Protecting Floating-Point Computation for DNN Binaries with MBA Obfuscation

2026-07-20

AI Summary

This work addresses the vulnerability of floating-point operations in deep neural network (DNN) binaries to reverse engineering, a limitation inadequately tackled by existing code obfuscation techniques. To this end, the paper introduces FLOB, the first framework that applies Mixed Boolean-Arithmetic (MBA) obfuscation to protect DNN floating-point computations. FLOB elevates floating-point values into a high-precision binary expansion space, applies MBA transformations therein, and then projects the results back to the original precision in a semantically equivalent manner—ensuring no additional rounding errors. This approach preserves numerical correctness while substantially enhancing resistance to reverse engineering, with flexible trade-offs between protection strength and performance. Experimental results demonstrate that FLOB reduces operator recovery rates to an average of 4.51%, a 32.82-percentage-point improvement over current methods.

Abstract

Deep neural networks (DNNs) have become a foundational component of modern computing systems with a wide range of applications, such as computer vision, edge intelligence, etc. For the sake of low latency and data privacy, DNN models are increasingly compiled into executables and deployed on local devices. However, that exposes the models to model theft, enabling adversaries to recover proprietary assets via reverse engineering techniques. While code obfuscation naturally emerges for protecting executables from reverse engineering, existing schemes are primarily designed for traditional programs, focusing on complex control flow structures and integer-based operations. They are fundamentally inappropriate for DNN binaries, which exhibit relatively simple code structures and heavily rely on floating-point computation. In this paper, we propose FLOB, an obfuscation framework that protects floating-point computations in DNN binaries using Mixed Boolean-Arithmetic (MBA) transformations. Our approach lifts floating-point values into a higher-precision binary expansion space and performs MBA transformations at the representation level. The computation is carried out entirely in the lifted space, and the final result is then projected back to the target precision, yielding outputs that are semantically equivalent under the original floating-point format, without introducing additional rounding error. The experimental results show that FLOB outperforms the state-of-the-art obfuscation schemes against existing reverse engineering analyzers and deobfuscators, while preserving computational correctness without introducing additional numerical error, allowing flexible trade-off between protection and performance. Specifically, FLOB reduces the operator recovery rate to 4.51% on average, achieving a 32.82 percentage-point reduction compared to existing methods.

Problem

• DNN binaries
• floating-point computation
• code obfuscation
• reverse engineering
• model theft

Innovation

• MBA obfuscation
• floating-point computation
• DNN binary protection
• binary lifting
• reverse engineering resistance