DiMEx: Breaking the Cold Start Barrier in Data-Free Model Extraction via Latent Diffusion Priors

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Thesia, Yash, Suthar, Meera
Format: Preprint
Published: 2026
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866914244421746688
author Thesia, Yash
Suthar, Meera
author_facet Thesia, Yash
Suthar, Meera
contents Model stealing attacks pose an existential threat to Machine Learning as a Service (MLaaS), allowing adversaries to replicate proprietary models for a fraction of their training cost. While Data-Free Model Extraction (DFME) has emerged as a stealthy vector, it remains fundamentally constrained by the "Cold Start" problem: GAN-based adversaries waste thousands of queries converging from random noise to meaningful data. We propose DiMEx, a framework that weaponizes the rich semantic priors of pre-trained Latent Diffusion Models to bypass this initialization barrier entirely. By employing Random Embedding Bayesian Optimization (REMBO) within the generator's latent space, DiMEx synthesizes high-fidelity queries immediately, achieving 52.1 percent agreement on SVHN with just 2,000 queries - outperforming state-of-the-art GAN baselines by over 16 percent. To counter this highly semantic threat, we introduce the Hybrid Stateful Ensemble (HSE) defense, which identifies the unique "optimization trajectory" of latent-space attacks. Our results demonstrate that while DiMEx evades static distribution detectors, HSE exploits this temporal signature to suppress attack success rates to 21.6 percent with negligible latency.
format Preprint
id arxiv_https___arxiv_org_abs_2601_01688
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle DiMEx: Breaking the Cold Start Barrier in Data-Free Model Extraction via Latent Diffusion Priors
Thesia, Yash
Suthar, Meera
Machine Learning
I.2.6; K.6.5
Model stealing attacks pose an existential threat to Machine Learning as a Service (MLaaS), allowing adversaries to replicate proprietary models for a fraction of their training cost. While Data-Free Model Extraction (DFME) has emerged as a stealthy vector, it remains fundamentally constrained by the "Cold Start" problem: GAN-based adversaries waste thousands of queries converging from random noise to meaningful data. We propose DiMEx, a framework that weaponizes the rich semantic priors of pre-trained Latent Diffusion Models to bypass this initialization barrier entirely. By employing Random Embedding Bayesian Optimization (REMBO) within the generator's latent space, DiMEx synthesizes high-fidelity queries immediately, achieving 52.1 percent agreement on SVHN with just 2,000 queries - outperforming state-of-the-art GAN baselines by over 16 percent. To counter this highly semantic threat, we introduce the Hybrid Stateful Ensemble (HSE) defense, which identifies the unique "optimization trajectory" of latent-space attacks. Our results demonstrate that while DiMEx evades static distribution detectors, HSE exploits this temporal signature to suppress attack success rates to 21.6 percent with negligible latency.
title DiMEx: Breaking the Cold Start Barrier in Data-Free Model Extraction via Latent Diffusion Priors
topic Machine Learning
I.2.6; K.6.5
url https://arxiv.org/abs/2601.01688