Machine learning is vulnerable to a wide variety of attacks

This paper looks at how threat actors can manipulate and poison machine learning training, to twist the systems to their goals. The goals are achieved without impacting the integrity of the underlying data sets.

Manipulating SGD with Data Ordering Attacks (arxiv.org)

References

Shumailov, I., Shumaylov, Z., Kazhdan, D., Zhao, Y., Papernot, N., Erdogdu, M. A., & Anderson, R. J. (2021). Manipulating sgd with data ordering attacks. Advances in Neural Information Processing Systems34, 18021-18032.

Further Reading

Data ordering attacks | Light Blue Touchpaper

Machine learning – Wikipedia

machine learning 200