Insertion-based Decoding with Automatically Inferred Generation Order

• Published in: Transactions of the Association for Computational Linguistics Vol. 7; pp. 661 - 676
• Main Authors: Gu, Jiatao, Liu, Qi, Cho, Kyunghyun
• Format: Journal Article
• Language: English
• Published: One Rogers Street, Cambridge, MA 02142-1209, USA MIT Press 01.11.2019; MIT Press Journals, The; The MIT Press
• Subjects: Algorithms; Decoding; Insertion; Linguistics; Machine translation; Natural language generation; Word order
• ISSN: 2307-387X; 2307-387X
• DOI: 10.1162/tacl_a_00292

Conventional neural autoregressive decoding commonly assumes a fixed left-to-right generation order, which may be sub-optimal. In this work, we propose a novel decoding algorithm— InDIGO—which supports flexible sequence generation in arbitrary orders through insertion operations. We extend Transformer, a state-of-the-art sequence generation model, to efficiently implement the proposed approach, enabling it to be trained with either a pre-defined generation order or adaptive orders obtained from beam-search. Experiments on four real-world tasks, including word order recovery, machine translation, image caption, and code generation, demonstrate that our algorithm can generate sequences following arbitrary orders, while achieving competitive or even better performance compared with the conventional left-to-right generation. The generated sequences show that InDIGO adopts adaptive generation orders based on input information.