# AdaPrivate-TS: Private Thompson Sampling for Contextual Bandits with Privacy Amplification

Source: [AdaPrivate-TS: Private Thompson Sampling for Contextual Bandits with Privacy Amplification](<https://www.sota2.com/research/paper/6a3a1356206542251d7c733a>)

Collection index: [llms.txt](<https://www.sota2.com/research/llms.txt>)

Data updated: 2026-06-24T05:00:13.022Z

Authors: Mohammadreza Riyazat, Eranga Ukwatta

Published: 2026-06-19T21:18:33.000Z

## Abstract

We present AdaPrivate-TS, a differentially private contextual bandit algorithm that combines Thompson Sampling with batched zCDP composition. Our key insight is that differential privacy noise inflates the posterior covariance in a structured way: adding Gaussian noise $N(0,\\sigma^2 I)$ to $b$ yields sampling covariance $v^2 A^{-1} + \\sigma^2 A^{-2}$, which Thompson Sampling interprets as increased uncertainty rather than pure corruption. Under event-level privacy (protecting individual interactions) with stochastic contexts, we prove that the privacy cost is only $O(\\sqrt{d}\\,\\log T/\\sqrt{\\rho})$, logarithmic in $T$, because parallel composition amortizes noise across batches. Additionally, we explore privacy amplification via Poisson subsampling, which can reduce effective noise at stringent privacy budgets. Experiments on synthetic and real-world datasets demonstrate: (1) AdaPrivate-TS achieves 93-99% of non-private performance at $\\varepsilon \\in \[0.5, 5\]$, outperforming UCB by 0.5-3.7% and up to 18% with tuned adaptive exploration at extreme $\\varepsilon$; (2) privacy amplification provides additional 2-5% gains at low $\\varepsilon$; (3) on MovieLens and Jester, AdaPrivate-TS achieves the best overall performance among event-level baselines, dominating at $\\varepsilon \\geq 2$; (4) under DP-SVD private features, TS's advantage over UCB grows to +11%, confirming noise-as-uncertainty is not limited to reward privacy. We provide rigorous proofs for privacy guarantees under interactive zCDP composition and comprehensive evaluation including convergence curves, 12-seed CIs, and DP-SVD feature ablation.

## Recorded benchmark results

A research result applies to its stated dataset, split, metric and evaluation configuration. Compare like-for-like settings, preserve the metric's unit and direction, and distinguish paper-reported results from independent reproduction. The best result recorded here is not a claim of exhaustive, current global coverage.

- [Contextual Bandits on MovieLens 25M](<https://www.sota2.com/research/sota/contextual-bandits-on-movielens-25m.md>): Contextual Bandits; MovieLens 25M; Success Rate vs Non-Private LinUCB: 90.6.

- [Contextual Bandits on Jester](<https://www.sota2.com/research/sota/contextual-bandits-on-jester.md>): Contextual Bandits; Jester; Non-private LinUCB Performance: 83.3.

## Original sources and implementation

[Original paper](<https://arxiv.org/abs/2606.21757>)

[PDF](<https://arxiv.org/pdf/2606.21757.pdf>)