arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2604.06644cs.CVcs.LG
收藏

变分特征压缩用于模型特定的表示

Variational Feature Compression for Model-Specific Representations

The University of Queensland(昆士兰大学);University of Wollongong(伍伦贡大学);City University of Hong Kong(香港城市大学)

Zinan Guo, Zihan Wang, Chuan Yan, Liuhuo Wan, Ethan Ma, Guangdong Bai

AI总结:

本文提出一种变分特征压缩框架,通过任务驱动的交叉熵目标和KL正则化,在不进行像素级重建的情况下,压缩输入到紧凑的潜在空间,同时抑制跨模型转移,提升目标分类器的准确性。

摘要:

随着深度学习推理越来越多地部署在共享和云环境,输入再利用成为一大问题,即提交给一个任务的数据被未经授权的模型用于其他任务。现有隐私防护主要限制数据访问,但对释放的表示的下游使用控制有限。本文提出一个特征提取框架,通过变分潜在瓶颈,结合任务驱动的交叉熵目标和KL正则化,不进行像素级重建,将输入编码到紧凑的潜在空间。动态二进制掩码通过每维KL散度和基于梯度的显著性计算,抑制对目标任务无信息的潜在维度。由于显著性计算需要梯度访问,编码器在白盒环境下训练,而推理只需通过冻结的目标模型进行前向传递。在CIFAR-100上,处理后的表示对目标分类器仍然有效,而所有未授权分类器的准确性降至低于2%,抑制比超过45倍。初步实验在CIFAR-10、Tiny ImageNet和Pascal VOC上提供了跨任务的证据,但需要进一步评估以评估对自适应对手的鲁棒性。

英文摘要:

As deep learning inference is increasingly deployed in shared and cloud-based settings, a growing concern is input repurposing, in which data submitted for one task is reused by unauthorized models for another. Existing privacy defenses largely focus on restricting data access, but provide limited control over what downstream uses a released representation can still support. We propose a feature extraction framework that suppresses cross-model transfer while preserving accuracy for a designated classifier. The framework employs a variational latent bottleneck, trained with a task-driven cross-entropy objective and KL regularization, but without any pixel-level reconstruction loss, to encode inputs into a compact latent space. A dynamic binary mask, computed from per-dimension KL divergence and gradient-based saliency with respect to the frozen target model, suppresses latent dimensions that are uninformative for the intended task. Because saliency computation requires gradient access, the encoder is trained in a white-box setting, whereas inference requires only a forward pass through the frozen target model. On CIFAR-100, the processed representations retain strong utility for the designated classifier while reducing the accuracy of all unintended classifiers to below 2%, yielding a suppression ratio exceeding 45 times relative to unintended models. Preliminary experiments on CIFAR-10, Tiny ImageNet, and Pascal VOC provide exploratory evidence that the approach extends across task settings, although further evaluation is needed to assess robustness against adaptive adversaries.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