Description
Multitask Soft Option Learning
Maximilian Igl (Oxford)*; Andrew Gambardella (University of Oxford); Jinke He (Delft University of Technology); Nantas Nardelli (University of Oxford); N Siddharth (Unversity of Oxford); Wendelin Boehmer (University of Oxford); Shimon Whiteson (University of Oxford)
We present Multitask Soft Option Learning (MSOL), a hierarchical multitask framework based on Planning as Inference. MSOL extends the concept of options, using separate variational posteriors for each task, regularized by a shared prior. This “soft” version of options avoids several instabilities during training in a multitask setting, and provides a natural way to learn both intra-option policies and their terminations. Furthermore, it allows fine-tuning of options for new tasks without forgetting their learned policies, leading to faster training without reducing the expressiveness of the hierarchical policy. We demonstrate empirically that MSOL significantly outperforms both hierarchical and flat transfer-learning baselines.