Hwanjun Song
KAIST
E2-3110, 291 Daehak-ro, Yuseong-gu Email: songhwanjunxkxkxk@kaist.ac.kr / ghkswns91xkxkxk@gmail.com |
Mar 2024: Three research papers on multimodal data creation, hallucination evaluation, semi-supervised text summarization was accepted at the main track of NAACL.
Jan 2024: We received a silver prize at the Samsung Humantech Paper Awards (Signal Processing/NLP).
Jan 2024: A Paper on 'Time-series Anomaly Detection' accepted at TheWebConf (WWW) 2024.
Jan 2024: Papers on 'Noisy Labels' and 'Continual Learning' accepted at AAAI 2024.
See our Lab Webpage.
Publications Google scholar profile An underline indicates the corresponding author.H. Aboutalebi, H. Song, Y. Xie, A. Gupta, J. Sun, H. Su, I. Shalyminov, N. Pappas, S. Signh, S. Mansour. MAGID: An Automated Pipeline for Generating Synthetic Multi-modal Datasets. Annual Conference of the North American Chapter of the Association for Computational Linguistics (NAACL-main) 2024. [pdf] | |
L. Tang, I. Shalyminov, A. W. Wong, J. Burnsky, J. W Vincent, Y. Yang, S. Singh, S. Feng, H. Song, H. Su, L. Sun, Y. Zhang, S. Mansour, K. McKeown. TofuEval: Evaluating Hallucinations of LLMs on Topic-Focused Dialogue Summarization. Annual Conference of the North American Chapter of the Association for Computational Linguistics (NAACL-main) 2024. [pdf] | |
J. He, H, Su, J. Cai, I, Shalyminov, H. Song, S. Mansour. Semi-Supervised Dialogue Abstractive Summarization via High-Quality Pseudolabel Selection. Annual Conference of the North American Chapter of the Association for Computational Linguistics (NAACL-main) 2024. | |
Y. Nam, S. Yoon, Y. Shin, M. Bae, H. Song, JG. Lee, BS. Lee M. Breaking the Time-Frequency Granularity Discrepancy in Time-Series Anomaly Detection. TheWebConf (WWW) 2024. To Appear | |
H. Song, M. Kim, JG. Lee. Toward Robustness in Multi-label Classification: A Data Augmentation Strategy against Imbalance and Noise. The AAAI Conference on Artificial Intelligence (AAAI) 2024. [pdf] | |
D. Kim, D. Park, Y. Shin, H. Song, JG. Lee. Adaptive Shortcut Debiasing for Online Continual Learning. The AAAI Conference on Artificial Intelligence (AAAI) 2024. |
H. Song, I. Shalyminov, H. Su, S. Singh, K. Yao, S. Mansour. Enhancing Abstractiveness of Summarization Models through Calibrated Distillation. International Conference on Empirical Methods in Natural Language Processing (EMNLP Findings) 2023. [Amazon Science Blog] | |
S. Bae, J. Ko, H. Song, SY. Yun. Fast and Robust Early-Exiting Framework for Autoregressive Language Models with Synchronized Parallel Decoding. International Conference on Empirical Methods in Natural Language Processing (EMNLP main) 2023. | |
D. Park, S. Choi, D. Kim, H. Song, JG. Lee. Robust Data Pruning under Label Noise via Maximizing Re-labeling Accuracy. Annual Conference on Neural Information Processing Systems (NeurIPS) 2023. To Appear | |
J. Lee, Y. Kwon, J. Park, M. Yu, S. Park, H. Song Q-HyViT: Post-Training Quantization for Hybrid Vision Transformer with Bridge Block Reconstruction. Preprint arXiv 2023. [pdf] | |
D. Jung, D. Han, J. Bang, H. Song. Generating Instance-level Prompts for Rehearsal-free Continual Learning. International Conference on Computer Vision (ICCV) 2023. Oral Presentation. | |
Y. Shin, S. Yoon, H. Song, D. Park, B. Kim, JG. Lee, BS. Lee. Context Consistency Regularization for Label Sparsity in Time Series. International Conference on Machine Learning (ICML) 2023. [pdf] | |
H. Song, J. Bang. Prompt-Guided Transformers for End-to-End Open-Vocabulary Object Detection. Preprint arXiv 2023. [pdf] [code] | |
S. Kim, S, Bae, H. Song, SY. Yun. Re-thinking Federated Active Learning based on Inter-class Diversity. International Conference on Computer Vision and Pattern Recognition (CVPR) 2023. [pdf] | |
H. Koh, M. Seo, J. Bang, H. Song, D. Hong, S. Park, JW. Ha, J. Choi. Online Boundary-Free Continual Learning by Scheduled Data Prior. International Conference on Learning Representations (ICLR) 2023 . [pdf] | |
S. E. Whang, Y. Roh, H. Song, JG. Lee. Data Collection and Quality Challenges in Deep Learning: A Data-Centric AI Perspective. The VLDB Journal 2023. [pdf] |
D. Park, Y. Shin, J. Bang, Y. Lee, H. Song, JG. Lee. Meta-Query-Net: Resolving Purity-Informativeness Dilemma in Open-set Active Learning. Annual Conference on Neural Information Processing Systems (NeurIPS) 2022. [pdf] [code] | |
J. Oh, S. Kim, N. Ho, JH. Kim, H. Song, SY. Yun. Understanding Cross-domain Few-shot Learning: An Experimental Study. Annual Conference on Neural Information Processing Systems (NeurIPS) 2022. [pdf] [code] | |
D. Park, J. Kang, H. Song, S. Yoon, JG Lee. Multi-view POI-level Cellular Trajectory Reconstruction for Digital Contact Tracing of Infectious Diseases. International Conference on Data Minig (ICDM) 2022. | |
