# Power Mechanism: Private Tabular Representation Release for Model Agnostic Consumption

arXiv: https://arxiv.org/abs/2510.05581 · HTML: https://mlsanity.com/paper/2510.05581/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Builds on 11 of its 24 references that are in the map.
- First posted on arXiv 7 Oct 2025.

## Builds on (most cited first, 11 of 11)

- [Deep Learning with Differential Privacy](https://mlsanity.com/paper/1607.00133/) (2016)
- [Renyi Differential Privacy](https://mlsanity.com/paper/1702.07476/) (2017)
- [Differentially Private Learning Needs Better Features (or Much More Data)](https://mlsanity.com/paper/2011.11660/) (2020)
- [Adversary Instantiation: Lower Bounds for Differentially Private Machine Learning](https://mlsanity.com/paper/2101.04535/) (2021)
- [Concentrated Differential Privacy](https://mlsanity.com/paper/1603.01887/) (2016)
- [Protection Against Reconstruction and Its Applications in Private Federated Learning](https://mlsanity.com/paper/1812.00984/) (2018)
- [Unleashing the Tiger: Inference Attacks on Split Learning](https://mlsanity.com/paper/2012.02670/) (2020)
- [Pufferfish Privacy Mechanisms for Correlated Data](https://mlsanity.com/paper/1603.03977/) (2016)
- [Optimality of the Laplace Mechanism in Differential Privacy](https://mlsanity.com/paper/1504.00065/) (2015)
- [Blowfish Privacy: Tuning Privacy-Utility Trade-offs using Policies](https://mlsanity.com/paper/1312.3913/) (2013)
- [Gradual Release of Sensitive Data under Differential Privacy](https://mlsanity.com/paper/1504.00429/) (2015)

## Abstract (arXiv)

Traditional collaborative learning approaches are based on sharing of model weights between clients and a server. However, there are advantages to resource efficiency through schemes based on sharing of embeddings (activations) created from the data. Several differentially private methods were developed for sharing of weights while such mechanisms do not exist so far for sharing of embeddings. We propose Ours to learn a privacy encoding network in conjunction with a small utility generation network such that the final embeddings generated from it are equipped with formal differential privacy guarantees. These privatized embeddings are then shared with a more powerful server, that learns a post-processing that results in a higher accuracy for machine learning tasks. We show that our co-design of collaborative and private learning results in requiring only one round of privatized communication and lesser compute on the client than traditional methods. The privatized embeddings that we share from the client are agnostic to the type of model (deep learning, random forests or XGBoost) used on the server in order to process these activations to complete a task.