Multistep Inverse Is Not All You Need

By Alexander Levine, Peter Stone, and Amy Zhang

Reinforcement Learning Journal, vol. 2, 2024, pp. 884–925.

Presented at the Reinforcement Learning Conference (RLC), Amherst Massachusetts, August 9–12, 2024.


Download:

Abstract:

In real-world control settings, the observation space is often unnecessarily high-dimensional and subject to time-correlated noise. However, the *controllable* dynamics of the system are often far simpler than the dynamics of the raw observations. It is therefore desirable to learn an encoder to map the observation space to a simpler space of control-relevant variables. In this work, we consider the Ex-BMDP model, first proposed by Efroni et al. (2022), which formalizes control problems where observations can be factorized into an action-dependent latent state which evolves deterministically, and action-independent time-correlated noise. Lamb et al. (2022) proposes the ""AC-State"" method for learning an encoder to extract a complete action-dependent latent state representation from the observations in such problems. AC-State is a *multistep-inverse* method, in that it uses the encoding of the the first and last state in a path to predict the *first* action in the path. However, we identify cases where AC-State will fail to learn a correct latent representation of the agent-controllable factor of the state. We therefore propose a new algorithm, ACDF, which combines multistep-inverse prediction with a latent forward model. ACDF is guaranteed to correctly infer an action-dependent latent state encoder for a large class of Ex-BMDP models. We demonstrate the effectiveness of ACDF on tabular Ex-BMDPs through numerical simulations; as well as high-dimensional environments using neural-network-based encoders. Code is available at https://github.com/midi-lab/acdf.


Citation Information:

Alexander Levine, Peter Stone, and Amy Zhang. "Multistep Inverse Is Not All You Need." Reinforcement Learning Journal, vol. 2, 2024, pp. 884–925.

BibTeX:

@article{levine2024multistep,
    title={Multistep Inverse Is Not All You Need},
    author={Levine, Alexander and Stone, Peter and Zhang, Amy},
    journal={Reinforcement Learning Journal},
    volume={2},
    pages={884--925},
    year={2024}
}