社会层面 · privacy论文趋势 2023–2026

主要结论

• 社会层面 · privacy论文从 2023 年的 21 篇增长到 2026 年的 138 篇，三年间扩大到约 6.6 倍。
• NeurIPS 是发社会层面 · privacy论文最多的会议，2023–2026 年共 215 篇，占四个会议总量的 50%。
• 交叉方向里涨得最快的是社会层面 · accountability transparency and interpretability：同时属于两个主题的论文从 2023 年的 0 篇增至 2026 年的 93 篇。
• 在「对齐与负责任 AI」的 5 个主题里，它的 2026 年论文数排第 4；同组增长最快的社会层面 · safety是 93 倍，社会层面 · privacy是 6.6 倍。

代表论文

按年份和会议均匀挑选，方便快速看到每一届的样貌

• NeurIPS 2026Assessing Per-Sample Membership Inference Vulnerability without Retraining
• ICML 2026Accuracy-First Rényi Differential Privacy and Post-Processing Immunity
• ICLR 2026Mitigating Privacy Risk via Forget Set-Free Unlearning
• CVPR 2026Protego User-Centric Pose-Invariant Privacy Protection Against Face Recognition-Induced Digital Footprint
• NeurIPS 2025A Generalized Binary Tree Mechanism for Private Approximation of All-Pair Shortest Distances
• ICML 2025Breaking the $n^{1.5}$ Additive Error Barrier for Private and Efficient Graph Sparsification via Private Expander Decomposition
• ICLR 2025Machine Unlearning via Simulated Oracle Matching
• CVPR 2025Enhancing Facial Privacy Protection via Weakening Diffusion Purification
• NeurIPS 2024Adaptive Randomized Smoothing: Certified Adversarial Robustness for Multi-Step Defences
• ICML 2024Privacy Preserving Adaptive Experiment Design
• ICLR 2024Beyond Memorization: Violating Privacy via Inference with Large Language Models
• CVPR 2024Validating Privacy-Preserving Face Recognition under a Minimum Assumption
• NeurIPS 2023Faster approximate subgraph counts with privacy
• ICML 2023GuardHFL: Privacy Guardian for Heterogeneous Federated Learning
• ICLR 2023Decepticons: Corrupted Transformers Breach Privacy in Federated Learning for Language Models
• CVPR 2023Privacy-Preserving Representations Are Not Enough: Recovering Scene Content From Camera Poses
• NeurIPS 2026Asymmetric Invertible Threat: Learning Reversible Privacy Defense for Face Recognition
• ICML 2026Accurate, Private, Secure, Federated U-statistics with Higher Degree
• ICLR 2026Measuring Physical-World Privacy Awareness of Large Language Models: An Evaluation Benchmark
• CVPR 2026LDP-Slicing Local Differential Privacy for Images via Randomized Bit-Plane Slicing

