Research map / Learning theory and algorithms / Fairness
Trade-offs and Guarantees of Adversarial Representation Learning for Information Obfuscation
Han Zhao, Jianfeng Chi, Yuan Tian, Geoffrey J. Gordon
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 |
|---|---|---|---|
| Increase lambda from 0 to 5 (ablation) | MAX-ENT · Adult, gender | target task accuracy: 0.8501 -> 0.7891 | Hurt |
| Increase lambda from 0 to 5 (ablation) | GRL · Adult, gender | target task accuracy: 0.8501 -> 0.8387 | Hurt |
| Increase lambda from 0 to 5 (ablation) | ALT-UP · Adult, gender | target task accuracy: 0.8501 -> 0.8456 | Hurt |
| Increase lambda from 0 to 1 (ablation) | ALT-UP · Adult, gender | adversary accuracy on gender: 0.7408 -> 0.6627 | Improved |
| Increase lambda from 0 to 5 (ablation) | MAX-ENT · Adult, gender | adversary accuracy on gender: 0.7408 -> 0.5695 | Improved |
| Increase lambda from 0 to 1 (ablation) | GRL · Adult, age | adversary accuracy on age: 0.7190 -> 0.5361 | Improved |
| Increase lambda from 0 to 5 (ablation) | MAX-ENT · Adult, age | target task accuracy: 0.8467 -> 0.7795 | Hurt |
| Increase lambda from 0 to 5 (ablation) | ALT-UP · Adult, age | target task accuracy: 0.8467 -> 0.8451 | No clear effect |
| Increase lambda from 0 to 5 (ablation) | ALT-UP · Adult, education | target task accuracy: 0.8494 -> 0.8494 | No clear effect |
| Increase lambda from 0 to 5 (ablation) | GRL · Adult, education | adversary accuracy on education: 0.7088 -> 0.5528 | Improved |
| Increase lambda from 0 to 0.1 (ablation) | GRL · Adult, education | target task accuracy: 0.8494 -> 0.8525 | Improved |
| Increase lambda from 0 to 5 (ablation) | MAX-ENT · Adult, education | target task accuracy: 0.8494 -> 0.8087 | Hurt |
Closest papers by meaning
- Learning Robust and Privacy-Preserving Representations via Information Theory (AAAI 2025 · similarity 0.89)
- Adversarially Learned Representations for Information Obfuscation and Inference (ICML 2019 · similarity 0.88)
- PASS: Private Attributes Protection with Stochastic Data Substitution (ICML 2025 · similarity 0.85)
- Bits Leaked per Query: Information-Theoretic Bounds for Adversarial Attacks on LLMs (NeurIPS 2025 · similarity 0.84)
- A Unified Framework for Adversary-Aware Differential Privacy Bounds (NeurIPS 2026 · similarity 0.84)
- Provable Robustness against Backdoor Attacks via the Primal-Dual Perspective on Differential Privacy (NeurIPS 2026 · similarity 0.84)
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.
