The paper formalizes slate recommendation as a two-stage process: a randomized score learner followed by deterministic selection. It shows that a differential‑privacy guarantee can be preserved through the selection stage when selector inputs are public or independent, and introduces a logged margin certificate that bounds score‑induced objective changes, ensuring the slate order remains stable. Experiments on OULAD, MovieLens‑25M, and Amazon Musical Instruments demonstrate that higher anchor weight reduces ranking churn, while simulations confirm bounded target drift and utility dependent on settings.
arXiv:2601. 10237v3 Announce Type: replace Abstract: Differentially Private Stochastic Gradient Descent (DP-SGD) is the dominant paradigm for private training, but its fundamental limitations under worst-case adversarial privacy definitions remain poorly understood.
arXiv:2601. 14033v2 Announce Type: replace Abstract: Machine learning models are increasingly served behind APIs.
arXiv:2610.07212v1 Announce Type: new Abstract: Reinforcement learning with verifiable rewards (RLVR) trains a language model on problems that may themselves be confidential, and the trained model ca...
The paper introduces Private Best-of-N (PrivBoN), a method that adds calibrated Gumbel noise to reward scores during inference-time alignment, achieving both ε-differential privacy and KL-regularized alignment. When the privacy budget exceeds a critical threshold ε*, the noise becomes regret-optimal, matching the theoretical alignment skyline. The authors also propose Private Inference-Time Pessimism (PrivITP), which uses χ^2-regularized rejection sampling and a two-phase Gaussian mechanism to provide ex-post (ε,δ)-DP with a privacy cost independent of the number of responses, and demonstrate that both methods outperform standard Best-of-N across multiple models and datasets.
The paper introduces a differentially private framework for sparse reward estimation in reinforcement learning with human feedback (RLHF). It assumes that only a small subset of features influences human preferences and provides theoretical error bounds for both Sample-DP and Label-DP settings under the Bradley-Terry-Luce model. Empirical results show that the proposed methods outperform dense baselines on sentiment generation and safety alignment tasks, achieving a better privacy‑utility trade‑off.
arXiv:2606. 12733v2 Announce Type: replace Abstract: Privacy auditing provides an important safeguard by estimating the actual information leaked by a model, thus ensuring that theoretical privacy guarantees hold in practice.
Contextual dueling bandits learn from pairwise comparisons, the same feedback used to fine-tune language models from human preferences. Standard algorithms explore aggressively and can show users two...
The paper introduces conservative contextual dueling bandits, which enforce that each displayed item’s cumulative score remains above a specified fraction of a trusted baseline’s score. It presents algorithms—MaxInP-BC for linear utilities and BTIGW-BC for general utilities—that guarantee safety with high probability while achieving sublinear regret. Experiments show these methods satisfy the safety constraint more reliably than unconstrained counterparts, with only a moderate increase in regret.
arXiv:2606. 02563v1 Announce Type: new Abstract: Heterogeneous Differential Privacy (HDP) in Federated Learning (FL) allows clients to select individual privacy budgets ($\varepsilon_i$) according to institutional policies and data sensitivity.
The paper extends the abstract gradient training (AGT) framework to provide tighter differential privacy guarantees for both private prediction and private learning. It introduces Abstract Gradient Sampling (AGS) to analyze privacy in continuous, unbounded regression and offers theoretical and empirical evidence that these methods yield tighter bounds than global-sensitivity baselines, even in previously unbounded settings. The authors also demonstrate that their private learning algorithm can outperform standard private learners under comparable conditions.
Heterogeneous Differential Privacy (HDP) in Federated Learning (FL) allows clients to select individual privacy budgets ($\varepsilon_i$) according to institutional policies and data sensitivity. In practice, many HDP-FL systems employ $\varepsilon$-aware server aggregation to improve model utility by re-weighting client updates according to their declared privacy budgets.
arXiv:2303. 07152v3 Announce Type: replace-cross Abstract: Achieving optimal statistical performance while ensuring the privacy of personal data is a challenging yet crucial objective in modern data analysis.