Data Augmentation for Learning to Play in Text-Based Games

Abstract

Improving generalization in text-based games serves as a useful stepping-stone towards reinforcement learning (RL) agents with generic linguistic ability. Data augmentation for generalization in RL has shown to be very successful in classic control and visual tasks, but there is no prior work for text-based games. We propose Transition-Matching Permutation, a novel data augmentation technique for text-based games, where we identify phrase permutations that match as many transitions in the trajectory data. We show that applying this technique results in state-of-the-art performance in the Cooking Game benchmark suite for text-based games.

Cite

Text

Kim and Kim. "Data Augmentation for Learning to Play in Text-Based Games." International Joint Conference on Artificial Intelligence, 2022. doi:10.24963/IJCAI.2022/436

Markdown

[Kim and Kim. "Data Augmentation for Learning to Play in Text-Based Games." International Joint Conference on Artificial Intelligence, 2022.](https://mlanthology.org/ijcai/2022/kim2022ijcai-data/) doi:10.24963/IJCAI.2022/436

BibTeX

@inproceedings{kim2022ijcai-data,
  title     = {{Data Augmentation for Learning to Play in Text-Based Games}},
  author    = {Kim, Jinhyeon and Kim, Kee-Eung},
  booktitle = {International Joint Conference on Artificial Intelligence},
  year      = {2022},
  pages     = {3143-3149},
  doi       = {10.24963/IJCAI.2022/436},
  url       = {https://mlanthology.org/ijcai/2022/kim2022ijcai-data/}
}