Bounding an Attack ’ s Complexity for a Simple Learning Model

By Blaine Nelson, Anthony D. Joseph. 2006

How researchers see this paper

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

In Statistical Learning Theory

Citations in the corpus

Towards the Science of Security and Privacy in Machine Learning (2016, background): “Thus, all the attacks below bound the adversary in their attacks [42].”
Security analysis of online centroid anomaly detection (2010, background): “The poisoning attack against online centroid anomaly detection has been considered by Nelson and Joseph (2006) for the case of infinite window, i.e., when a learning algorithm memorizes all data seen so far.”
SoK: Security and Privacy in Machine Learning (2018, background): “Thus, all the attacks below bound the adversary in their attacks [35].”
Evaluation and Design of Robust Neural Network Defenses (2018, background): “Nelson and Joseph [95] study the problem of poisoning in the space of a anomaly detection system.”
Poisoning Adaptive Biometric Systems (2012, background): “The analysis of [4, 6] was made under the worst-case assumption that the attacker perfectly knows the targeted system, which is typical in security problems.”
A Game Theoretical Model for Adversarial Learning (2009, background): “As a result, a new subfield of “adversarial learning” has emerged to understand and design classifier which are robust to adversarial transformation[1], [2], [3], [4], [5], [6], [7].”
Adversarial machine learning (2019, methodology): “If the past is wholly incorporated into the next generation of model, the advantage seems to lie with the learning model as we explored in prior work with retraining hypersphere anomaly detectors [49].”
Neural network Trojan (2013, unknown): “References dealing with such analysis is given in [23,24,35].”
A framework for quantitative security analysis of machine learning (2009, background): “Obviously, most closely related is the work of Nelson and Joseph [7] who also addressed a poisoning attack against online anomaly detection.”
Online Anomaly Detection under Adversarial Impact (2010, background): “Experimental evaluation carried out on real HTTP and exploit traces confirms the tightness of our theoretical bounds.”
Open problems in the security of learning (2008, background): “Nelson and Joseph [20] bound the minimum number of examples an adversary must submit to a hypersphere-based anomaly detector in order to shift the hypersphere s center.”