Research map / Label-efficient and robust learning / Recognition and long-tail
DISCO: Dynamic and Invariant Sensitive Channel Obfuscation for Deep Neural Networks
Abhishek Singh, Ayush Chopra, Ethan Garza, Emily Zhang, Praneeth Vepakomma, Vivek Sharma, Ramesh Raskar
Check 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 |
|---|---|---|---|
| Use dynamic channel pruning to obfuscate sensitive activations | DISCO · CIFAR-10 sensitive attribute leakage | sensitive-attribute adversary accuracy: 0.3573 -> 0.2282 | Improved |
| Use dynamic channel pruning to obfuscate sensitive activations | DISCO · CelebA reconstruction | SSIM: 0.68 -> 0.38 | Improved |
| Use dynamic channel pruning to obfuscate sensitive activations | DISCO · CelebA reconstruction | PSNR: 20.49 -> 11.61 | Improved |
| Add the spatial-decoupling pre-processing module (ablation) | DISCO · Sensitive-input reconstruction | reconstruction quality | Improved |
| Turn off the pre-processing module (ablation) | DISCO · Sensitive-input reconstruction with supervised decoder | reconstruction quality | Mixed |
| Use Gaussian activation noise strong enough to prevent reconstruction | Collaborative inference · Sensitive-input reconstruction | task utility | Hurt |
| Protect Big Nose instead of Mouth Open while predicting Smiling | DISCO · CelebA | sensitive-attribute adversary accuracy: 0.783 -> 0.559 | Improved |
| Protect Mouth Open instead of Big Nose while predicting Smiling | DISCO · CelebA | sensitive-attribute adversary accuracy: 0.814 -> 0.783 | Mixed |
Later papers that compared with this method
3 later accepted papers report DisCo as a baseline in their results tables; 3 of them report a better number on most of the results they share with it.
- Simple Unsupervised Knowledge Distillation With Space Similarity (ECCV 2024 · better on 4 of 5 shared results)
- Bag of Instances Aggregation Boosts Self-supervised Distillation (ICLR 2022 · better on 6 of 6 shared results)
- Improving Self-Supervised Lightweight Model Learning via Hard-Aware Metric Distillation (ECCV 2022 · better on 5 of 5 shared results)
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- DISCO: Adversarial Defense with Local Implicit Functions (NeurIPS 2022 · similarity 0.86)
- Robust NAS under adversarial training: benchmark, theory, and beyond (ICLR 2024 · similarity 0.84)
- Certification for Differentially Private Prediction in Gradient-Based Training (ICML 2025 · similarity 0.84)
- Backdoor Channels Hidden in Latent Space: Cryptographic Undetectability in Modern Neural Networks (NeurIPS 2026 · similarity 0.83)
- United We Stand: Accelerating Privacy-Preserving Neural Inference by Conjunctive Optimization with Interleaved Nexus (AAAI 2024 · similarity 0.83)
- Certified Unlearning for Neural Networks (ICML 2025 · similarity 0.83)
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.
