# An Interactive Framework for Finding the Optimal Trade-off in Differential Privacy

arXiv: https://arxiv.org/abs/2509.04290 · HTML: https://mlsanity.com/paper/2509.04290/

## Key facts

- Cited by 1 papers in the map (data through 14 Oct 2026).
- Builds on 32 of its 62 references that are in the map.
- First posted on arXiv 4 Sep 2025.

## Builds on (most cited first, 30 of 32)

- [An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale](https://mlsanity.com/paper/2010.11929/) (2020)
- [Deep Learning with Differential Privacy](https://mlsanity.com/paper/1607.00133/) (2016)
- [Optuna: A Next-generation Hyperparameter Optimization Framework](https://mlsanity.com/paper/1907.10902/) (2019)
- [ImageNet-21K Pretraining for the Masses](https://mlsanity.com/paper/2104.10972/) (2021)
- [Rotation Equivariant CNNs for Digital Pathology](https://mlsanity.com/paper/1806.03962/) (2018)
- [Concentrated Differential Privacy: Simplifications, Extensions, and Lower Bounds](https://mlsanity.com/paper/1605.02065/) (2016)
- [Unlocking High-Accuracy Differentially Private Image Classification through Scale](https://mlsanity.com/paper/2204.13650/) (2022)
- [Numerical Composition of Differential Privacy](https://mlsanity.com/paper/2106.02848/) (2021)
- [Concentrated Differential Privacy](https://mlsanity.com/paper/1603.01887/) (2016)
- [Reward-rational (implicit) choice: A unifying formalism for reward learning](https://mlsanity.com/paper/2002.04833/) (2020)
- [Hyperparameter Tuning with Renyi Differential Privacy](https://mlsanity.com/paper/2110.03620/) (2021)
- [Max-value Entropy Search for Multi-Objective Bayesian Optimization with Constraints](https://mlsanity.com/paper/2009.01721/) (2020)
- [Private Selection from Private Candidates](https://mlsanity.com/paper/1811.07971/) (2018)
- [Preferential Bayesian Optimization](https://mlsanity.com/paper/1704.03651/) (2017)
- [Prediction-Oriented Bayesian Active Learning](https://mlsanity.com/paper/2304.08151/) (2023)
- [Multi-objective Bayesian Optimization using Pareto-frontier Entropy](https://mlsanity.com/paper/1906.00127/) (2019)
- [TAN Without a Burn: Scaling Laws of DP-SGD](https://mlsanity.com/paper/2210.03403/) (2022)
- [Privacy Loss in Apple's Implementation of Differential Privacy on MacOS 10.12](https://mlsanity.com/paper/1709.02753/) (2017)
- [Preference Exploration for Efficient Bayesian Optimization with Multiple Outcomes](https://mlsanity.com/paper/2203.11382/) (2022)
- [qEUBO: A Decision-Theoretic Acquisition Function for Preferential Bayesian Optimization](https://mlsanity.com/paper/2303.15746/) (2023)
- [Multi-Attribute Bayesian Optimization With Interactive Preference Learning](https://mlsanity.com/paper/1911.05934/) (2019)
- [Practical Differentially Private Hyperparameter Tuning with Subsampling](https://mlsanity.com/paper/2301.11989/) (2023)
- ["I need a better description'': An Investigation Into User Expectations For Differential Privacy](https://mlsanity.com/paper/2110.06452/) (2021)
- [Automatic Discovery of Privacy-Utility Pareto Fronts](https://mlsanity.com/paper/1905.10862/) (2019)
- [What Are the Chances? Explaining the Epsilon Parameter in Differential Privacy](https://mlsanity.com/paper/2303.00738/) (2023)
- [Multi-Objective Bayesian Optimization with Active Preference Learning](https://mlsanity.com/paper/2311.13460/) (2023)
- [On the Efficacy of Differentially Private Few-shot Image Classification](https://mlsanity.com/paper/2302.01190/) (2023)
- [A New Linear Scaling Rule for Private Adaptive Hyperparameter Optimization](https://mlsanity.com/paper/2212.04486/) (2022)
- [Interactive Hyperparameter Optimization in Multi-Objective Problems via Preference Learning](https://mlsanity.com/paper/2309.03581/) (2023)
- [Visualizing Privacy-Utility Trade-Offs in Differentially Private Data Releases](https://mlsanity.com/paper/2201.05964/) (2022)

## Built on by (most cited first, 1 of 1)

- [Multi-Objective Bayesian Optimization via Adaptive varepsilon-Constraints Decomposition](https://mlsanity.com/paper/2604.15959/) (2026)

## Abstract (arXiv)

Differential privacy (DP) is the standard for privacy-preserving analysis, and introduces a fundamental trade-off between privacy guarantees and model performance. Selecting the optimal balance is a critical challenge that can be framed as a multi-objective optimization (MOO) problem where one first discovers the set of optimal trade-offs (the Pareto front) and then learns a decision-maker's preference over them. While a rich body of work on interactive MOO exists, the standard approach -- modeling the objective functions with generic surrogates and learning preferences from simple pairwise feedback -- is inefficient for DP because it fails to leverage the problem's unique structure: a point on the Pareto front can be generated directly by maximizing accuracy for a fixed privacy level. Motivated by this property, we first derive the shape of the trade-off theoretically, which allows us to model the Pareto front directly and efficiently. To address inefficiency in preference learning, we replace pairwise comparisons with a more informative interaction. In particular, we present the user with hypothetical trade-off curves and ask them to pick their preferred trade-off. Our experiments on differentially private logistic regression and deep transfer learning across six real-world datasets show that our method converges to the optimal privacy-accuracy trade-off with significantly less computational cost and user interaction than baselines.