Performance Evaluation for the Infer Engine–Optimization on Dynamic Advertising

Different from traditional online advertising, dynamic media campaigns show different impressions (visitors to a web page) with different advertisement based on the characteristics of the impression. Infer Engine is a system that optimizes bidding for media campaigns by maximizing total profit for every impression. The theoretical foundation of the Engine is a by statistics based learning theory for predictive models and mathematics analysis for optimization. The proposed project is on performance valuation of the Infer Engine with different learning algorithms. Comparison with other similar products will also be investigated. The metrics are the accuracy of the prediction and the learning time. The partner organization is the first company to apply these highly efficient techniques in this area. Performance analysis and comparisons will help the further improvement of the Engine.

Kevin Mak & Ben McInroy
Faculty Supervisor: 
Dr. Wenying Feng