Fully Quantum Latent Diffusion for Synthetic Malware Records: Effective Denoising, Limited Class Information, and Costly Post-Selection
Abstract
Malware benchmarks are difficult to share, because the samples are harmful and a corpus can reveal the incidents of its holder, while synthetic substitutes can leak training records. Quantum diffusion models, however, have been evaluated only on images or quantum states, leaving the utility, privacy, and sampling cost of their samples unknown. In this research, we introduce a fully quantum latent diffusion model for malware records, in which a parameterized circuit with one mid-circuit measurement denoises amplitude-encoded latent vectors of opcode sequences without classical neural layers. On 17 malware families with ten seeds, we compared it with random states, a perfect generator, a parameter-matched classical control, and a classical latent diffusion model. The circuit lowers its denoising infidelity by 63\%, yet classifiers trained on its records reach a macro-F1 score of 0.11 on real records, against 0.08 for random states, 0.61 for a perfect generator, and 0.44 for a classical diffusion model of comparable size, and adding its records to real training data lowers the score from 0.635 to 0.593. Post-selected sampling accepts 3.0×10−7 of runs, and a training penalty that raises the acceptance to 0.14 removes most of the denoising. No synthetic record copies a training record, and smoothing the decoded lengths lowers the area under the curve of a length-based membership test from 0.80 to 0.51, which a differentially private length histogram matches with a formal guarantee. These results indicate that quantum generators of shareable security data require a joint evaluation of utility, privacy, and sampling cost.
Related
- Acceleration of CTQMC via Denoising Diffusion Models with Auxiliary Spectral Functions
- Capacity Overflow, Effective Stochasticity, and Phase-B Invariants: Critical Deficit, Markov Closure, and Invariant Selection under Finite Projection
- Reduced Precision Diffusion Synthetic Acceleration for SN Neutron Transport in LLNL's ARDRA using Hypre
- From Effective Temperature to Non-Boltzmann State Selection in Driven-Dissipative Quantum Criticality
- All-Order Helicity Selection Rules in Effective Field Theories
BibTeX
@misc{tran2026fully,
title = {{Fully Quantum Latent Diffusion for Synthetic Malware Records: Effective Denoising, Limited Class Information, and Costly Post-Selection}},
author = {Ban Q. Tran and Quang Duy Tran and Anh H. Nguyen and Ziwen Pan and Tommy Dang and Susan Mengel},
howpublished = {arXiv},
year = {2026},
month = oct,
doi = {10.48550/arxiv.2610.38974},
eprint = {2610.38974},
archivePrefix = {arXiv},
url = {https://arxiv.org/abs/2610.38974},
note = {CC BY 4.0},
}
RIS
TY - UNPB
TI - Fully Quantum Latent Diffusion for Synthetic Malware Records: Effective Denoising, Limited Class Information, and Costly Post-Selection
AU - Ban Q. Tran
AU - Quang Duy Tran
AU - Anh H. Nguyen
AU - Ziwen Pan
AU - Tommy Dang
AU - Susan Mengel
PY - 2026
DA - 2026/10/28
PB - arXiv
DO - 10.48550/arxiv.2610.38974
UR - https://arxiv.org/abs/2610.38974
AB - Malware benchmarks are difficult to share, because the samples are harmful and a corpus can reveal the incidents of its holder, while synthetic substitutes can leak training records. Quantum diffusion models, however, have been evaluated only on images or quantum states, leaving the utility, privacy, and sampling cost of their samples unknown. In this research, we introduce a fully quantum latent diffusion model for malware records, in which a parameterized circuit with one mid-circuit measurement denoises amplitude-encoded latent vectors of opcode sequences without classical neural layers. On 17 malware families with ten seeds, we compared it with random states, a perfect generator, a parameter-matched classical control, and a classical latent diffusion model. The circuit lowers its denoising infidelity by 63\%, yet classifiers trained on its records reach a macro-F1 score of 0.11 on real records, against 0.08 for random states, 0.61 for a perfect generator, and 0.44 for a classical diffusion model of comparable size, and adding its records to real training data lowers the score from 0.635 to 0.593. Post-selected sampling accepts 3.0×10^−7 of runs, and a training penalty that raises the acceptance to 0.14 removes most of the denoising. No synthetic record copies a training record, and smoothing the decoded lengths lowers the area under the curve of a length-based membership test from 0.80 to 0.51, which a differentially private length histogram matches with a formal guarantee. These results indicate that quantum generators of shareable security data require a joint evaluation of utility, privacy, and sampling cost.
ER -