arXiv:2608. 11003v1 Announce Type: cross Abstract: In this work, we study the information bottleneck under perfect privacy, with particular emphasis on the active-rate regime, where the representation-rate constraint is binding and directly limits the achievable utility.
arXiv:2508.04800v2 Announce Type: replace-cross Abstract: We introduce a novel privatization framework for high-dimensional controlled variable selection. Our framework enables rigorous False Discove...
arXiv:2505. 22703v2 Announce Type: replace Abstract: Many problems in trustworthy ML can be expressed as constraints on prediction rates across subpopulations, including group fairness constraints (demographic parity, equalized odds, etc.
arXiv:2605. 05905v2 Announce Type: replace Abstract: Objective perturbation is a standard mechanism in differentially private empirical risk minimization.
arXiv:2603. 19040v2 Announce Type: replace Abstract: Differentially private wireless federated learning (DPWFL) is a promising framework for protecting sensitive user data.
arXiv:2602. 17284v2 Announce Type: replace Abstract: We consider the privacy amplification properties of a sampling scheme in which a user's data isused in $k$ steps chosen randomly and uniformly from a sequence (or set) of $t$ steps.
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.
arXiv:2407. 08233v3 Announce Type: replace Abstract: Current differentially private learning paradigms face a severe utility bottleneck: DP-SGD degrades performance through noise accumulation over training steps, while aggregation-based approaches such as PATE suffer from data inefficiency due to disjoint data partitioning.
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:2606. 04384v1 Announce Type: new Abstract: Machine learning's reliance on sensitive data necessitates privacy-preserving techniques like Differentially Private Stochastic Gradient Descent (DPSGD).
arXiv:2512. 04008v2 Announce Type: replace Abstract: Training with differential privacy (DP) guarantees dataset members that they cannot be identified by users of the released model.
The paper introduces a general learning framework that protects privacy in federated learning by distorting model parameters, enabling a trade‑off between privacy and utility. The algorithm supports arbitrary privacy measurements and delivers personalized utility‑privacy balances for each parameter, client, and communication round. The authors prove that the gap between their algorithm’s utility loss and the optimal loss is sub‑linear in iterations, provide a convergence rate, and demonstrate empirically that their method outperforms baselines under the same privacy budget.
The paper proposes a privacy‑aligned personalized federated learning method that releases a private client context once and limits repeated adaptation to a fixed coefficient space, thereby reducing dimensionality misalignment. A factorized generator creates an adaptive optimization geometry that reshapes noisy updates, and most of the private‑training benefit is preserved by radial evolution. Variable‑length Gaussian quantization is used for coefficient updates, allowing the quantization error to act as the privacy perturbation and cutting protected uplink communication by a factor of 2.67 on CIFAR‑10 at ε=16 while maintaining comparable future‑client accuracy.