Cross-Platform Program Analysis Using Universal Logical Features
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Manual program analysis is time-consuming and impractical for large numbers of applications across different platforms, such as Android and iOS, due to platform-specific features that hinder effective machine learning-based classification of malicious or benign programs.
Innovation Solution
A method using machine learning to train discriminatory models on platform-independent logical features extracted from programs across different platforms, enabling cross-platform analysis by converting platform-specific features into universal features for distinguishing between labeled and unlabeled programs.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If manual program analysis is performed to accurately distinguish between malicious and benign programs, then classification accuracy is improved, but analysis time and resource consumption increase significantly
Solution Approach 1:
The system enables programs to be automatically analyzed through self-service mechanisms where the machine learning model independently processes and classifies programs without requiring manual expert intervention for each program, thus achieving high classification accuracy while minimizing time loss
Solution Approach 2:
The invention transforms program analysis from manual expert review to automated machine learning classification by changing the operational parameters from human-driven sequential analysis to algorithm-driven parallel processing, maintaining accuracy through trained models while dramatically reducing analysis time
2Measurement precision
If machine learning models are trained on platform-specific features to improve classification accuracy, then discriminatory performance is improved, but cross-platform applicability deteriorates
Solution Approach 1:
The system achieves universality by training machine learning models on platform-independent logical features that can be extracted from any platform's programs. The model learns from Android, iOS, and other platform programs using the same feature extraction and classification methodology, making the discriminatory performance applicable across all platforms without requiring platform-specific model training
Solution Approach 2:
The invention introduces platform-independent logical features as an intermediary layer between platform-specific program implementations and the machine learning model. This intermediary transformation allows the model to receive standardized feature inputs from different platforms, enabling cross-platform discriminatory performance without direct platform-specific training
3Measurement precision
If extensive manual program analysis is performed to create training data, then model training accuracy is improved, but data preparation time and expert effort increase
Solution Approach 1:
The system performs preliminary automated feature extraction from programs before model training, creating structured training datasets through automated processing rather than manual annotation. This preliminary action prepares the training data in advance using algorithmic feature extraction, maintaining model training accuracy while eliminating the need for time-consuming manual data preparation by experts
Data Source
AI summary
A method for performing program analysis includes receiving programs of a first platform that have been assigned a first label and programs of the first platform that have been assigned a second label. Each of the programs of the first platform is expressed as platform-independent logical features. A discriminatory model or classifier is trained, using machine learning, based on the expression of the programs of the first platform as platform-independent logical features, to distinguish between programs of the first label and programs of the second label. An unlabeled program of a second platform is received and is expressed as platform-independent logical features. The trained discriminatory model or classifier is used to determine if the unlabeled program warrants the first label or the second label, based on the expression of the unlabeled program as platform-independent logical features.