Fine-tuning large language models on private data for downstream applications poses significant privacy risks in potentially exposing sensitive information. Several popular community platforms now offer convenient distribution of a large variety of pre-trained models, allowing anyone to publish without rigorous verification. This scenario creates a privacy threat, as pre-trained models can be intentionally crafted to compromise the privacy of fine-tuning datasets. In this study, we introduce a novel poisoning technique that uses model-unlearning as an attack tool. This approach manipulates a pre-trained language model to increase the leakage of private data during the fine-tuning process. Our method enhances both membership inference and data extraction attacks while preserving model utility. Experimental results across different models, datasets, and fine-tuning setups demonstrate that our attacks significantly surpass baseline performance. This work serves as a cautionary note for users who download pretrained models from unverified sources, highlighting the potential risks involved.

In distributed learning settings, models are iteratively updated with shared gradients computed from potentially sensitive user data. While previous work has studied various privacy risks of sharing gradients, our paper aims to provide a systematic approach to analyze private information leakage from gradients. We present a unified game-based framework that encompasses a broad range of attacks including attribute, property, distributional, and user disclosures. We investigate how different uncertainties of the adversary affect their inferential power via extensive experiments on five datasets across various data modalities. Our results demonstrate the inefficacy of solely relying on data aggregation to achieve privacy against inference attacks in distributed learning. We further evaluate five types of defenses, namely, gradient pruning, signed gradient descent, adversarial perturbations, variational information bottleneck, and differential privacy, under both static and adaptive adversary settings. We provide an information-theoretic view for analyzing the effectiveness of these defenses against inference from gradients. Finally, we introduce a method for auditing attribute inference privacy, improving the empirical estimation of worst-case privacy through crafting adversarial canary records.

In this paper, we propose a new Bayesian model to solve the Robust PCA problem - recovering the underlying low-rank matrix and sparse matrix from their noisy compositions. We first derive and analyze a new objective function, which is proven to be equivalent to the fundamental minimizing "rank+sparsity" objective. To solve this objective, we develop a concise Sparse Bayesian Learning (SBL) method that has minimum assumptions and effectively deals with the crux of the problem. The concise modeling allows simple and effective Empirical Bayesian inference via MAP-EM. Simulation studies demonstrate the superiority of the proposed method over the existing state-of-the-art methods. The efficacy of the method is further verified through a text extraction image processing task.

We study the problem of low-rank and sparse decomposition from possibly noisy observations. We propose a novel objective function with nuclear norm on the low-rank term and ℓ0-`norm' on the sparse term, as well as ℓ1-norm on the additive noise term. When there is no dense inlier noise, the proposed method shares the same theoretical guarantee as the Principal Component Pursuit (PCP), i.e., it can recover the low-rank component and sparse component exactly with high probability. Simulations in the noisy case demonstrate that the proposed method outperforms existing state-of-the-art methods. Results on a surveillance video application further verify the effectiveness of the proposed method.