Search Results for « Robust Causal Bandits »

We study a contextual bandit setting where the agent has access to causal side information, in addition to the ability to perform multiple targeted experiments corresponding to potentially different context-action pairs—simultaneously in one-shot within a budget. This new formalism provides a natural model for several real-world scenarios where parallel targeted experiments can be conducted and where some domain knowledge of causal relationships is available. We propose a new algorithm that utilizes a novel entropy-like measure that we introduce. We perform several experiments, both using purely synthetic data and using a real-world dataset. In addition, we study sensitivity of our algorithm's performance to various aspects of the problem setting. The results show that our algorithm performs better than baselines in all of the experiments. We also show that the algorithm is sound; that is, as budget increases, the learned policy eventually converges to an optimal policy. Further, we theoretically bound our algorithm's regret under additional assumptions. Finally, we provide ways to achieve two popular notions of fairness, namely counterfactual fairness and demographic parity, with our algorithm.

Causal knowledge is sought after throughout data-driven fields due to its explanatory power and potential value to inform decision-making. If the targeted system is well-understood in terms of its causal components, one is able to design more precise and surgical interventions so as to bring certain desired outcomes about. The idea of leveraging the causal understanding of a system to improve decision-making has been studied in the literature under the rubric of structural causal bandits (Lee and Bareinboim, 2018). In this setting, (1) pulling an arm corresponds to performing a causal intervention on a set of variables, while (2) the associated rewards are governed by the underlying causal mechanisms. One key assumption of this work is that any observed variable (X) in the system is manipulable, which means that intervening and making do(X = x) is always realizable. In many real-world scenarios, however, this is a too stringent requirement. For instance, while scientific evidence may support that obesity shortens life, it’s not feasible to manipulate obesity directly, but, for example, by decreasing the amount of soda consumption (Pearl, 2018). In this paper, we study a relaxed version of the structural causal bandit problem when not all variables are manipulable. Specifically, we develop a procedure that takes as argument partially specified causal knowledge and identifies the possibly-optimal arms in structural bandits with non-manipulable variables. We further introduce an algorithm that uncovers non-trivial dependence structure among the possibly-optimal arms. Finally, we corroborate our findings with simulations, which shows that MAB solvers enhanced with causal knowledge and leveraging the newly discovered dependence structure among arms consistently outperform causal-insensitive solvers.