Abstract
Generalization outside the scope of one's training data requires leveragingprior knowledge about the effects that transfer, and the effects that don't,between different data sources. Bayesian transfer learning is a principledparadigm for specifying this knowledge, and refining it on the basis of datafrom the source (training) and target (prediction) tasks. We address thechallenging transfer learning setting where the learner (i) cannot fine-tune inthe target task, and (ii) does not know which source data points correspond tothe same task (i.e., the data sources are unknown). We propose a proxy-informedrobust method for probabilistic transfer learning (PROMPT), which provides aposterior predictive estimate tailored to the structure of the target task,without requiring the learner have access to any outcome information from thetarget task. Instead, PROMPT relies on the availability of proxy information.PROMPT uses the same proxy information for two purposes: (i) estimation ofeffects specific to the target task, and (ii) construction of a robustreweighting of the source data for estimation of effects that transfer betweentasks. We provide theoretical results on the effect of this reweighting on therisk of negative transfer, and demonstrate application of PROMPT in twosynthetic settings.