通过剪枝降低神经网络隐私泄露，同时保持模型精度。
CutClean: Neural Network Pruning for Privacy-Preserving Inference
- 在每层网络后加隐私头，量化敏感信息泄露程度。
- 逐步增加稀疏性，使最后一层隐私头准确率降至15%以下。
- 适合需要保护用户隐私的高风险应用场景。
神经网络在高风险应用中部署日益广泛，但隐私泄露问题愈发突出。我们发现，即使没有传统数据偏差导致的表示不平衡，隐私泄露仍可能发生，这在处理敏感属性时构成重大风险。为此，我们提出CutClean，一种面向隐私的剪枝方法，可在提升网络稀疏性的同时减少隐私信息流动。该方法在每一网络块后设置辅助线性隐私头，用于量化信息泄露，并逐步施加稀疏化以消除私有属性泄露，其效果以最后一层隐私头的分类准确率衡量。在合成与真实数据集上的实验表明，该方法能有效降低私有信息流，实现高达80%以上的稀疏率，同时保持目标分类任务的高精度。
原文摘要 · Abstract (English)
Neural networks are increasingly deployed in high-stakes applications with growing privacy leakage concerns. We show that this privacy leakage can occur even in the absence of representation imbalances that lead to traditional dataset biases. This poses significant privacy risks when deploying models that process sensitive attributes. In this context, we propose CutClean, a privacy-aware pruning method that allows to reduce privacy information flow through the network, while increasing its sparsity. Our approach employs auxiliary linear privacy heads placed at each network's block to quantify information leakage, and further applies increasing levels of sparsity to remove the private attribute leakage, measured in terms of the accuracy of the privacy head attached to the last block. Experiments on synthetic and real-world datasets demonstrate that our approach effectively minimizes private information flow while achieving high sparsity rates and preserving classification target accuracy.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。