2,999 Hits in 2.0 sec

Intrusion Detection using Spatial-Temporal features based on Riemannian Manifold [article]

Amardeep Singh, Julian Jang-Jaccard
2021 arXiv   pre-print
In future research, we will try to use conditional Generative adversarial networks along with tangent space features to detect minority attacks.  ...  Dataset II DescriptionUNSW-NB15 dataset is benchmark dataset that contains nine families of intrusion attacks, namely, Shellcode, Fuzzers, Generic, DoS, Backdoors, Analysis, Generic, Worms and Reconnaissance  ... 
arXiv:2111.00626v1 fatcat:6ozpq2pfjjgqfnoeqdonxsnm6e

Training data requirement for a neural network to predict aerodynamic coefficients

Rajkumar Thirumalainambi, Jorge Bardina, Anthony J. Bell, Mladen V. Wickerhauser, Harold H. Szu
2003 Independent Component Analyses, Wavelets, and Neural Networks  
Basic aerodynamic coefficients are modeled as functions of angle of attack, speed brake deflection angle, Mach number, and side slip angle.  ...  The training data for the neural network are derived from wind tunnel test measurements and numerical simulations.  ...  At lower angles of attack, the first three neural network architectures performance are reasonable in comparison to the 4" neural network architecturc.  ... 
doi:10.1117/12.486343 fatcat:ze6nfnp5rbcylnnb2sw6g334pa


I. V. Zhukovyts'kyy, V. M. Pakhomova, D. O. Ostapets, O. I. Tsyhanok
2020 Nauka ta Progres Transportu  
The first neural network to determine the category of attack class (DoS, R2L, U2R, Probe) or the fact that there was no attack; other neural networksto detect the type of attack, if any (each of these  ...  four neural networks corresponds to one class of attack and is able to identify types that belong only to this class).  ...  General characteristics of the software model.  ... 
doi:10.15802/stp2020/218318 fatcat:kde3tllwvrarvohwj4c6zz42r4

Detecting Simulated Attacks in Computer Networks Using Resilient Propagation Artificial Neural Networks

Mario A. Garcia, Tung Trinh
2015 POLIBITS Research Journal on Computer Science and Computer Engineering With Applications  
Intrusion detection systems using neural networks have been deemed a promising solution to detect such attacks.  ...  Many studies have been done on applying neural networks in intrusion detection systems. This work presents a study of applying resilient propagation neural networks to detect simulated attacks.  ...  In general, the proposed solution does not perform well in detecting all types of attacks.  ... 
doi:10.17562/pb-51-1 fatcat:6c3dmnumsbbirg4or6xi2ym63m

3D Point Cloud Completion with Geometric-Aware Adversarial Augmentation [article]

Mengxi Wu, Hao Huang, Yi Fang
2021 arXiv   pre-print
In contrast to the PGD-k attack, our method generates adversarial samples that keep the geometric features in clean samples and contain few outliers.  ...  We propose a novel approach to generate adversarial samples that benefit both the performance of clean and adversarial samples.  ...  Thus, the adversarial samples generated by GD may have higher attack strength than the ones generated by PMPD and PMCD.  ... 
arXiv:2109.10161v1 fatcat:ynqicl6ubjgsdgf26seuygotnm

DeepLaser: Practical Fault Attack on Deep Neural Networks [article]

Jakub Breier, Xiaolu Hou, Dirmanto Jap, Lei Ma, Shivam Bhasin, and Yang Liu
2018 arXiv   pre-print
In this paper, we initiate the first study of leveraging physical fault injection attacks on Deep Neural Networks (DNNs), by using laser injection technique on embedded systems.  ...  ., malicious faults and attacks become a tremendous concern, which potentially could lead to catastrophic consequences.  ...  Studying these functions under fault attacks allows to derive general conclusions on susceptibility of deep learning to fault attacks.  ... 
arXiv:1806.05859v2 fatcat:2puc4lxqybbsjpr3nsvlfvhthu

Truth or Backpropaganda? An Empirical Investigation of Deep Learning Theory [article]

Micah Goldblum, Jonas Geiping, Avi Schwarzschild, Michael Moeller, Tom Goldstein
2020 arXiv   pre-print
not optimal for generalization; (3) demonstrate that ResNets do not conform to wide-network theories, such as the neural tangent kernel, and that the interaction between skip connections and batch normalization  ...  plays a role; (4) find that rank does not correlate with generalization or robustness in a practical setting.  ...  . • Neural tangent kernels and the wide-network limit: We investigate theoretical results concerning neural tangent kernels of realistic architectures.  ... 
arXiv:1910.00359v3 fatcat:oas2iunoyfantiepiklcz5pude

A new intrusion detection and alarm correlation technology based on neural network

Yansong Liu, Li Zhu
2019 EURASIP Journal on Wireless Communications and Networking  
on artificial neural network.  ...  unknown attacks.  ...  For the attacks that do not appear in the training set, there is a high detection rate, which indicates that the artificial neural network-based intrusion detection can detect the unknown attack and the  ... 
doi:10.1186/s13638-019-1419-z fatcat:sphztuhc2fgxxhhrseg5iu2zsy

