Research map / Adversarial robustness and security / Certified and empirical robustness
Self-Healing Robust Neural Networks via Closed-Loop Control
Zhuotong Chen, Qianxiao Li, Zheng Zhang
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 changed | Where | Measured | Effect |
|---|---|---|---|
| add closed-loop self-healing control | standard RN-18 · CIFAR-10 | accuracy under AA l_inf: 0. / 63.89 | Improved |
| add self-healing control | standard RN-18 · CIFAR-10 | accuracy under AA l2: 0. / 82.1 | Improved |
| add self-healing control | standard RN-18 · CIFAR-10 | clean accuracy: 94.71 / 92.81 | Hurt |
| add self-healing control | TRADES RN-18 · CIFAR-10 | accuracy under AA l1: 9.86 / 42.85 | Improved |
| add self-healing control | TRADES RN-18 · CIFAR-10 | clean accuracy: 82.39 / 87.51 | Improved |
| add self-healing control | TRADES WRN-28-10 · CIFAR-100 | accuracy under AA l1: 3.24 / 16.43 | Improved |
| add self-healing control | TRADES WRN-28-10 · CIFAR-100 | clean accuracy: 56.96 / 56.84 | No clear effect |
| add self-healing control | EfficientNet-b0 · Tiny-ImageNet | accuracy under AA l_inf: 0.21 / 46.08 | Improved |
| use optimal embedding function instead of sub-optimal one (ablation) | self-healing RN-18 · CIFAR-10 | accuracy under AA l_inf: 40.29 -> 63.89 | Improved |
| use optimal embedding function instead of sub-optimal one (ablation) | self-healing RN-18 · CIFAR-10 | clean accuracy: 82.97 -> 92.81 | Improved |
Closest papers by meaning
- Self-Healing Machine Learning: A Framework for Autonomous Adaptation in Real-World Environments (NeurIPS 2024 · similarity 0.83)
- Towards Trustworthy Deep Learning (AAAI 2024 · similarity 0.83)
- FEEL-SNN: Robust Spiking Neural Networks with Frequency Encoding and Evolutionary Leak Factor (NeurIPS 2024 · similarity 0.83)
- REGLO: Provable Neural Network Repair for Global Robustness Properties (AAAI 2024 · similarity 0.83)
- Robust Stable Spiking Neural Networks (ICML 2024 · similarity 0.82)
- Conserve-Update-Revise to Cure Generalization and Robustness Trade-off in Adversarial Training (ICLR 2024 · similarity 0.82)
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.
