• Proposes a novel cross-lingual deep learning approach inspired by Neural Machine Translation to compare binary basic blocks across different ISAs.
• Solves the cross-architecture code containment problem for the first time, going beyond function-level equivalence.
• Implements INNEREYE and demonstrates superior accuracy, efficiency, and scalability over existing basic-block comparison methods.
• Shows the effective transfer of NLP techniques to large-scale binary code analysis, enabling fine-grained vulnerability discovery and plagiarism detection.