全部社会层面 · privacy论文

共 431 篇，第 2 / 9 页

51. ICML 2026Leveraging Machine Unlearning for Cost-Efficient Preference Alignment
52. ICML 2026LMCleaner: Efficient and Certified Online Unlearning via Influence Propagation Truncation
53. ICML 2026Mind the Gap: Mixtures of Gaussians in Approximate Differential Privacy
54. ICML 2026MINIM: Privacy-Aware Minimal View for Agents via Trusted Local Sanitization
55. ICML 2026Multilingual Unlearning in LLMs: Transfer, Dynamics, and Reversibility
56. ICML 2026MultiPriv: Benchmarking Individual-Level Privacy Reasoning in Vision-Language Models
57. ICML 2026Near-Optimal Private Linear Regression via Iterative Hessian Mixing
58. ICML 2026OptiFluence: Principled Design of Privacy Canaries
59. ICML 2026Optimal conversion from Rényi Differential Privacy to $f$-Differential Privacy
60. ICML 2026OSNIP: Balancing the Privacy-Utility-Efficiency Trilemma in LLM Inference via Obfuscated Semantic Null Space
61. ICML 2026PEARL: Differentially Private and Entropy-Aware Regulated Language Generation
62. ICML 2026Persuasive Privacy
63. ICML 2026Preference-Calibrated Optimization with Score-Level Distribution Alignment for Text-to-Image Diffusion Model Unlearning
64. ICML 2026PRISM: Gauge-Invariant Tangent-Space Differentially Private LoRA
65. ICML 2026Privacy Amplification in Differentially Private Zeroth-Order Optimization with Hidden States
66. ICML 2026Privacy-Aware Data Integration for Enhanced Quantile Inference under Heterogeneity
67. ICML 2026Privasis: Synthesizing the Largest "Public" Private Dataset from Scratch
68. ICML 2026Private Learning with Public Feature Conditioning
69. ICML 2026Privately Fine-Tuned LLMs Preserve Temporal Dynamics in Tabular Data
70. ICML 2026PrivCode++: Latent-Conditioned Differentially Private Code Generation for Comprehensive Guarantees
71. ICML 2026PrivGate: Steering Contextual Integrity in LLMs via Latent Space Geometry
72. ICML 2026Provably Protecting Fine-Tuned LLMs from Training Data Extraction while Preserving Utility
73. ICML 2026RECOVER: Reliable Detection of Unauthorized Data Usage in Text-to-Image Diffusion Models via Inversion Robustness
74. ICML 2026Reverse-Engineering Model Editing on Language Models
75. ICML 2026Revisiting Asymmetries in Black-box Link Stealing against Graph Neural Networks
76. ICML 2026Revisiting ML Training under Fully Homomorphic Encryption: Convergence Guarantees, Differential Privacy, and Efficient Algorithms
77. ICML 2026Selective Disclosure Watermarking for Large Language Models
78. ICML 2026Shift-Dependent Asymmetry: Orthogonal Inverse Low-Rank Adaptation for Federated Medical Segmentation
79. ICML 2026Shuffling-Aware Optimization for Private Vector Mean Estimation
80. ICML 2026SlaClip: Gradient Norm Slacks can be Indicator for Adaptive Clipping in DP-SGD
81. ICML 2026Social Hippocampus Memory Learning
82. ICML 2026Solving Positive Linear Programs with Differential Privacy
83. ICML 2026Tackling Fake Forgetting through Uncertainty Quantification
84. ICML 2026The Double-Edged Nature of the Rashomon Set for Trustworthy Machine Learning
85. ICML 2026The Forgetting-Retention Dilemma: Certified Unlearning Theory in Continual Learning
86. ICML 2026The Hidden Risk: Membership Inference Attacks on Multimodal Federated Learning via Modality Imbalance
87. ICML 2026Towards Whole-corpus Reconstruction of Heterogeneous RAG Knowledge Bases
88. ICML 2026Trajectory-Aware Certified Decentralized Unlearning via SGD Stability
89. ICML 2026Unlearning Isn’t Forgetting: Revealing Hidden Leakage in Class Unlearning Evaluations
90. ICML 2026Unlearning with Asymmetric Sources: Improved Unlearning-Utility Trade-off with Public Data
91. ICML 2026Unlearning’s Blind Spots: Over‑Unlearning and Prototypical Relearning Attack
92. ICML 2026Variance-Reduced $(\varepsilon, \delta)-$Unlearning using Forget Set Gradients
93. ICML 2026VPD-100K: Towards Generalizable and Fine-grained Visual Privacy Protection
94. ICML 2026Who’s in Charge? Disempowerment Patterns in Real-World LLM Usage
95. NeurIPS 2026A Private Empirical Defense Against Privacy Audits
96. NeurIPS 2026Assessing Per-Sample Membership Inference Vulnerability without Retraining
97. NeurIPS 2026Asymmetric Invertible Threat: Learning Reversible Privacy Defense for Face Recognition
98. NeurIPS 2026Asymptotically Optimal Best Arm Identification with Fixed-Budget under Differential Privacy
99. NeurIPS 2026Auditing Privacy Leakage in Tabular Foundation Model Embeddings
100. NeurIPS 2026BEAGLE: Behavior-Enforced Agent for Grounded Learner Emulation

常见问题

2023–2026 年四大会议一共有多少篇社会层面 · privacy论文？

共 431 篇，按年份分别是 2023 年 21 篇、2024 年 117 篇、2025 年 155 篇、2026 年 138 篇。

NeurIPS 2026 有多少篇社会层面 · privacy论文？

NeurIPS 2026 有 44 篇社会层面 · privacy论文，上一届是 80 篇。

ICLR 2026 有多少篇社会层面 · privacy论文？

ICLR 2026 有 2 篇社会层面 · privacy论文，上一届是 2 篇。

ICML 2026 有多少篇社会层面 · privacy论文？

ICML 2026 有 90 篇社会层面 · privacy论文，上一届是 71 篇。

CVPR 2026 有多少篇社会层面 · privacy论文？

CVPR 2026 有 2 篇社会层面 · privacy论文，上一届是 2 篇。

哪个会议的社会层面 · privacy论文最多？

NeurIPS，2023–2026 年共 215 篇。