Sehyun Choi

Sehyun Choi

Auto-generated profile, based on publicly available information

Sehyun Choi

Department of Computer Science and Engineering, HKUST

Research Focus & Metrics

Academic Career Length

7Years (2020-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
1 January 2021
Verified
Conference Paper
Open access
PDF file
TFTianqing Fang
Tianqing Fang
WWWeiqi Wang
Weiqi Wang
SCSehyun Choi
Sehyun Choi
SHShibo Hao
Shibo Hao
HZHongming Zhang
Hongming Zhang
BHBin He
Bin He
Tianqing Fang, Weiqi Wang, Sehyun Choi, Shibo Hao, Hongming Zhang, Yangqiu Song, Bin He. Proceedings of the 2021 Conference on Empirical Methods in Natural Language Processing. 2021.

13 October 2023 in arXiv
Verified
Preprint
Open access
PDF file
SCSehyun Choi
Sehyun Choi
TFTianqing Fang
Tianqing Fang
ZWZhaowei Wang
Zhaowei Wang
YSYangqiu Song
Yangqiu Song
Large Language Models (LLMs) have demonstrated remarkable human-level natural language generation capabilities. However, their potential to generate misinformation, often called the hallucination problem, poses a significant risk to their deployment. A common approach to address this issue is to retrieve relevant knowledge and fine-tune the LLM with the knowledge in its input. Unfortunately, this method incurs high training costs and may cause catastrophic forgetting for multi-tasking models. To overcome these limitations, we propose a knowledge-constrained decoding method called KCTS (Knowledge-Constrained Tree Search), which guides a frozen LM to generate text aligned with the reference knowledge at each decoding step using a knowledge classifier score and MCTS (Monte-Carlo Tree Search). To adapt the sequence-level knowledge classifier to token-level guidance, we also propose a novel token-level hallucination detection method called RIPA (Reward Inflection Point Approximation). Our empirical results on knowledge-grounded dialogue and abstractive summarization demonstrate the strength of KCTS as a plug-and-play, model-agnostic decoding method that can effectively reduce hallucinations in natural language generation.

29 February 2020 in Journal of Agriculture & Life Science
Verified
Journal Article
Closed access
KKKi-Wan Kim
Ki-Wan Kim
SKSe-Hyuk Kim
Se-Hyuk Kim
SCSe-Hyun Choi
Se-Hyun Choi