Filters








17,847 Hits in 6.9 sec

Precise Statistical Analysis of Classification Accuracies for Adversarial Training [article]

Adel Javanmard, Mahdi Soltanolkotabi
2022 arXiv   pre-print
of the standard and robust accuracy for a class of minimax adversarially trained models.  ...  Our comprehensive analysis allows us to theoretically explain several intriguing empirical phenomena and provide a precise understanding of the role of different problem parameters on standard and robust  ...  [JS20] Adel Javanmard and Mahdi Soltanolkotabi, Supplementary material to "precise statistical analysis of classification accuracies for adversarial training", 2020.  ... 
arXiv:2010.11213v2 fatcat:avkvexobxzb6bkrd2hk4r7t2m4

AdvAndMal: Adversarial Training for Android Malware Detection and Family Classification

Chenyue Wang, Linlin Zhang, Kai Zhao, Xuhui Ding, Xusheng Wang
2021 Symmetry  
of the overall framework for the adversarial training.  ...  malware classification layer is trained by RGB image visualized from the sequence of system calls.  ...  Acknowledgments: The authors would like to thank editors and referees for their precious remarks and comments. Conflicts of Interest: The authors declare no conflict of interest.  ... 
doi:10.3390/sym13061081 fatcat:mewauplxavfdva3pe2z2s6fj6e

Expansion of Cyber Attack Data From Unbalanced Datasets Using Generative Techniques [article]

Ibrahim Yilmaz, Rahat Masum
2019 arXiv   pre-print
We have tested the accuracy of our model with the imbalance dataset initially and then with the increasing the attack samples and found improvement of classification performance for the latter.  ...  Multilayer Perceptron (MLP) technique will provide improvement in accuracy and increase the performance of detecting the attack and benign data from a balanced dataset.  ...  In our analysis, we have used 60% of the dataset as a training set and 40% of the dataset as a test set for the classifier. Accuracy and analysis of this classifier are discussed in Section V-A.  ... 
arXiv:1912.04549v1 fatcat:vqzfsywppvflpcr22idh767hje

CyberPulse: A Machine Learning based Link Flooding Attack Mitigation System for Software Defined Networks

Raihan Ur Rasool, Usman Ashraf, Khandakar Ahmed, Hua Wang, Wajid Rafique, Zahid Anwar
2019 IEEE Access  
CyberPulse was evaluated for its accuracy, false positive rate, and effectiveness as compared to competing approaches on realistic networks generated using Mininet.  ...  Software-defined networking (SDN) offers a novel paradigm for effective network management by decoupling the control plane from the data plane thereby allowing a high level of manageability and programmability  ...  In the next step, the analysis is performed to assess accuracy and training time.  ... 
doi:10.1109/access.2019.2904236 fatcat:m2p3y4ttbngtxffra6vua4nhnq

QuantifyML: How Good is my Machine Learning Model?

Muhammad Usman, Divya Gopinath, Corina S. Păsăreanu
2021 Electronic Proceedings in Theoretical Computer Science  
The efficacy of machine learning models is typically determined by computing their accuracy on test data sets.  ...  QuantifyML enables i) evaluating learnability by comparing the counts for the outputs to ground truth, expressed as logical predicates, ii) comparing the performance of models built with different machine  ...  Quantifying adversarial robustness for image classification models: We trained a decision-tree classifier on the popular MNIST benchmark, which is a collection of handwritten digits classified to one of  ... 
doi:10.4204/eptcs.348.6 fatcat:l3t63j3lcfbdxfw4wzyduu4xj4

An Adversarial Attack Analysis on Malicious Advertisement URL Detection Framework [article]

Ehsan Nowroozi, Abhishek, Mohammadreza Mohammadi, Mauro Conti
2022 arXiv   pre-print
The combination set of six different kinds of features precisely overcome the obfuscation in fraudulent URL classification.  ...  Based on different statistical properties, we use twelve different formatted datasets for detection, prediction and classification task.  ...  We also looked into the vulnerability of various ensembles for adversarial training.  ... 
arXiv:2204.13172v1 fatcat:i4tuwzuijrertcy7s2noflj5ry

ASP:A Fast Adversarial Attack Example Generation Framework based on Adversarial Saliency Prediction [article]

Fuxun Yu, Qide Dong, Xiang Chen
2018 arXiv   pre-print
attack based on the adversarial saliency analysis.  ...  However, with the appearance of the Adversarial Attack, the NN based system performance becomes extremely vulnerable:the image classification results can be arbitrarily misled by the adversarial examples  ...  This model could achieve 99.2% classification accuracy after training 10 epochs.  ... 
arXiv:1802.05763v3 fatcat:m3xw27nh3nesvphrrljbhowhei

Deep Adversarial Learning on Google Home devices [article]

Andrea Ranieri, Davide Caputo, Luca Verderame, Alessio Merlo, Luca Caviglione
2021 arXiv   pre-print
Smart speakers and voice-based virtual assistants are core components for the success of the IoT paradigm.  ...  To cope with that, deep adversarial learning approaches can be used to build black-box countermeasures altering the network traffic (e.g., via packet padding) and its statistical information.  ...  Fig. 4 shows the classification accuracy for ML techniques trained on the original data (not subjected to any adversarial technique) of the Utility/Media/Travel scenario.  ... 
arXiv:2102.13023v1 fatcat:yi3noammfreyjgs5gm5itp3kee

