Lean Classifier Model for Mobile Device Security
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing anti-virus, firewall, and encryption products for mobile devices are inadequate in identifying and addressing the complex factors contributing to performance degradation, often relying on resource-intensive scanning engines and limited to detecting known viruses, failing to efficiently restore devices to their original condition.
Innovation Solution
A behavior-based security system that monitors device behaviors, generates behavior vectors, and applies them to classifier models to classify behaviors as benign or non-benign, identifying and mitigating performance-degrading applications while informing users of the contributing factors, using a lean classifier model to balance resource usage and accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If resource-intensive scanning engines are used to detect malware, then detection capability is improved, but device performance and power consumption are degraded
Solution Approach 1:
The patent extracts only the essential classification functionality from complex full classifier models, creating lean classifier models that retain malware detection capability while removing unnecessary computational overhead. This extraction principle allows the system to maintain security effectiveness while significantly reducing resource consumption on mobile devices.
Solution Approach 2:
The patent changes the parameter of classifier model complexity by generating lean versions with reduced numbers of decision stumps and features. This parameter transformation maintains the core detection functionality while adapting the model to run efficiently on resource-constrained mobile devices, resolving the contradiction between detection capability and device performance.
2Measurement precision
If comprehensive behavior monitoring is implemented to identify non-benign behaviors, then security detection accuracy is improved, but processing resources and energy consumption increase
Solution Approach 1:
The patent applies partial action by implementing behavior monitoring for only the most critical features and using lean classifier models that evaluate a subset of all possible behaviors. This selective monitoring approach maintains sufficient classification accuracy to identify non-benign behaviors while consuming acceptable levels of processing resources and energy on mobile devices.
3Measurement precision
If complex classifier models are used to classify device behaviors, then classification accuracy is improved, but device complexity and resource requirements increase
Solution Approach 1:
The patent extracts the essential classification logic from complex full classifier models to create lean classifier models suitable for mobile devices. This extraction process removes unnecessary decision stumps and features while preserving the core functionality needed to classify behaviors as benign or non-benign, thereby reducing model complexity without significantly compromising accuracy.
Solution Approach 2:
The patent applies local quality by creating device-specific lean classifier models that are tailored to individual mobile devices based on their specific behavior patterns. This customization allows each device to have a simplified model optimized for its local characteristics, reducing overall system complexity while maintaining classification accuracy for that specific device context.
Data Source
AI summary
A computing device processor may be configured with processor-executable instructions to implement methods of detecting and responding non-benign behaviors of the computing device. The processor may be configured to monitor device behaviors to collect behavior information, generate a behavior vector information structure based on the collected behavior information, apply the behavior vector information structure to a classifier model to generate analysis results, use the analysis results to classify a behavior of the device, use the analysis results to determine the features evaluated by the classifier model that contributed most to the classification of the behavior, and select the top “n” (e.g., 3) features that contributed most to the classification of the behavior. The computing device may display the selected features on an electronic display of the computing device.