Emergent Communication Pretraining for Few-Shot Machine Translation

Abstract

While state-of-the-art models that rely upon massively multilingual pretrained encoders achieve sample efficiency in downstream applications, they still require abundant amounts of unlabelled text. Nevertheless, most of the world's languages lack such resources. Hence, we investigate a more radical form of unsupervised knowledge transfer in the absence of linguistic data. In particular, for the first time we pretrain neural networks via emergent communication from referential games. Our key assumption is that grounding communication on images---as a crude approximation of real-world environments---inductively biases the model towards learning natural languages. On the one hand, we show that this substantially benefits machine translation in few-shot settings. On the other hand, this also provides an extrinsic evaluation protocol to probe the properties of emergent languages ex vitro. Intuitively, the closer they are to natural languages, the higher the gains from pretraining on them should be. For instance, in this work we measure the influence of communication success and maximum sequence length on downstream performances. Finally, we introduce a customised adapter layer and annealing strategies for the regulariser of maximum-a-posteriori inference during fine-tuning. These turn out to be crucial to facilitate knowledge transfer and prevent catastrophic forgetting. Compared to a recurrent baseline, our method yields gains of $59.0\%$$\sim$$147.6\%$ in BLEU score with only $500$ NMT training instances and $65.1\%$$\sim$$196.7\%$ with $1,000$ NMT training instances across four language pairs. These proof-of-concept results reveal the potential of emergent communication pretraining for both natural language processing tasks in resource-poor settings and extrinsic evaluation of artificial languages.

References in corpus (16)

• Adam: A Method for Stochastic Optimization
• Sequence to Sequence Learning with Neural Networks
• Empirical Evaluation of Gated Recurrent Neural Networks on Sequence Modeling
• Overcoming catastrophic forgetting in neural networks
• Categorical Reparameterization with Gumbel-Softmax
• Cross-lingual Language Model Pretraining
• The Concrete Distribution: A Continuous Relaxation of Discrete Random Variables
• Multilingual Denoising Pre-training for Neural Machine Translation
• MASS: Masked Sequence to Sequence Pre-training for Language Generation
• Learning multiple visual domains with residual adapters
• Phrase-Based & Neural Unsupervised Machine Translation
• Imagination improves Multimodal Translation
• Emergent Translation in Multi-Agent Communication
• From Zero to Hero: On the Limitations of Zero-Shot Cross-Lingual Transfer with Multilingual Transformers
• Ease-of-Teaching and Language Structure from Emergent Communication
• Universal linguistic inductive biases via meta-learning

Cited by in corpus (3)

• Towards More Human-like AI Communication: A Review of Emergent Communication Research
• Emergent Language: A Survey and Taxonomy
• On Bilingual Lexicon Induction with Large Language Models