Yan Sun (孙彦)

Yan Sun (孙彦)

PhD Candidate, National Univesrity of Singapore

Publications

You can also find my articles on my Google Scholar profile.

Conference Papers

Unsupervised Multiple Kernel Learning for Graphs via Ordinality Preservation

Published in The Thirteen International Conference on Learning Representations, 2025

Graph Metric Learning: Proposes UMKL-G, which combines multiple graph kernels in an unsupervised way, preserving the ordinal relationships between graphs.

Citation: Sun, Y., & Kok, S. (2025). Unsupervised Multiple Kernel Learning for Graphs via Ordinality Preservation. In The Thirteen International Conference on Learning Representations.
Download Paper

MMD Graph Kernel: Effective Metric Learning for Graphs via Maximum Mean Discrepancy

Published in The Twelfth International Conference on Learning Representations (Spotlight), 2024

Graph Metric Learning: Proposes MMD-GK, a graph kernel method with unsupervised and supervised variants, achieving superior results in graph clustering and classification tasks.

Citation: Sun, Y., & Fan, J. (2024). MMD Graph Kernel: Effective Metric Learning for Graphs via Maximum Mean Discrepancy. In The Twelfth International Conference on Learning Representations.
Download Paper

Deep orthogonal hypersphere compression for anomaly detection

Published in The Twelfth International Conference on Learning Representations (Spotlight), 2024

Anomaly Detection: Introduces hypersphere-based methods (DOHSC, DO2HSC) for anomaly detection, including graph-level cases, offering compact decision regions and improved accuracy.

Citation: Zhang, Y., Sun, Y., Cai, J., & Fan, J. (2024). Deep orthogonal hypersphere compression for anomaly detection. In The Twelfth International Conference on Learning Representations.
Download Paper

Journal Articles

Laplacian-based Cluster-Contractive t-SNE for High-Dimensional Data Visualization

Published in ACM Transactions on Knowledge Discovery from Data, 2023

Dimension Reduction: Enhances t-SNE by leveraging graph Laplacian for better cluster preservation, outperforming t-SNE and UMAP in structure and visualization.

Citation: Sun, Y., Han, Y., & Fan, J. (2023). Laplacian-based Cluster-Contractive t-SNE for High-Dimensional Data Visualization. ACM Transactions on Knowledge Discovery from Data, 18(1), 1-22.
Download Paper