ACL Findings2025
gMBA: Expression Semantic Guided Mixed Boolean-Arithmetic Deobfuscation Using Transformer Architectures
Youjeong Noh, Joon-Young Paik, Jingun Kwon, Eun-Sun Cho
Abstract
Mixed Boolean-Arithmetic (MBA) obfuscation protects intellectual property by converting programs into forms that are more complex to analyze. However, MBA has been increasingly exploited by malware developers to evade detection and cause significant real-world problems. Traditional MBA deobfuscation methods often consider these expressions as part of a black box and overlook their internal semantic information. To bridge this gap, we propose a truth table, which is an automatically constructed semantic representation of an expression's behavior that does not rely on external resources. The truth table is a mathematical form that represents the output of expression for all possible combinations of input. We also propose a general and extensible guided MBA deobfuscation framework (gMBA) that modifies a Transformer-based neural encoderdecoder Seq2Seq architecture to incorporate this semantic guidance. Experimental results and in-depth analysis show that integrating expression semantics significantly improves performance and highlights the importance of internal semantic expressions in recovering obfuscated code to its original form. 1
Ask Lune
Know this paper in minutes, not an afternoon.
Lune can read every page of it and the top-tier papers that cite it, and gives you its claims, its evidence and its limits, each one cited to the paper it comes from. No skimming, no AI slop.
Lineage
- Simplifying Mixed Boolean-Arithmetic Obfuscation by Program Synthesis and Term Rewriting cites Loki: Hardening Code Obfuscation Against Automated Attacks
- Simplifying Mixed Boolean-Arithmetic Obfuscation by Program Synthesis and Term Rewriting cites NeuReduce: Reducing Mixed Boolean-Arithmetic Expressions by Recurrent Neural Network
- Simplifying Mixed Boolean-Arithmetic Obfuscation by Program Synthesis and Term Rewriting cites MBA-Blast: Unveiling and Simplifying Mixed Boolean-Arithmetic Obfuscation
- Loki: Hardening Code Obfuscation Against Automated Attacks cites Backward-Bounded DSE: Targeting Infeasibility Questions on Obfuscated Codes
- Loki: Hardening Code Obfuscation Against Automated Attacks cites NeuReduce: Reducing Mixed Boolean-Arithmetic Expressions by Recurrent Neural Network
- Loki: Hardening Code Obfuscation Against Automated Attacks cites MBA-Blast: Unveiling and Simplifying Mixed Boolean-Arithmetic Obfuscation
- NeuReduce: Reducing Mixed Boolean-Arithmetic Expressions by Recurrent Neural Network cites Backward-Bounded DSE: Targeting Infeasibility Questions on Obfuscated Codes
- MBA-Blast: Unveiling and Simplifying Mixed Boolean-Arithmetic Obfuscation cites Backward-Bounded DSE: Targeting Infeasibility Questions on Obfuscated Codes
Builds on5
- Backward-Bounded DSE: Targeting Infeasibility Questions on Obfuscated CodesSébastien Bardin, Robin David, Jean-Yves MarionS&P 2017 · 63 citations
- MBA-Blast: Unveiling and Simplifying Mixed Boolean-Arithmetic ObfuscationBinbin Liu, Junfu Shen, Jiang Ming, Qilong Zheng et al.USENIX Security 2021 · 37 citations
- NeuReduce: Reducing Mixed Boolean-Arithmetic Expressions by Recurrent Neural NetworkWeijie Feng, Binbin Liu, Dongpeng Xu, Qilong Zheng et al.EMNLP Findings 2020 · 14 citations
- Simplifying Mixed Boolean-Arithmetic Obfuscation by Program Synthesis and Term RewritingJaehyung Lee, Woosuk LeeCCS 2023 · 8 citations
- Loki: Hardening Code Obfuscation Against Automated AttacksMoritz Schloegel, Tim Blazytko, Moritz Contag, Cornelius Aschermann et al.USENIX Security 2022
Related papers
- Boosting SMT solver performance on mixed-bitwise-arithmetic expressionsDongpeng Xu, Binbin Liu, Weijie Feng, Jiang Ming et al.PLDI 2021 · 23 citations
- Syntia: Synthesizing the Semantics of Obfuscated CodeTim Blazytko, Moritz Contag, Cornelius Aschermann, Thorsten HolzUSENIX Security 2017 · 99 citations
- Large Language Models for Opaque Predicate Resolution: A Universal Control Flow Deobfuscation FrameworkXiao Chen, Qiuyun Wang, Shuwei Wang, Weize Zhang et al.FSE 2026
- Augmenting Search-based Program Synthesis with Local Inference Rules to Improve Black-box DeobfuscationVidal Attias, Nicolas Bellec, Grégoire Menguy, Sébastien Bardin et al.CCS 2025
- Redefining Grey-Box Deobfuscation through Context-Aware Program Synthesis, and Application to MBAsVidal Attias, Grégoire Menguy, Nicolas Bellec, Sébastien Bardin et al.CCS 2026
