Filters








1,891 Hits in 4.1 sec

Stealing Links from Graph Neural Networks [article]

Xinlei He and Jinyuan Jia and Michael Backes and Neil Zhenqiang Gong and Yang Zhang
2020 arXiv   pre-print
Recently, neural networks were extended to graph data, which are known as graph neural networks (GNNs).  ...  In this work, we propose the first attacks to steal a graph from the outputs of a GNN model that is trained on the graph.  ...  Graph Neural Networks Many important real-world datasets come in the form of graphs or networks, e.g., social networks, knowledge graph, and chemical networks.  ... 
arXiv:2005.02131v2 fatcat:7nmav57hbrf53kwphte4qiooa4

Quantifying (Hyper) Parameter Leakage in Machine Learning [article]

Vasisht Duddu, D. Vijay Rao
2020 arXiv   pre-print
Specifically, we use Bayesian Networks to capture uncertainty in estimating the target model under various extraction attacks based on the subjective notion of probability.  ...  This provides a practical tool to infer actionable details about extracting blackbox models and help identify the best attack combination which maximises the knowledge extracted (or information leaked) from  ...  Typically, deeper the Neural Network the higher the performance due to which the ML community has focuses on scaling Neural Networks to large number of layers [16] [22] . • Layer Type.  ... 
arXiv:1910.14409v2 fatcat:kz2e5l37pbcj3gv7au65kndtjy

10 Security and Privacy Problems in Self-Supervised Learning [article]

Jinyuan Jia, Hongbin Liu, Neil Zhenqiang Gong
2021 arXiv   pre-print
neural networks [29] .  ...  , an attacker could try to reconstruct the remaining links in the graph.  ... 
arXiv:2110.15444v2 fatcat:mroo7j7dhvgf5cymtugmmulsxa

Adversarial Model Extraction on Graph Neural Networks [article]

David DeFazio, Arti Ramesh
2019 arXiv   pre-print
Graph Neural Networks (GNNs) are a popular deep learning framework to perform machine learning tasks over relational data.  ...  Along with the advent of deep neural networks came various methods of exploitation, such as fooling the classifier or contaminating its training data.  ...  Unlike neural network extraction on iid data, we need to modify the underlying graph in order to obtain new label information from the victim.  ... 
arXiv:1912.07721v1 fatcat:764mkqyql5dctchokxsjfdpr7e

Neural Datalog Through Time: Informed Temporal Modeling via Logical Specification [article]

Hongyuan Mei and Guanghui Qin and Minjie Xu and Jason Eisner
2020 arXiv   pre-print
Rules serve to prove facts from other facts and from past events.  ...  In both synthetic and real-world domains, we show that neural probabilistic models derived from concise Datalog programs improve prediction by encoding appropriate domain knowledge in their architecture  ...  We thank Karan Uppal, Songyun Duan and Yujie Zha from Bloomberg L.P. for helpful comments and support to apply the framework to Bloomberg's real-world data.  ... 
arXiv:2006.16723v2 fatcat:fzrz5xnubvdatinnf4uyhxxesy

LPGNet: Link Private Graph Networks for Node Classification [article]

Aashish Kolluri, Teodora Baluta, Bryan Hooi, Prateek Saxena
2022 arXiv   pre-print
Graph convolutional networks (GCNs) are one such widely studied neural network architecture that perform well on this task.  ...  However, powerful link-stealing attacks on GCNs have recently shown that even with black-box access to the trained model, inferring which links (or edges) are present in the training graph is practical  ...  We are grateful for the constructive feedback from the anonymous reviewers and for conducting the revision procedure. We thank Kunwar Preet Singh for his help while revising the paper.  ... 
arXiv:2205.03105v2 fatcat:r76pwrmcsrckdbnjpqky2rj4zy

I Know What You Trained Last Summer: A Survey on Stealing Machine Learning Models and Defences [article]

Daryna Oliynyk, Rudolf Mayer, Andreas Rauber
2022 arXiv   pre-print
We address this by categorising and comparing model stealing attacks, assessing their performance, and exploring corresponding defence techniques in different settings.  ...  This raises the necessity for a thorough systematisation of the field of model stealing, to arrive at a comprehensive understanding why these attacks are successful, and how they could be holistically  ...  If authors claim their attack to be a behavior stealing attack but indeed provide a method for stealing parameters, we define their goal as parameter stealing in Table 3 .  ... 
arXiv:2206.08451v1 fatcat:umkkakjnzbgefm6fr3z6iusegu

Temporal Patterns Discovery of Evolving Graphs for Graph Neural Network (GNN)-based Anomaly Detection in Heterogeneous Networks

Jongmo Kim, Kunyoung Kim, Gi-Yoon Jeon, Mye M. Sohn
2022 Journal of Internet Services and Information Security  
with Graph Neural Networks (GNN).  ...  As a result, we can obtain an evolving graph that is simplified and temporal patternsenhanced from original networks. It is used an input graph for a GNN-based anomaly detection model.  ...  Acknowledgements Temporal patterns discovery of evolving graphs for GNN-based anomaly detection J. Kim, K. Kim, G. Jeon, and M.  ... 
doi:10.22667/jisis.2022.02.28.072 dblp:journals/jisis/KimKJS22 fatcat:uncpemjenbgavni65vrnhzszzy

