Filters








45 Hits in 1.8 sec

Adversarially Robust Classification based on GLRT [article]

Bhagyashree Puranik, Upamanyu Madhow, Ramtin Pedarsani
2020 arXiv   pre-print
In this paper, we explore, in the setting of classical composite hypothesis testing, a defense strategy based on the generalized likelihood ratio test (GLRT), which jointly estimates the class of interest  ...  We show that the GLRT approach yields performance competitive with that of the minimax approach under the worst-case attack, and observe that it yields a better robustness-accuracy trade-off under weaker  ...  of the uncertainty sets Θ k depends on the constraints on the adversary.  ... 
arXiv:2011.07835v1 fatcat:wysgs7c745gx7jiyljmkvxrf7u

Generalized Likelihood Ratio Test for Adversarially Robust Hypothesis Testing [article]

Bhagyashree Puranik, Upamanyu Madhow, Ramtin Pedarsani
2021 arXiv   pre-print
We interpret an adversarial perturbation as a nuisance parameter, and propose a defense based on applying the generalized likelihood ratio test (GLRT) to the resulting composite hypothesis testing problem  ...  For non-asymptotic regimes, we show via simulations that the GLRT defense is competitive with the minimax approach under the worst-case attack, while yielding a better robustness-accuracy tradeoff under  ...  Mittal, “Lower bounds on adversarial sification Based on GLRT,” in ICASSP 2021-2021 IEEE International robustness from optimal transport,” in Advances in Neural Information Conference  ... 
arXiv:2112.02209v1 fatcat:7lh3gixnd5e7zh3b44qnwiyxdy

Locally optimal detection of stochastic targeted universal adversarial perturbations [article]

Amish Goel, Pierre Moulin
2020 arXiv   pre-print
In this paper, we derive the locally optimal generalized likelihood ratio test (LO-GLRT) based detector for detecting stochastic targeted universal adversarial perturbations (UAPs) of the classifier inputs  ...  We also describe a supervised training method to learn the detector's parameters, and demonstrate better performance of the detector compared to other detection methods on several popular image classification  ...  We then evaluate the detector on three key metrics and show that LO-GLRT detector achieves better performance than PRN and PCA based detectors on all the metrics.  ... 
arXiv:2012.04692v1 fatcat:h2ovqr2jpjfc5ajn3ncdsrys7u

Fast Locally Optimal Detection of Targeted Universal Adversarial Perturbations

Amish Goel, Pierre Moulin
2022 IEEE Transactions on Information Forensics and Security  
This paper proposes a locally-optimal generalized likelihood ratio test (LO-GLRT) for detecting targeted attacks on a classifier, where the attacks add a norm-bounded targeted universal adversarial perturbation  ...  The LO-GLRT outperforms the PRN detector on both counts, with a running time at least 100 times lower than that of the PRN detector.  ...  Defenses based on robustifying a classifier include [30] and [31] . We proposed a LO-GLRT for the detection of adversarial inputs in [15] .  ... 
doi:10.1109/tifs.2022.3169922 fatcat:uobgfosckjfppptsdbzxilnyie

Learning to Detect with Constant False Alarm Rate [article]

Tzvi Diskin, Uri Okun, Ami Wiesel
2022 arXiv   pre-print
We consider the use of machine learning for hypothesis testing with an emphasis on target detection. Classical model-based solutions rely on comparing likelihoods.  ...  In contrast, data-driven machine learning is often more robust and yields classifiers with fixed computational complexity.  ...  The architecture is based on four non-linear features: the sample mean of x, its sample variance and robust versions of the two based on the median.  ... 
arXiv:2206.05747v1 fatcat:p7htc2vvljbsjpvh3nax5relta

Detecting DGA domains with recurrent neural networks and side information [article]

Ryan R. Curtin, Andrew B. Gardner, Slawomir Grzonkowski, Alexey Kleymenov, Alejandro Mosquera
2019 arXiv   pre-print
This is a reasonable baseline for classification using both the domain name and the side information (WHOIS features). • glrt-lstm: a GLRT LSTM model built only on the full domain name (no side information  ...  These adversarial attacks have been successfully applied to fields outside of images, including audio [8] and malware classification [18] .  ... 
arXiv:1810.02023v2 fatcat:eaixlavwxra3tmeepwnvajv6ui

CFARnet: deep learning for target detection with constant false alarm rate [article]

Tzvi Diskin, Yiftach Beer, Uri Okun, Ami Wiesel
2022 arXiv   pre-print
In contrast, data-driven machine learning is often more robust and yields classifiers with fixed computational complexity.  ...  Classical model-based solutions to composite hypothesis testing are sensitive to imperfect models and are often computationally expensive.  ...  The architecture is based on four non-linear features: the sample mean of x, its sample variance and robust versions of the two based on the median.  ... 
arXiv:2208.02474v1 fatcat:jqjfh2zs45gyjobbdshr7qjskq

