Abstract | ||
---|---|---|
Meta-learning considers the problem of learning an efficient learning process that can leverage its past experience to accurately solve new tasks. However, the efficacy of meta-learning crucially depends on the distribution of tasks available for training, and this is often assumed to be known a priori or constructed from limited supervised datasets. In this work, we aim to provide task distributions for meta-learning by considering self-supervised tasks automatically proposed from unlabeled text, to enable large-scale meta-learning in NLP. We design multiple distributions of self-supervised tasks by considering important aspects of task diversity, difficulty, type, domain, and curriculum, and investigate how they affect meta-learning performance. Our analysis shows that all these factors meaningfully alter the task distribution, some inducing significant improvements in downstream few-shot accuracy of the meta-learned models. Empirically, results on 20 downstream tasks show significant improvements in few-shot learning -- adding up to +4.2% absolute accuracy (on average) to the previous unsupervised meta-learning method, and perform comparably to supervised methods on the FewRel 2.0 benchmark. |
Year | Venue | DocType |
---|---|---|
2021 | EMNLP | Conference |
Volume | Citations | PageRank |
2021.emnlp-main | 0 | 0.34 |
References | Authors | |
0 | 5 |
Name | Order | Citations | PageRank |
---|---|---|---|
Trapit Bansal | 1 | 131 | 8.33 |
Karthick Prasad Gunasekaran | 2 | 0 | 0.34 |
Tong Wang | 3 | 85 | 10.63 |
Tsendsuren Munkhdalai | 4 | 169 | 13.49 |
Andrew Kachites McCallumzy | 5 | 19203 | 1588.22 |