Privacy-Preserving Click Pattern Anomaly Detection for Mobile In-App Browser Advertising Fraud

Abstract

Mobile in-app browser advertising fraud poses significant economic threats to digital marketing ecosystems, costing advertisers billions annually through sophisticated click manipulation schemes. This research presents a privacy-preserving anomaly detection framework specifically designed for identifying fraudulent click patterns within mobile in-app browser environments. The proposed methodology integrates differential privacy mechanisms with temporal sequence analysis to detect abnormal user interaction patterns while maintaining user privacy compliance. Through comprehensive evaluation on real-world advertising datasets containing 2.3 million click events, our approach achieves 94.7% detection accuracy with minimal privacy budget consumption. The framework analyzes multi-dimensional features including click timing intervals, touch pressure distributions, and device sensor signals to distinguish genuine user interactions from automated fraud attempts. Experimental results demonstrate superior performance compared to existing methods while ensuring ε-differential privacy guarantees, achieving optimal balance between detection effectiveness and privacy protection in mobile advertising environments.