Table of contents

2019 IEEE Geoscience and Remote Sensing Letters  
Dewen, and C About the Cover: A parameterized model capable of estimating residual motion error (RME) was developed based on the geometry of a single-baseline interferogram.  ...  GLRT Detection of Micromotion Targets for the Multichannel SAR-GMTI System .............. W. Zhang and Y.  ... 
doi:10.1109/lgrs.2018.2886732 fatcat:hbivgb5ovfdjphz7qwmbmykus4

Papertitles

2019 2019 International Conference on Control, Automation and Information Sciences (ICCAIS)  
Radars for Multi-Target Tracking Robust Distributed Sonar CFAR Detection Based on Modified VI-CFAR Detector Robust GLRT Detection Exploiting Persymmetry in Partially Homogeneous Environments Robust H∞  ...  on Generative Adversarial Nets in Through-wall Radar Imaging Target Localization in CEMS Based on Shunt-Wound Radial Basis Function Network The Outlier Elimination Methodology of Ultra-Short BaseLine  ... 
doi:10.1109/iccais46528.2019.9074559 fatcat:srkln5llk5czjgcksdtuhlgwdy

A Survey of Blind Modulation Classification Techniques for OFDM Signals

Anand Kumar, Sudhan Majhi, Guan Gui, Hsiao-Chun Wu, Chau Yuen
2022 Sensors  
We focus on the most widely used statistical and machine learning (ML) models and emphasize their advantages and limitations.  ...  Blind modulation classification (MC) is an integral part of designing an adaptive or intelligent transceiver for future wireless communications.  ...  Furthermore, CNN-based MC techniques are robust to prediction errors on carrier phase offset and SNR.  ... 
doi:10.3390/s22031020 pmid:35161766 pmcid:PMC8840120 fatcat:aebc3apjtbbelaia6nhtjyu5sq

RadarConf21 2021 Blank Page

2021 2021 IEEE Radar Conference (RadarConf21)  
Robust Adaptive Beamforming Based on the Direct Biconvex Optimization Modeling by Xiny- ing Zou, Qiping Zhang, Weijian Zhang, Jinfeng Hu 4.  ...  Distributed GLRT-Based Detection of Target in SIRP Clutter and Noise by Batu Chalise, Kevin Wagner 3.  ... 
doi:10.1109/radarconf2147009.2021.9455240 fatcat:mfgpxueblfdvtli4uh6vyfgw4m

Provable tradeoffs in adversarially robust classification [article]

Edgar Dobriban, Hamed Hassani, David Hong, Alexander Robey
2022 arXiv   pre-print
We derive exact and approximate Bayes-optimal robust classifiers for the important setting of two- and three-class Gaussian classification problems with arbitrary imbalance, for ℓ_2 and ℓ_∞ adversaries  ...  We develop and leverage new tools, including recent breakthroughs from probability theory on robust isoperimetry, which, to our knowledge, have not yet been used in the area.  ...  The approach is based on the generalized likelihood ratio test (GLRT) and can be applied to general multi-class Gaussian settings.  ... 
arXiv:2006.05161v5 fatcat:ic3unrv27vaerfofok4d4jnpbi

2020 Index IEEE Signal Processing Letters Vol. 27

2020 IEEE Signal Processing Letters  
Fonseca, E., +, LSP 2020 1235-1239 Amphibian Sounds Generating Network Based on Adversarial Learning.  ...  Jung, A., +, LSP 2020 825-829 Fourier analysis Amphibian Sounds Generating Network Based on Adversarial Learning.  ... 
doi:10.1109/lsp.2021.3055468 fatcat:wfdtkv6fmngihjdqultujzv4by

2020 Index IEEE Transactions on Information Forensics and Security Vol. 15

2020 IEEE Transactions on Information Forensics and Security  
., +, TIFS 2020 2514-2527 Fourier transforms A Robust Approach for Securing Audio Classification Against Adversarial Attacks.  ...  Hua, G., +, TIFS 2020 1868-1878 Audio signal processing A Robust Approach for Securing Audio Classification Against Adversarial Attacks.  ...  G Gait analysis Deep Learning-Based Gait Recognition Using Smartphones in the Wild.  ... 
doi:10.1109/tifs.2021.3053735 fatcat:eforexmnczeqzdj3sc2j4yoige

Table of Contents

2020 IEEE Signal Processing Letters  
Kang 446 Amphibian Sounds Generating Network Based on Adversarial Learning . . . . . S. Park, M. Elhilali, D. K. Han, and H.  ...  Xu 655 Multi-Scale Shape Index-Based Local Binary Patterns for Texture Classification .. . . . . . . . . . N. Alpaslan and K.  ...  Zou 775 Adversarial Transfer Learning for Deep Learning Based Automatic Modulation Classification . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .  ... 
doi:10.1109/lsp.2020.3040844 fatcat:xpovskhrvfgctk3hhufuvpyyne
« Previous Showing results 1 — 15 out of 45 results