ML Framework Reconstructing Binary Controllers to Interpretable Models
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing machine-learning frameworks fail to effectively reconstruct feedback controllers from binary files in a format interpretable by subject matter experts, leading to challenges in security and access to proprietary control techniques, and are sensitive to noisy data and require manual feature engineering.
Innovation Solution
A machine-learning framework that generates datasets through code generation processes, uses natural language processing-inspired transformers for model reconstruction, and supports multiple SME-interpretable modeling languages and computational architectures, enabling the transformation of binary code into readable source code and ensuring model interpretability.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of information
If existing machine-learning frameworks are used to reconstruct feedback controllers from binary files, then model reconstruction capability is achieved, but the output is not interpretable by subject matter experts and security vulnerabilities remain
Solution Approach 1:
The patent introduces an intermediary translation layer that converts binary controller code into human-readable source code formats (C, C++, Python, MATLAB). This intermediary representation maintains functional equivalence while enabling expert interpretation and verification, thus resolving the contradiction between reconstruction capability and interpretability without compromising security
Solution Approach 2:
The patent replaces traditional binary executable formats with high-level programming language representations. This substitution transforms the controller model from an opaque mechanical binary structure into an interpretable textual form that preserves functionality while enabling human analysis and verification of security properties
2Extent of automation
If traditional machine-learning frameworks are used for controller reconstruction, then reconstruction is performed, but the systems are sensitive to noisy data and require manual feature engineering
Solution Approach 1:
The patent implements self-service through automated feature extraction and model reconstruction pipelines that operate without manual intervention. The system automatically processes binary files, extracts relevant features, and generates interpretable models end-to-end, eliminating the need for manual feature engineering while maintaining high automation levels
Solution Approach 2:
The patent applies preliminary action by pre-processing binary controller files through automated analysis stages before reconstruction. The system performs preliminary feature extraction, noise filtering, and structure identification automatically, preparing the data for seamless model generation without requiring manual preprocessing steps
Data Source
AI summary
A method includes receiving a subject-matter expert (SME) interpretable model. The method further includes converting, by a processing device, the SME interpretable model into a functional mockup unit (FMU). The method further includes integrating the FMU into a control software project (CSP). The method further includes compiling the CSP into binary code.