ATTACKS ON MACHINE LEARNING MODELS

Summary from the TLDR newsletter

This blog attempts to shine a light on the topic of attacks on machine learning models. It aims to inform readers about neural network attacks, a topic worthy of attention for those deploying machine learning models in production. Machine learning models are just as susceptible to attacks as other systems, but they are extra hard to protect as they have both code and data vulnerabilities. Current defenses are not enough, and real fixes may require massive changes to model development. As long as model development involves pattern matching, attacks can never be truly prevented.