Convolutional Neural Networks Applied to Human Face Classification

Brian Cheung
2012 2012 11th International Conference on Machine Learning and Applications  
We trained a convolutional neural network to distinguish between images of human faces from computer generated avatars as part of the ICMLA 2012 Face Recognition Challenge.  ...  Convolutional neural network models have covered a broad scope of computer vision applications, achieving competitive performance with minimal domain knowledge.  ...  Such performance far exceeds brute force attack and would enable an attacker to bypass the Avatar CAPTCHA with far fewer expected attempts.  ... 
doi:10.1109/icmla.2012.177 dblp:conf/icmla/Cheung12 fatcat:cwskodp2vzgptegc6iwcbunwdu

Finding Dynamics Preserving Adversarial Winning Tickets [article]

Xupeng Shi, Pengfei Zheng, A. Adam Ding, Yuan Gao, Weizhong Zhang
2022 arXiv   pre-print
Modern deep neural networks (DNNs) are vulnerable to adversarial attacks and adversarial training has been shown to be a promising method for improving the adversarial robustness of DNNs.  ...  Based on recent works of Neural Tangent Kernel (NTK), we systematically study the dynamics of adversarial training and prove the existence of trainable sparse sub-network at initialization which can be  ...  All experiments are performed in JAX (Bradbury et al., 2018) , together with the neural-tangent library (Novak et al., 2020) .  ... 
arXiv:2202.06488v3 fatcat:lzewk5c5pzdd5kn73a7lj6jtiu

Classification Agent-Based Techniques for Detecting Intrusions in Databases [chapter]

Cristian Pinzón, Yanira De Paz, Rosa Cano
2008 Lecture Notes in Computer Science  
The agent incorporates a Case-based reasoning mechanism whose main characteristic involves a mixture of neural networks that carry out the task of filtering attacks.  ...  Moreover, the approaches based on models for detecting SQL injection attacks are very sensitive. With only slight variations of accuracy, they generate a large number of false positive and negatives.  ...  To check the validity of the proposed model, a series of test were elaborated which were executed on a memory of cases, specifically developed for these tests, and which generated attack consults.  ... 
doi:10.1007/978-3-540-87656-4_7 fatcat:rdqclh7qhnhipiyjhksht3qtrq

Contribution of wavelets to cybersecurity: Intrusion detection systems using neural networks

Saiida Lazaar
2021 General Letters in Mathematics  
Wavelets combined to neural networks can be useful for modelling intrusion detection with the main challenges to reduce the false alarms, increase the test accuracy and increase novel attacks detection  ...  In this work, we present a major contribution in the research field to better understand how wavelets and neural networks can be combined for modelling efficient IDS.  ...  considered as a unit which is generally expressed by an activation function (sigmoid function, etc.).  ... 
doi:10.31559/glm2021.10.2.2 fatcat:oe2px6kgbfbqtaxct77kiuzzeq

Neural Network Assisted Inverse Dynamic Guidance for Terminally Constrained Entry Flight

Hao Zhou, Tawfiqur Rahman, Wanchun Chen
2014 The Scientific World Journal  
Applying this approximation, an inverse dynamic system for an entry flight is solved to generate guidance command.  ...  Results from simulations indicate improved performance of the neural network assisted method.  ...  Neural Network Polynomial for Terminal Velocity Control. The polynomial expression of 2 with respect to V . was generated during parameter optimization.  ... 
doi:10.1155/2014/686040 pmid:24723821 pmcid:PMC3958667 fatcat:j6iq34wdyrak5hpbkdhu3hkvg4

Key Feature Recognition Algorithm of Network Intrusion Signal Based on Neural Network and Support Vector Machine

Kai Ye
2019 Symmetry  
Therefore, a key feature recognition algorithm for network intrusion signals based on neural network and support vector machine is proposed.  ...  The experimental results show that the algorithm has the advantages of high precision, low false positive rate and the recognition time of key features of R2L (it is a common way of network intrusion attack  ...  For the identification of U2R and R2L attacks, BP network uses 41-50-40-3 network structure, the hidden layer uses S-type transfer function with tangent property, the output layer uses S-type For the  ... 
doi:10.3390/sym11030380 fatcat:qtes2lkntfc2xawkelhkomsxiy

Robustness of Generalized Learning Vector Quantization Models against Adversarial Attacks [article]

Sascha Saralajew and Lars Holdijk and Maike Rees and Thomas Villmann
2019 arXiv   pre-print
The evaluation suggests that both Generalized LVQ and Generalized Tangent LVQ have a high base robustness, on par with the current state-of-the-art in robust neural network methods.  ...  We therefore present an extensive evaluation of three LVQ models: Generalized LVQ, Generalized Matrix LVQ and Generalized Tangent LVQ.  ...  This changes if the squared Euclidean distance in GLVQ is replaced by adaptive dissimilarity measures such as in Generalized Matrix LVQ (GMLVQ) [8] or Generalized Tangent LVQ (GTLVQ) [9] .  ... 
arXiv:1902.00577v2 fatcat:j3gviumyyvffdmm2abchvrr6pe
« Previous Showing results 1 — 15 out of 2,999 results