Counterfactual Reasoning and Learning Systems
This work addresses the challenge of optimizing interactive learning systems, such as ad placement, by providing a method for counterfactual reasoning, which is incremental in applying causal inference to a specific domain.
The paper tackles the problem of predicting the consequences of changes to complex learning systems by leveraging causal inference, enabling improved short-term and long-term performance, as demonstrated through experiments on the Bing search engine's ad placement system.
This work shows how to leverage causal inference to understand the behavior of complex learning systems interacting with their environment and predict the consequences of changes to the system. Such predictions allow both humans and algorithms to select changes that improve both the short-term and long-term performance of such systems. This work is illustrated by experiments carried out on the ad placement system associated with the Bing search engine.