Adversarial Pattern Classification Using Multiple Classifiers and Randomisation

By Battista Biggio, Giorgio Fumera, Fabio Roli. 2008

70 citations recorded by the source corpus.

Subfields: Artificial Intelligence, Information Systems, Signal Processing

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

In Information Systems

In Signal Processing

Citations in the corpus

Wild Patterns: Ten Years After the Rise of Adversarial Machine Learning (2017, background): “The latter approach has been firstly proposed in 2008 [43] as an effective way to hide information about the classification function to the attacker, with recent follow-ups in [7, 45] to counter adversarial examples.”
Randomized Prediction Games for Adversarial Machine Learning (2016, background): “the attacker has to increase the number of modifications made to the attack samples [28].”
Adversarial Attacks and Defences: A Survey (2018, background): “classifiers Biggio et al. [19], [20] Poisoning based attacks: Crafted training data for Support vector Machines Dalvi et al. [24] Adversarial Classification, Email Spam detection, fraud Biggio et al. [16], [15] Pattern recognition detection, intrusion detection, biometric identification Papernot et ”
Multiple classifier systems for robust classifier design in adversarial environments (2010, methodology): “In [4] we analysed randomisation strategies based on MCSs to prevent an adversary from gaining sufficient knowledge on a classifier to evade it.”
A Theoretical Framework for Robustness of (Deep) Classifiers against Adversarial Examples (2016, methodology): “, 2010) and (Biggio et al., 2008) propose a method to introduce some randomness in the selection of classification boundaries; (2) A few recent studies (Xiao et al.”
Multiple Classifier Systems under Attack (2010, methodology): “In [15] we analysed a randomization strategy based on MCSs to improve robustness by preventing an adversary from gaining too much knowledge on a classifier.”
Scalable Optimization of Randomized Operational Decisions in Adversarial Classification Settings (2015, methodology): “Extensive experiments demonstrate the efficacy of our approach.”
Pattern Recognition Systems under Attack: Design Issues and Research Challenges (2014, background): “We have investigated a specific implementation of the latter approach in [14]; however, it is still an open issue to understand whether randomization may be effective also against attacks that can be exploited to reverse-engineer the classification function [50, 57, 10], and whether and how it can i”
Optimal randomized classification in adversarial settings (2014, background): “[3] (who propose adding noise to the classification boundary) and Colbaugh and Glass [7] (who address randomization among two equally good classifiers), the vast majority of the work considers only deterministic classification schemes.”
Cracking Classifiers for Evasion: A Case Study on the Google's Phishing Pages Filter (2016, background): “Accordingly, some mitigation techniques have been proposed to against evasion attacks by reducing the leakage of exploitable knowledge as much as possible [8][11] or making the learning method more robust to evasion [15][36].”