Kyumin Lee

Kyumin Lee

Auto-generated profile, based on publicly available information

Kyumin Lee

Computer Science Department Worcester Polytechnic Institute Worcester MA USA

Research Focus & Metrics

Academic Career Length

9Years (2018-Present)

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

Representative Publications

View All Publications
1 December 2021 in Building and Environment
Verified
Journal Article
Closed access
XGXingtong Guo
Xingtong Guo
KLKyumin Lee
Kyumin Lee
ZWZhe Wang
Zhe Wang
SLShichao Liu
Shichao Liu

1 January 2023
Verified
Conference Paper
Open access
PDF file
YLYichuan Li
Yichuan Li
KDKaize Ding
Kaize Ding
KLKyumin Lee
Kyumin Lee
Yichuan Li, Kaize Ding, Kyumin Lee. Findings of the Association for Computational Linguistics: EMNLP 2023. 2023.

1 January 2024
Verified
Conference Paper
Open access
PDF file
YLYichuan Li
Yichuan Li
KDKaize Ding
Kaize Ding
JWJianling Wang
Jianling Wang
KLKyumin Lee
Kyumin Lee
Yichuan Li, Kaize Ding, Jianling Wang, Kyumin Lee. Findings of the Association for Computational Linguistics ACL 2024. 2024.


19 December 2023 in Proceedings of the ACM on Interactive, Mobile, Wearable and Ubiquitous Technologies
Verified
Journal Article
Open access
WGWen Ge
Wen Ge
GMGuanyi Mou
Guanyi Mou
EAEmmanuel Agu
Emmanuel Agu
KLKyumin Lee
Kyumin Lee
Human Activity Recognition (HAR) is a challenging, multi-label classification problem as activities may co-occur and sensor signals corresponding to the same activity may vary in different contexts (e.g., different device placements). This paper proposes a Deep Heterogeneous Contrastive Hyper-Graph Learning (DHC-HGL) framework that captures heterogenous Context-Aware HAR (CA-HAR) hypergraph properties in a message-passing and neighborhood-aggregation fashion. Prior work only explored homogeneous or shallow-node-heterogeneous graphs. DHC-HGL handles heterogeneous CA-HAR data by innovatively 1) Constructing three different types of sub-hypergraphs that are each passed through different custom HyperGraph Convolution (HGC) layers designed to handle edge-heterogeneity and 2) Adopting a contrastive loss function to ensure node-heterogeneity. In rigorous evaluation on two CA-HAR datasets, DHC-HGL significantly outperformed state-of-the-art baselines by 5.8% to 16.7% on Matthews Correlation Coefficient (MCC) and 3.0% to 8.4% on Macro F1 scores. UMAP visualizations of learned CA-HAR node embeddings are also presented to enhance model explainability. Our code is publicly available1 to encourage further research.

1 January 2023
Verified
Conference Paper
Open access
PDF file
HVHuma Varzgani
Huma Varzgani
NKNima Kordzadeh
Nima Kordzadeh
KLKyumin Lee
Kyumin Lee
Health misinformation on social media has become a major threat to users. To alleviate this issue, platforms such as Twitter have started labeling posts considered as misinformation to warn users. However, the effectiveness of such labels on user perceptions and actions are not clear, as it has not yet been examined by researchers in prior studies. We aim to address this gap through a model, which draws upon concepts from color theory and construal level theory and focuses on the impact of three misinformation label characteristics: background color of the label, abstractness of the message, and assertiveness of the message language. We propose that the effectiveness of these warning labels will lead users to verify, avoid using, and avoid sharing such labeled posts on social media. This paper provides important theoretical contributions and aids policymakers and platform providers by offering insights on what motivates users to take protective actions.