AI Paper Atlas

Research map / Adversarial robustness and security / Black-box and transfer attacks

On the Perils of Cascading Robust Classifiers

Ravi Mangal, Zifan Wang, Chi Zhang, Klas Leino, Corina Pasareanu, Matt Fredrikson

ICLR 2023 · Adversarial robustness and security › Black-box and transfer attacks · approach: Robustness benchmarks and defenses · 0 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
Apply CasA to the cascading ensembleMNIST ℓ∞ ε=0.1 Small Exact cascade · MNIST ℓ∞ ε=0.1FPR (%): 88.71Hurt
Apply CasA to the cascading ensembleMNIST ℓ∞ ε=0.1 Small Exact cascade · MNIST ℓ∞ ε=0.1ERA (%): 11.17Hurt
Use cascading instead of the single best constituentCIFAR-10 ℓ∞ ε=2/255 Large · CIFAR-10 ℓ∞ ε=2/255ERA (%): 52.65 -> 58.15Hurt
Use sound weighted voting instead of the single modelMNIST ℓ∞ ε=0.1 Small · MNIST ℓ∞ ε=0.1CRA (%): 94.94 -> 94.94No clear effect
Use uniform voting instead of the single modelMNIST ℓ∞ ε=0.1 Small · MNIST ℓ∞ ε=0.1CRA (%): 94.94 -> 9.29Hurt
Use uniform voting instead of the single modelCIFAR-10 ℓ∞ ε=2/255 Small · CIFAR-10 ℓ∞ ε=2/255CRA (%): 46.43 -> 18.58Hurt
Replace dual-network attack objectives with surrogate objectives (ablation)CasA on MNIST ℓ∞ ε=0.3 Small cascade · MNIST ℓ∞ ε=0.3ERA (%): 4.93 -> 7.67Mixed
Replace dual-network attack objectives with surrogate objectives (ablation)CasA on CIFAR-10 ℓ∞ ε=2/255 Small cascade · CIFAR-10 ℓ∞ ε=2/255ERA (%): 23.79 -> 22.92Mixed
Train constituents non-sequentially on the full training dataset (ablation)Uniform voting ensemble · CIFAR-10 ℓ∞ ε=2/255 SmallCRA (%): 18.58 -> 43.48Mixed
Train constituents non-sequentially on the full training dataset (ablation)Uniform voting ensemble · CIFAR-10 ℓ∞ ε=2/255 SmallCRA (%): 46.43 -> 43.48Hurt
Use weighted voting with learned weightsNon-sequentially trained ensemble · CIFAR-10 ℓ∞ ε=2/255 SmallCRA (%): 46.43 -> 42.35Hurt

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy