摘要：We consider the problem of displaying commer- cial advertisements on web pages, in the ＂cost per click＂ model. The advertisement server has to learn the appeal of each type of visitor for the different advertisements in order to maximize the profit. Advertisements have constraints such as a certain number of clicks to draw, as well as a lifetime. This problem is thus inherently dynamic, and intimately com- bines combinatorial and statistical issues. To set the stage, it is also noteworthy that we deal with very rare events of in- terest, since the base probability of one click is in the or- der of 10-4. Different approaches may be thought of, rang- ing from computationally demanding ones （use of Markov decision processes, or stochastic programming） to very fast ones. We introduce NOSEED, an adaptive policy learning al- gorithm based on a combination of linear programming and multi-arm bandits. We also propose a way to evaluate the extent to which we have to handle the constraints （which is directly related to the computation cost）. We investigate the performance of our system through simulations on a realistic model designed with an important commercial web actor.

摘要：An adaptable product can satisfy different customer requirements by changing its configuration and parameter values during the operation stage. Design of adaptable products aims at reducing the environment impact through replacement of multiple different products with single adaptable ones. Due to the complex architec- ture, multiple functional requirements, and changes of product configurations and parameter values in operation, impact of uncertainties to the functional performance measures needs to be considered in design of adaptable products. In this paper, a robust design approach is introduced to identify the optimal design configuration and parameters of an adaptable product whose functional performance measures are the least sensitive to uncertain- ties. An adaptable product in this paper is modeled by both configurations and parameters. At the configuration level, methods to model different product configuration candi- dates in design and different product configuration states in operation to satisfy design requirements are introduced. At the parameter level, four types of product/operating parameters and relations among these parameters are discussed. A two-level optimization approach is developed to identify the optimal design configuration and its parameter values of the adaptable product. A case study is implemented to illustrate the effectiveness of the newly developed robust adaptable design method.