Unsupervised Learning for Trustworthy IoT [article]

Nikhil Banerjee, Thanassis Giannetsos, Emmanouil Panaousis, Clive Cheong Took
2018 arXiv   pre-print
Our initial set of results clearly show that these unsupervised learning algorithms are prone to adversarial infection, thus, magnifying the need for further research in the field by leveraging a mix of  ...  clustering and classification processes.  ...  More precisely, we derive various results that show the accuracy of classification after the system has been re-trained with various levels of adversarial samples. IV.  ... 
arXiv:1805.10401v1 fatcat:62wy3r3txffh5fqqemngl2h25q

Unsupervised Learning for Trustworthy IoT

Nikhil Banerjee, Thanassis Giannetsos, Emmanouil Panaousis, Clive Cheong Took
2018 2018 IEEE International Conference on Fuzzy Systems (FUZZ-IEEE)  
Our initial set of results clearly show that these unsupervised learning algorithms are prone to adversarial infection, thus, magnifying the need for further research in the field by leveraging a mix of  ...  and classification processes.  ...  More precisely, we derive various results that show the accuracy of classification after the system has been re-trained with various levels of adversarial samples. IV.  ... 
doi:10.1109/fuzz-ieee.2018.8491672 dblp:conf/fuzzIEEE/BanerjeeGPT18 fatcat:7eszse5cynethjuw7nedp6tit4

Weighted Average Precision: Adversarial Example Detection in the Visual Perception of Autonomous Vehicles [article]

Yilan Li, Senem Velipasalar
2020 arXiv   pre-print
Several research work in adversarial machine learning started to focus on the detection of AEs in autonomous driving.  ...  We perform evaluation on Berkeley Deep Drive (BDD) and CityScapes datasets to show how our approach outperforms existing single-frame-mAP based AE detections by increasing 17.76% accuracy of performance  ...  Goodfellow et al. first introduced the idea of adversarial training [25] , which try to integrate existing adversarial existing AE generation methods into the training process so that the trained model  ... 
arXiv:2002.03751v2 fatcat:2ovzphkp3rdpfl7ruvw4upqqku

Detection of Tor Traffic using Deep Learning

Debmalya Sarkar, P. Vinod, Suleiman Y. Yerima
2020 2020 IEEE/ACS 17th International Conference on Computer Systems and Applications (AICCSA)  
The system achieved 99.89% accuracy in the classification of Tor and non-Tor traffic on the UNB-CIC Tor network dataset.  ...  Hence, in this paper we present a deep neural network (DNN) based system for the detection and classification of encrypted Tor traffic.  ...  The trained models failed to detect any of the adversarial examples despite being able to classify Tor and non-Tor traffic with very high accuracy.  ... 
doi:10.1109/aiccsa50499.2020.9316533 fatcat:5fvnudnokzhyzejpx6eng4ldfa

Accelerometer-Based Gait Segmentation: Simultaneously User and Adversary Identification [article]

Yujia Ding, Weiqing Gu
2019 arXiv   pre-print
of training data for small data sets.  ...  In particular, the new technology is being applied to cell phone recorded walking data and performs an accuracy of 98.79% for 6 classes classification (user-adversary identification) and 99.06% for binary  ...  or given based on geometric analysis integrated by statistics.  ... 
arXiv:1910.06149v1 fatcat:yrsrhs3x6fet5d2pe4ui6mnshm

Layerwise Perturbation-Based Adversarial Training for Hard Drive Health Degree Prediction [article]

Jianguo Zhang and Ji Wang and Lifang He and Zhao Li and Philip S. Yu
2018 arXiv   pre-print
Our extensive experiments on two real-world hard drive datasets demonstrate the superiority of the proposed schemes for both supervised and semi-supervised classification.  ...  Firstly, we design a layerwise perturbation-based adversarial training method which can add perturbations to any layers of a neural network to improve the generalization of the network.  ...  of datasets TABLE III : III Overall results in supervised setting ST-1 ST-2 Accuracy Precision Recall Macro-F1 Accuracy Precision Recall Macro-F1 DT 82.4 73.6 74.7 74.1 82.2 74.0  ... 
arXiv:1809.04188v4 fatcat:o6ecrsjvefhtxppg5yinmttqxa

Defending Hardware-based Malware Detectors against Adversarial Attacks [article]

Abraham Peedikayil Kuruvila, Shamik Kundu, Kanad Basu
2020 arXiv   pre-print
Our experimental results prove that the proposed defense is able to improve the classification accuracy of HPC traces that have been modified through an adversarial sample generator by up to 31.5  ...  In this paper, we propose a Moving target defense (MTD) for this adversarial attack by designing multiple ML classifiers trained on different sets of HPCs.  ...  Precision represents the proportion of positive classifications that were correct.  ... 
arXiv:2005.03644v2 fatcat:bccoigjhizgsxda4cuqmmm3w4i
« Previous Showing results 1 — 15 out of 17,847 results