Adversarial training for multi-context joint entity and relation extraction

EMNLP 2018 Giannis BekoulisJohannes DeleuThomas DemeesterChris Develder

Adversarial training (AT) is a regularization method that can be used to improve the robustness of neural network methods by adding small perturbations in the training data. We show how to use AT for the tasks of entity recognition and relation extraction... (read more)

PDF Abstract

Results from the Paper


TASK DATASET MODEL METRIC NAME METRIC VALUE GLOBAL RANK LEADERBOARD
Relation Extraction ACE 2004 multi-head + AT Entity+Relation F1 47.45 # 4
Entity F1 81.64 # 4
Relation Extraction ADE Corpus multi-head + AT Entity+Relation F1 75.52 # 3
Entity F1 86.73 # 3
Relation Extraction CoNLL04 multi-head + AT Entity F1 83.6 # 7
Relation F1 61.95 # 6
Relation F1 61.95 # 7

Methods used in the Paper


METHOD TYPE
🤖 No Methods Found Help the community by adding them if they're not listed; e.g. Deep Residual Learning for Image Recognition uses ResNet