Learning Task Decomposition with Ordered Memory Policy Network



Published on


Many complex real-world tasks are composed of several levels of sub-tasks. Humans leverage these hierarchical structures to accelerate the learning process and achieve better generalization. In this work, we study the inductive bias and propose Ordered Memory Policy Network (OMPN) to discover subtask hierarchy by learning from demonstration. The discovered subtask hierarchy could be used to perform task decomposition, recovering the subtask boundaries in an unstruc-tured demonstration. Experiments on Craft and Dial demonstrate that our modelcan achieve higher task decomposition performance under both unsupervised and weakly supervised settings, comparing with strong baselines. OMPN can also bedirectly applied to partially observable environments and still achieve higher task decomposition performance. Our visualization further confirms that the subtask hierarchy can emerge in our model.

This paper has been published at ICLR 2021

Please cite our work using the BibTeX below.

      title={Learning Task Decomposition with Ordered Memory Policy Network}, 
      author={Yuchen Lu and Yikang Shen and Siyuan Zhou and Aaron Courville and Joshua B. Tenenbaum and Chuang Gan},
Close Modal