Contextual Preference Distribution Learning
This addresses risk-averse decision-making problems in settings like ridesharing where preferences vary with context, representing an incremental improvement over existing methods.
The paper tackled uncertainty in decision-making from heterogeneous and context-dependent human preferences by proposing a sequential learning-and-optimization pipeline to learn preference distributions for risk-averse formulations, reducing average post-decision surprise by up to 114× compared to risk-neutral approaches and up to 25× compared to baselines in a synthetic ridesharing environment.
Decision-making problems often feature uncertainty stemming from heterogeneous and context-dependent human preferences. To address this, we propose a sequential learning-and-optimization pipeline to learn preference distributions and leverage them to solve downstream problems, for example risk-averse formulations. We focus on human choice settings that can be formulated as (integer) linear programs. In such settings, existing inverse optimization and choice modelling methods infer preferences from observed choices but typically produce point estimates or fail to capture contextual shifts, making them unsuitable for risk-averse decision-making. Using a bounded-variance score function gradient estimator, we train a predictive model mapping contextual features to a rich class of parameterizable distributions. This approach yields a maximum likelihood estimate. The model generates scenarios for unseen contexts in the subsequent optimization phase. In a synthetic ridesharing environment, our approach reduces average post-decision surprise by up to 114$\times$ compared to a risk-neutral approach with perfect predictions and up to 25$\times$ compared to leading risk-averse baselines.