# Defense Against Model Stealing Based on Account-Aware Distribution Discrepancy

Source: [Defense Against Model Stealing Based on Account-Aware Distribution Discrepancy](<https://www.sota2.com/research/paper/68fa77741e209131cea73358>)

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

Data updated: 2025-10-26T04:18:49.615Z

Authors: Jian-Ping Mei, Weibin Zhang, Jie Chen, Xuyun Zhang, Tiantian Zhu

Published: 2025-03-16T13:22:53.000Z

## Abstract

Malicious users attempt to replicate commercial models functionally at low cost by training a clone model with query responses. It is challenging to timely prevent such model-stealing attacks to achieve strong protection and maintain utility. In this paper, we propose a novel non-parametric detector called Account-aware Distribution Discrepancy (ADD) to recognize queries from malicious users by leveraging account-wise local dependency. We formulate each class as a Multivariate Normal distribution (MVN) in the feature space and measure the malicious score as the sum of weighted class-wise distribution discrepancy. The ADD detector is combined with random-based prediction poisoning to yield a plug-and-play defense module named D-ADD for image classification models. Results of extensive experimental studies show that D-ADD achieves strong defense against different types of attacks with little interference in serving benign users for both soft and hard-label settings.

## 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.

## Original sources and implementation

[Original paper](<https://arxiv.org/abs/2503.12497>)

[PDF](<https://arxiv.org/pdf/2503.12497.pdf>)