Adaptive Advertising Algorithm for Ad Optimization

USC EE Master Term Paper: Proposed a self-learning adaptive algorithm to automatically optimize the visual display of online advertisements (e.g., Google Adsense™) to maximize the Click-Through Rate (CTR).

* The algorithm eliminates the need for manual ad customization by adjusting display parameters (placement, size, colors) based on web analytics and CTR results.
* A sensitivity-matrix-based optimization approach is used to tune correlated ad parameters simultaneously.
* A Global/Local database structure and a central codebook are implemented to efficiently store, group, and search for optimized parameter settings based on specific web analytic groups (e.g., new vs. returning users).
* Optimization is systematically tracked and guided by an Ad Parameter Tree.