Generalizing to New Tasks via One-Shot Compositional Subgoals

TL;DR

This paper introduces CASE, a method that improves generalization and learning efficiency in long-horizon tasks by using adaptive subgoals in a learned latent space, enabling one-shot task generalization from a single reference trajectory.

Contribution

The paper presents a novel approach using compositional arithmetic in latent space for adaptive subgoals, enhancing one-shot generalization and efficiency in complex tasks.

Findings

Outperforms previous state-of-the-art by 30%

Enables one-shot generalization with a single reference trajectory

Improves learning efficiency in long-horizon tasks

Abstract

The ability to generalize to previously unseen tasks with little to no supervision is a key challenge in modern machine learning research. It is also a cornerstone of a future "General AI". Any artificially intelligent agent deployed in a real world application, must adapt on the fly to unknown environments. Researchers often rely on reinforcement and imitation learning to provide online adaptation to new tasks, through trial and error learning. However, this can be challenging for complex tasks which require many timesteps or large numbers of subtasks to complete. These "long horizon" tasks suffer from sample inefficiency and can require extremely long training times before the agent can learn to perform the necessary longterm planning. In this work, we introduce CASE which attempts to address these issues by training an Imitation Learning agent using adaptive "near future" subgoals. These subgoals are recalculated at each step using compositional arithmetic in a learned latent representation space. In addition to improving learning efficiency for standard long-term tasks, this approach also makes it possible to perform one-shot generalization to previously unseen tasks, given only a single reference trajectory for the task in a different environment. Our experiments show that the proposed approach consistently outperforms the previous state-of-the-art compositional Imitation Learning approach by 30%.