Detecting adversarial advertisements in the wild

By David Sculley, Matthew Eric Otey, Michael Pohl, Bridget Spitznagel, John Hainsworth, Yunkai Zhou. 2011

110 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

Practical Evasion of a Learning-Based Classifier: A Case Study (2014, background): “From an adversarial viewpoint, the more information about a learning-based system is available, the higher the chances become that the system can be successfully gamed.”
Evading Classifiers by Morphing in the Dark (2017, background): “rating adversarial examples with great diversity, further enhancing eectiveness of the adversarial training. 7. RELATED WORK Evasion attacks against classiers have been studied in numerous contexts [36, 23, 30, 35]. These works dier in their assumptions on the adversary’s knowledge about the detecto”
Towards Making Systems Forget with Machine Unlearning (2015, background): “They are widely used in many areas of science [25, 35, 37, 46, 55, 63–65].”
Systematic Poisoning Attacks on and Defenses for Machine Learning in Healthcare (2015, background): “These applications include healthcare [1], network intrusion detection systems[2], spam and fraud detection, phishing detection [3], political decision making [4], adversarial advertisement detection [5], and financial engineering [6].”
On the hardness of evading combinations of linear classifiers (2013, background): “Unions and intersections of half-spaces are also a practical way to build real-world classification systems, as demonstrated by Google’s use of a combination of linear classifiers to detect malicious advertisements [16].”
Cyberspace Security Using Adversarial Learning and Conformal Prediction (2015, background): “Detecting adversarial advertisements in the wild [9] expands on the adversarial aspect.”
Hidden Technical Debt in Machine Learning Systems (2015, background): “However, in many cases ensembles work well because the errors in the component models are uncorrelated.”
Machine Learning: The High Interest Credit Card of Technical Debt (2014, methodology): “This approach is useful in situations such as [8], in which sub-problems decompose n aturally, or in which the cost of maintaining separate models is outweighed by the benefits of enfo rced modularity.”
Predicting accurate probabilities with a ranking loss (2012, background): “mizes ranking performance; the idea is hinted at in (Sculley et al., 2011), but not discussed formally.”