Graph Neural Network-based Android Malware Classification with Jumping Knowledge [article]

Wai Weng Lo, Siamak Layeghy, Mohanad Sarhan, Marcus Gallagher, Marius Portmann
2022 arXiv   pre-print
This paper presents a new Android malware detection method based on Graph Neural Networks (GNNs) with Jumping-Knowledge (JK).  ...  Android function call graphs (FCGs) consist of a set of program functions and their inter-procedural calls.  ...  Graph Neural Networks Convolutional Neural Networks (CNNs) have been very successfully applied to the image classification problem. However, CNNs cannot be applied to non-Euclidean data structures.  ... 
arXiv:2201.07537v9 fatcat:cubrowiqjbffhgzujdxa2452ue

Phishing Website Detection using Supervised Deep Learning

2019 International journal of recent technology and engineering  
We can also increase the accuracy of our algorithm by adding certain more features and increasing the hidden layers in neural networks.  ...  In this algorithmic approach to detect genuine websites a feature set is used so by analyzing these features using deep neural networks we can detect a website is phished or not.  ...  Fig 2: Architecture of deep neural network An artificial neural network, or neural network, is a mathematical model inspired by biological neural networks.  ... 
doi:10.35940/ijrte.d8016.118419 fatcat:mr2nydaz55f27nc2vlklp3z2ba

Detecting Phishing Sites – An Overview [article]

P.Kalaharsha
2021 arXiv   pre-print
In phishing, attackers lure end-users and steal their personal in-formation. To minimize the damage caused by phishing must be detected as early as possible.  ...  Recurrent Neural Networks A class of artificial neural networks where connections between nodes form a directed graph along a temporal sequence is recurrent neural network (RNN).  ...  LSTM has feedback links, unlike normal feedforward neural networks.  ... 
arXiv:2103.12739v2 fatcat:yrmoe323pvcnfjq5ho7tkcinde

Node-Level Membership Inference Attacks Against Graph Neural Networks [article]

Xinlei He and Rui Wen and Yixin Wu and Michael Backes and Yun Shen and Yang Zhang
2021 arXiv   pre-print
To fully utilize the information contained in graph data, a new family of machine learning (ML) models, namely graph neural networks (GNNs), has been introduced.  ...  Many real-world data comes in the form of graphs, such as social networks and protein structure.  ...  These models, such as convolutional neural networks (CNNs) and recurrent neural networks (RNNs), are designed to extract fine-grained representation for each data sample from its own feature.  ... 
arXiv:2102.05429v1 fatcat:45kowsfzrbdovilnm2qpywtyhi

Membership Inference Attack on Graph Neural Networks [article]

Iyiola E. Olatunji, Wolfgang Nejdl, Megha Khosla
2021 arXiv   pre-print
Graph Neural Networks (GNNs), which generalize traditional deep neural networks on graph data, have achieved state-of-the-art performance on several graph analytical tasks.  ...  Introduction Graph neural networks (GNNs) have gained substantial attention from academia and industry in the past few years with high-impact applications ranging from the analysis of social networks,  ...  We will publish our code at the time of publication. 2 Background and Related Works Graph Neural Networks Graph Neural Networks popularized by graph convolutional networks (GCNs) and their variants,  ... 
arXiv:2101.06570v3 fatcat:czknpvcdsvdwtkdlcrbk37dvdm

UNTANGLE: Unlocking Routing and Logic Obfuscation Using Graph Neural Networks-based Link Prediction [article]

Lilas Alrahis, Satwik Patnaik, Muhammad Abdullah Hanif, Muhammad Shafique, Ozgur Sinanoglu
2021 arXiv   pre-print
Hence, UNTANGLE can infer the hidden timing paths by learning the composition of gates in the observed locked netlist or a circuit library leveraging graph neural networks.  ...  We focus on the latter since point function-based locking suffers from various structural vulnerabilities.  ...  ., the outputs from the previous SwB stage. Observed net Fig. 7 . Link prediction using graph neural networks (GNNs) (based on [38] ). C.  ... 
arXiv:2111.07062v1 fatcat:vckouaqgvrgjllhb74afdhejhi

Analysis of Machine Learning Algorithms to Protect from Phishing in Web Data Mining

N. Swapna
2017 International Journal of Computer Applications  
Our research work presents big data analytics that aims to prevent malicious email notifications & phishing from web service  ...  The neural networks used to predict the phishing websites multilayer neural networks shrink the error and elevate the performance conducted in the survey on phishing attacks.  ...  Data mining techniques like neural networks rule induction decision tree machine learning can be useful to the fuzzy logic model.  ... 
doi:10.5120/ijca2017912743 fatcat:nt4vh7gdazdqxa4r7u5ddqafa4
« Previous Showing results 1 — 15 out of 1,891 results