Robust Online Covariance and Sparse Precision Estimation Under Arbitrary Data Corruption. (arXiv:2309.08884v1 [cs.LG])

Abstract: Gaussian graphical models are widely used to represent correlations among entities but remain vulnerable to data corruption. In this work, we introduce a modified trimmed-inner-product algorithm to robustly estimate the covariance in an online scenario even in the presence of arbitrary and adversarial data attacks. At each time step, data points, drawn nominally independently and identically from a multivariate

Read more

Related Posts

• Sample Complexity of Robust Learning against Evasion Attacks. (arXiv:2308.12054v1 [cs.LG])a
• Adversarial training for tabular data with attack propagation. (arXiv:2307.15677v1 [cs.LG])a
• Why do universal adversarial attacks work on large language models?: Geometry might be the answer. (arXiv:2309.00254v1 [cs.LG])a
• Unlearnable Examples Give a False Sense of Security: Piercing through Unexploitable Data with Learnable Examples. (arXiv:2305.09241v3 [cs.LG] UPDATED)a
• It Is All About Data: A Survey on the Effects of Data on Adversarial Robustness. (arXiv:2303.09767v2 [cs.LG] UPDATED)a