KJCKorea
Journal Central

@article{ART002970003},
author={Euhee Kim},
title={Analysis of dynamic LSTM network activation in LSTM English learner language model: Subject-Verb Number Agreement task},
journal={Journal of The Korea Society of Computer and Information},
issn={1598-849X},
year={2023},
volume={28},
number={6},
pages={13-21},
doi={10.9708/jksci.2023.28.06.013}

TY - JOUR
AU - Euhee Kim
TI - Analysis of dynamic LSTM network activation in LSTM English learner language model: Subject-Verb Number Agreement task
JO - Journal of The Korea Society of Computer and Information
PY - 2023
VL - 28
IS - 6
PB - The Korean Society Of Computer And Information
SP - 13
EP - 21
SN - 1598-849X
AB - In this paper, we propose an approach to analyze the dynamic LSTM network activation in an LSTM English learner language model trained on a large corpus of English-speaking learners. The objective is to examine the relationship between network activation and performance on the Subject-Verb Number Agreement task. By employing a NA-task probing classifier and conducting ablation experiments, activation patterns are evaluated at each time step. The results reveal a strong link between network activation and the classifier's performance. The proposed model achieved 99.57% accuracy on the evaluation dataset for the NA-task, demonstrating its acquisition of correct grammar rules and accurate prediction ability. To analyze the influence of internal neurons on NA-task processing, specific LSTM neurons are removed and the model's performance is examined. Removing neuron 776 resulted in a more than 10% decrease in performance for plural subjects, while removing neuron 988 led to a 10% decrease in performance for singular subjects compared to the model before removal.
KW - probing classifier;LSTM language model;NLP;LSTM network;dynamic activation
DO - 10.9708/jksci.2023.28.06.013
ER -

Euhee Kim. (2023). Analysis of dynamic LSTM network activation in LSTM English learner language model: Subject-Verb Number Agreement task. Journal of The Korea Society of Computer and Information, 28(6), 13-21.

Euhee Kim. 2023, "Analysis of dynamic LSTM network activation in LSTM English learner language model: Subject-Verb Number Agreement task", Journal of The Korea Society of Computer and Information, vol.28, no.6 pp.13-21. Available from: doi:10.9708/jksci.2023.28.06.013

Euhee Kim "Analysis of dynamic LSTM network activation in LSTM English learner language model: Subject-Verb Number Agreement task" Journal of The Korea Society of Computer and Information 28.6 pp.13-21 (2023) : 13.

Euhee Kim. Analysis of dynamic LSTM network activation in LSTM English learner language model: Subject-Verb Number Agreement task. 2023; 28(6), 13-21. Available from: doi:10.9708/jksci.2023.28.06.013

Euhee Kim. "Analysis of dynamic LSTM network activation in LSTM English learner language model: Subject-Verb Number Agreement task" Journal of The Korea Society of Computer and Information 28, no.6 (2023) : 13-21.doi: 10.9708/jksci.2023.28.06.013

Euhee Kim. Analysis of dynamic LSTM network activation in LSTM English learner language model: Subject-Verb Number Agreement task. Journal of The Korea Society of Computer and Information, 28(6), 13-21. doi: 10.9708/jksci.2023.28.06.013

Euhee Kim. Analysis of dynamic LSTM network activation in LSTM English learner language model: Subject-Verb Number Agreement task. Journal of The Korea Society of Computer and Information. 2023; 28(6) 13-21. doi: 10.9708/jksci.2023.28.06.013

Euhee Kim. Analysis of dynamic LSTM network activation in LSTM English learner language model: Subject-Verb Number Agreement task. 2023; 28(6), 13-21. Available from: doi:10.9708/jksci.2023.28.06.013

Euhee Kim. "Analysis of dynamic LSTM network activation in LSTM English learner language model: Subject-Verb Number Agreement task" Journal of The Korea Society of Computer and Information 28, no.6 (2023) : 13-21.doi: 10.9708/jksci.2023.28.06.013