Imitation Learning as $f$-Divergence Minimization
This addresses imitation learning for robotics or AI systems where demonstrations are multi-modal, offering a more robust method compared to existing approaches, though it is incremental as it builds on prior divergence-based methods.
The paper tackles the problem of imitation learning with multi-modal demonstrations by proposing a framework that minimizes f-divergences, showing that using reverse KL divergence (I-projection) imitates multi-modal behaviors more reliably than GAIL and behavior cloning, with empirical results demonstrating improved performance.
We address the problem of imitation learning with multi-modal demonstrations. Instead of attempting to learn all modes, we argue that in many tasks it is sufficient to imitate any one of them. We show that the state-of-the-art methods such as GAIL and behavior cloning, due to their choice of loss function, often incorrectly interpolate between such modes. Our key insight is to minimize the right divergence between the learner and the expert state-action distributions, namely the reverse KL divergence or I-projection. We propose a general imitation learning framework for estimating and minimizing any f-Divergence. By plugging in different divergences, we are able to recover existing algorithms such as Behavior Cloning (Kullback-Leibler), GAIL (Jensen Shannon) and Dagger (Total Variation). Empirical results show that our approximate I-projection technique is able to imitate multi-modal behaviors more reliably than GAIL and behavior cloning.