Malware Detection Model Testing via Time-Shifted Dataset Segmentation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current malware detection systems lack regimented neural network model testing procedures, which hinders the assurance of correct functioning in detecting and classifying malware.
Innovation Solution
A method and apparatus for testing a malware detection machine learning model by training it with a first dataset and then testing it using a time-shifted version of the same dataset, allowing the model to learn from new malware samples and repeat the process until all samples are used.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If malware detection models are trained using traditional datasets without time-shifting, then the training process is simpler and faster, but the model's ability to detect new and evolving malware is compromised
Solution Approach 1:
The patent applies preliminary action by performing time-shifting on the dataset before testing the model. The dataset is pre-processed to create time-shifted versions that simulate future malware threats, allowing the model to be tested on data it hasn't seen before in a realistic temporal context. This preliminary preparation of the test data ensures the model is properly validated against evolving threats.
2Reliability
If regimented neural network model testing procedures are implemented, then model reliability is improved, but the testing process becomes more complex and time-consuming
Solution Approach 1:
The patent implements periodic action by repeatedly testing the model on multiple time-shifted versions of the dataset. The testing process is structured to systematically apply different time-shifted datasets in sequence, allowing comprehensive validation of the model's ability to detect malware across different time periods and threat evolutions, while maintaining an organized and efficient testing rhythm.
3Productivity
If the model is tested on the same dataset used for training, then the testing process is simpler, but the model's effectiveness against new malware cannot be properly evaluated
Solution Approach 1:
The patent applies segmentation by dividing the original dataset into multiple time-based segments and creating time-shifted versions. The dataset is segmented into training portions and testing portions with different temporal characteristics, allowing the model to be trained on one time period's malware patterns and tested on another time period's patterns, thereby validating its ability to generalize to new threats.
Data Source
AI summary
A method and apparatus for testing a malware detection machine learning model. The method trains a malware detection model using a first dataset containing malware samples from a particular time period. The trained model is then tested using a second dataset that is a time shifted version of the first dataset.