Learning To Recombine And Resample Data For Compositional Generalization

Ekin Akyürek, Afra Feyza Akyürek, Jacob Andreas. Arxiv 2020 – 40 citations

[Paper]    
RAG Reinforcement Learning Training Techniques

Flexible neural sequence models outperform grammar- and automaton-based counterparts on a variety of tasks. However, neural models perform poorly in settings requiring compositional generalization beyond the training data – particularly to rare or unseen subsequences. Past work has found symbolic scaffolding (e.g. grammars or automata) essential in these settings. We describe R&R, a learned data augmentation scheme that enables a large category of compositional generalizations without appeal to latent symbolic structure. R&R has two components: recombination of original training examples via a prototype-based generative model and resampling of generated examples to encourage extrapolation. Training an ordinary neural sequence model on a dataset augmented with recombined and resampled examples significantly improves generalization in two language processing problems – instruction following (SCAN) and morphological analysis (SIGMORPHON 2018) – where R&R enables learning of new constructions and tenses from as few as eight initial examples.

Similar Work