNN to classify them as any target labels. In a successful adversarial attack, the targeted mis-classification should be achieved with the minimal distortion added. In the literature, the added distortions are usually measured by \$L\_0\$, \$L\_1\$, \$L\_2\$, and \$L\_$\backslash$infty \$ norms, namely, L\_0, L\_1, L\_2, and L\_$ınfty$ attacks, respectively. However, there lacks a versatile framework for all types of adversarial attacks. This work for the first time unifies the methods of generating adversarial examples by leveraging ADMM (Alternating Direction Method of Multipliers), an operator splitting optimization approach, such that \$L\_0\$, \$L\_1\$, \$L\_2\$, and \$L\_$\backslash$infty \$ attacks can be effectively implemented by this general framework with little modifications. Comparing with the state-of-the-art attacks in each category, our ADMM-based attacks are so far the strongest, achieving both the 100% attack success rate and the minimal distortion.

Aidan, J. S., Zeenia, Garg, U..  2018.  Advanced Petya Ransomware and Mitigation Strategies. 2018 First International Conference on Secure Cyber Computing and Communication (ICSCCC). :23–28.

In this cyber era, the cyber threats have reached a new level of menace and maturity. One of the major threat in this cyber world nowadays is ransomware attack which had affected millions of computers. Ransomware locks the valuable data with often unbreakable encryption codes making it inaccessible for both organization and consumers, thus demanding heavy ransom to decrypt the data. In this paper, advanced and improved version of the Petya ransomware has been introduced which has a reduced anti-virus detection of 33% which actually was 71% with the original version. System behavior is also monitored during the attack and analysis of this behavior is performed and described. Along with the behavioral analysis two mitigation strategies have also been proposed to defend the systems from the ransomware attack. This multi-layered approach for the security of the system will minimize the rate of infection as cybercriminals continue to refine their tactics, making it difficult for the organization's complacent development.

Brasser, Ferdinand, Davi, Lucas, Dhavlle, Abhijitt, Frassetto, Tommaso, Dinakarrao, Sai Manoj Pudukotai, Rafatirad, Setareh, Sadeghi, Ahmad-Reza, Sasan, Avesta, Sayadi, Hossein, Zeitouni, Shaza et al..  2018.  Advances and Throwbacks in Hardware-assisted Security: Special Session. Proceedings of the International Conference on Compilers, Architecture and Synthesis for Embedded Systems. :15:1–15:10.
Hardware security architectures and primitives are becoming increasingly important in practice providing trust anchors and trusted execution environment to protect modern software systems. Over the past two decades we have witnessed various hardware security solutions and trends from Trusted Platform Modules (TPM), performance counters for security, ARM's TrustZone, and Physically Unclonable Functions (PUFs), to very recent advances such as Intel's Software Guard Extension (SGX). Unfortunately, these solutions are rarely used by third party developers, make strong trust assumptions (including in manufacturers), are too expensive for small constrained devices, do not easily scale, or suffer from information leakage. Academic research has proposed a variety of solutions, in hardware security architectures, these advancements are rarely deployed in practice.
Li, Xu, Zhong, Jinghua, Wu, Xixin, Yu, Jianwei, Liu, Xunying, Meng, Helen.  2020.  Adversarial Attacks on GMM I-Vector Based Speaker Verification Systems. ICASSP 2020 - 2020 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP). :6579—6583.
This work investigates the vulnerability of Gaussian Mixture Model (GMM) i-vector based speaker verification systems to adversarial attacks, and the transferability of adversarial samples crafted from GMM i-vector based systems to x-vector based systems. In detail, we formulate the GMM i-vector system as a scoring function of enrollment and testing utterance pairs. Then we leverage the fast gradient sign method (FGSM) to optimize testing utterances for adversarial samples generation. These adversarial samples are used to attack both GMM i-vector and x-vector systems. We measure the system vulnerability by the degradation of equal error rate and false acceptance rate. Experiment results show that GMM i-vector systems are seriously vulnerable to adversarial attacks, and the crafted adversarial samples are proved to be transferable and pose threats to neural network speaker embedding based systems (e.g. x-vector systems).
Zügner, Daniel, Akbarnejad, Amir, Günnemann, Stephan.  2018.  Adversarial Attacks on Neural Networks for Graph Data. Proceedings of the 24th ACM SIGKDD International Conference on Knowledge Discovery & Data Mining. :2847-2856.
Deep learning models for graphs have achieved strong performance for the task of node classification. Despite their proliferation, currently there is no study of their robustness to adversarial attacks. Yet, in domains where they are likely to be used, e.g. the web, adversaries are common. Can deep learning models for graphs be easily fooled? In this work, we introduce the first study of adversarial attacks on attributed graphs, specifically focusing on models exploiting ideas of graph convolutions. In addition to attacks at test time, we tackle the more challenging class of poisoning/causative attacks, which focus on the training phase of a machine learning model.We generate adversarial perturbations targeting the node's features and the graph structure, thus, taking the dependencies between instances in account. Moreover, we ensure that the perturbations remain unnoticeable by preserving important data characteristics. To cope with the underlying discrete domain we propose an efficient algorithm Nettack exploiting incremental computations. Our experimental study shows that accuracy of node classification significantly drops even when performing only few perturbations. Even more, our attacks are transferable: the learned attacks generalize to other state-of-the-art node classification models and unsupervised approaches, and likewise are successful even when only limited knowledge about the graph is given.