Principled approaches to robust machine learning and beyond

48 citations recorded by the source corpus.

How researchers see this paper

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

Stronger data poisoning attacks break data sanitization defenses (2018, methodology): “st larger than across D c, so any ^ outperforming can only have slightly higher loss than ^ across D c. 41 Koh, Steinhardt, and Liang work along this direction (we refer the interested reader to Li (2018) or Steinhardt (2018) for two recent overviews). There appears to be plenty of room both to imp”
Cronus: Robust and Heterogeneous Collaborative Learning with Black-Box Knowledge Transfer (2019, methodology): “The theoretical error bound and sample complexity 1 of the existing robust aggregation algorithms [13], [37], [16], [17], [15], [33] depend on the dimensionality of model parameters.”
Dataset Security for Machine Learning: Data Poisoning, Backdoor Attacks, and Defenses (2020, background): “reader to [169], [170], and [171] for additional references on this line of work.”
Manipulating the Byzantine: Optimizing Model Poisoning Attacks and Defenses for Federated Learning (2021, methodology): “Specifically, these defenses use spectral analysis [16], [15], [28], [26], e.g., singular value decomposition, to detect and ﬁlter outliers in poisoned data.”