All slides

Investigating Imperceptibility of Adversarial Attacks on Tabular Data: An Empirical Analysis

ADSN Conference 2024. Dec 2024, Perth, Australia

Poster proposing seven properties that define an imperceptible adversarial attack on tabular data (proximity, sparsity, deviation, sensitivity, immutability, feasibility and feature interdependency), metrics for each, and an empirical evaluation of five attack methods against them. With Chun Ouyang, Laith Alzubaidi, Alistair Barros and Catarina Moreira.

If the poster does not render here, use Open PDF above.