Learning Parameterized Task Structure for Generalization to Unseen Entities

TL;DR

This paper introduces PSGI, a method that models hierarchical task structures with parameterized subtask graphs using first-order logic, enabling efficient learning and generalization to unseen entities and subtasks.

Contribution

The paper proposes PSGI, a novel approach that learns task dependencies with first-order logic and entity attributes, allowing zero-shot generalization to new subtasks and entities.

Findings

PSGI accurately learns latent task structures.

PSGI generalizes to unseen subtasks and entities.

Outperforms prior methods in hierarchical task learning.

Abstract

Real world tasks are hierarchical and compositional. Tasks can be composed of multiple subtasks (or sub-goals) that are dependent on each other. These subtasks are defined in terms of entities (e.g., "apple", "pear") that can be recombined to form new subtasks (e.g., "pickup apple", and "pickup pear"). To solve these tasks efficiently, an agent must infer subtask dependencies (e.g. an agent must execute "pickup apple" before "place apple in pot"), and generalize the inferred dependencies to new subtasks (e.g. "place apple in pot" is similar to "place apple in pan"). Moreover, an agent may also need to solve unseen tasks, which can involve unseen entities. To this end, we formulate parameterized subtask graph inference (PSGI), a method for modeling subtask dependencies using first-order logic with subtask entities. To facilitate this, we learn entity attributes in a zero-shot manner, which are used as quantifiers (e.g. "is_pickable(X)") for the parameterized subtask graph. We show this approach accurately learns the latent structure on hierarchical and compositional tasks more efficiently than prior work, and show PSGI can generalize by modelling structure on subtasks unseen during adaptation.