Cost-aware prediction service pricing with incomplete information

Huanhuan Peng, Xiaoye Miao, Jinshan Zhang, Yunjun Gao, Shuiguang Deng, Jianwei Yin

Abstract: Trading the machine learning-based prediction services has been up-and-coming for individuals and small companies. It serves to directly provide the predictions, e.g., classifications, for consumers without domain knowledge. Existing prediction service pricing methods closely rely on the strong assumption of completely known information on service quality and consumers’ valuations. In this paper, we study the profit maximization problem of pricing prediction services under incomplete information for the first time. We propose a novel Service Market model, named SMELT, considering multiple types of customers with dEmand and quaLity-aware valuaTions. We first derive the theoretical optimal solution to maximize service profit with complete information. Then, we develop an effective framework PSPricer under the profit ratio guarantee to solve the profit maximization problem with incomplete information. It is capable of not only efficiently getting the sub-optimal service price with bounded revenue loss, but also effectively learning the service quality function with the maximum likelihood estimation. Moreover, due to the resource-intensive and costly characteristic of machine learning model inference, we further extend the SMELT model to consider the inevitable inference cost in service trading. We formulate a novel cost-aware profit maximization problem and derive the general optimal solution. The PSPricer framework is tailored with an effective heuristic to maximize the cost-aware profit with the theoretical profit ratio guarantee. Extensive experiments on real-life datasets demonstrate our theoretical findings and the effectiveness and efficiency of PSPricer in various settings, compared with the state-of-the-art approaches.