S. Kim, W. Shin, S. Jang, H. Song, SY. Yun. FedRN: Exploiting k-Reliable Neighbors Towards Robust Federated Learning. International Conference on Information and Knowledge Management (CIKM) 2022. [pdf] | |
W. Shin, J. Park, T. Woo, Y. Cho, K. Oh, H. Song. e-CLIP: Large-Scale Vision-Language Representation Learning in E-commerce. International Conference on Information and Knowledge Management (CIKM) 2022. The first large-scale industry study investigating a unified multimodal transformer model. [pdf] | |
J. Oh, S. Kim, N. Ho, JH. Kim, H. Song, SY. Yun. ReFine: Re-randomization before Fine-tuning for Cross-domain Few-shot Learning. International Conference on Information and Knowledge Management (CIKM) 2022. This paper was also presented at ICML UpML workshop, 2022. [pdf] | |
H. Song, D. Sun, S. Chun, V. Jampani, D. Han, B. Heo, W. Kim, and M-H Yang. An Extendable, Efficient and Effective Transformer-based Object Detector. Preprint arXiv 2022. An extended version of ViDT to support multi-task learning. [pdf] [code] | |
S. Kim, S. Bae, H. Song, SY. Yun. LG-FAL: Federated Active Learning Strategy using Local and Global Models. International Conference on Machine Learning (ReALML Workshop) 2022. | |
S. Yun, J. Kim, D. Han, H. Song, JW. Ha, J. Shin. Time Is MattEr: Temporal Self-supervision for Video Transformers. International Conference on Machine Learning (ICML) 2022. Short Presentation. | |
JH. Kim, J. Kim, SJ. Oh, S. Yun, H. Song, J. Jeong, JW. Ha, HO. Song. Dataset Condensation via Efficient Synthetic-Data Parameterization. International Conference on Machine Learning (ICML) 2022. Short Presentation. [pdf] [code] | |
J. Bang, H. Koh, S. Park, H. Song, JW. Ha, J. Choi. Online Continual Learning on a Contaminated Data Stream with Blurry Task Boundaries. International Conference on Computer Vision and Pattern Recognition (CVPR) 2022. [pdf] [code] | |
H. Song, M. Kim, D. Park, Y. Shin, JG. Lee. Learning from Noisy Labels with Deep Neural Networks: A Survey. IEEE Transactions on Neural Networks and Learning Systems (TNNLS) 2022. The most cited survey paper on handling noisy labels with DNNs. [pdf] [code] | |
H. Song, D. Sun, S. Chun, V. Jampani, D. Han, B. Heo, W. Kim, and M-H Yang. ViDT: An Efficient and Effective Fully Transformer-based Object Detector. International Conference on Learning Representations (ICLR) 2022. [pdf] [code] | |
Y. Shin, S. Yoon, S. Kim, H. Song, JG. Lee, B. S. Lee. Coherence-based Label Propagation over Time Series for Accelerated Active Learning. International Conference on Learning Representations (ICLR) 2022. [pdf] [code] | |
M. Kim, H. Song, Y. Shin, D. Park, K. Shin, JG. Lee. Meta-Learning for Online Update of Recommender Systems. The AAAI Conference on Artificial Intelligence (AAAI) 2022. [pdf] | |
D. Kim, H. Min, Y. Nam, H. Song, S. Yoon, M. Kim, JG. Lee. COVID-EENet: Predicting Fine-Grained Impact of COVID-19 on Local Economies . The AAAI Conference on Artificial Intelligence (AAAI) 2022. Oral Presentation. [pdf] [code] |
H. Song, E. Kim, V. Jampani, D. Sun, M-H. Yang. Exploiting Scene Depth for Object Detection with Multimodal Transformers. British Machine Vision Conference (BMVC) 2021. [pdf] [code] | |
D. Park, H. Song, M. Kim, JG. Lee. Task-Agnostic Undesirable Feature Deactivation Using Out-of-Distribution Data. Annual Conference on Neural Information Processing Systems (NeurIPS) 2021. [pdf] [code] | |
H. Song, M. Kim, D. Park, Y. Shin, JG. Lee. Robust Learning by Self-Transition for Handling Noisy Labels. International Conference on Knowledge Discovery and Data Mining (KDD) 2021. Oral Presentation. [pdf] | |
JG. Lee, Y. Roh, S. E. Whang. Machine Learning Robustness, Fairness, and their Convergence. . International Conference on Knowledge Discovery and Data Mining (KDD) 2021. 2+ Hour Live Tutorial. [webpage] [pdf] [video] [slide] | |
M. Kim, H. Song, D. Kim, K. Shin, JG. Lee. PREMERE: Meta-Reweighting via Self-Ensembling for Point-of-Interest Recommendation. The AAAI Conference on Artificial Intelligence (AAAI) 2021. Oral Presentation. [pdf] [code] |
H. Song, M, Kim, S. Kim, JG. Lee. Carpe Diem, Seize the Samples Uncertain "At the Moment" for Adaptive Batch Selection. International Conference on Information and Knowledge Management (CIKM) 2020. Oral Presentation. [pdf] [code] | |
H. Song, S. Kim, M. Kim, JG. Lee. Ada-Boundary: Accelerating DNN Training via Adaptive Boundary Batch Selection. Machine Learning (ML) 2020. Invited Paper and Oral Presentation at ECML-PKDD 2020. [pdf] [code] | |
M. Kim, J. Kang, Dim, H. Song, H. Min, Y. Nam, D. Park, JG. Lee. Hi-COVIDNet: Deep Learning Approach to Predict Inbound COVID-19 Patients and Case Study in South Korea . International Conference on Knowledge Discovery and Data Mining (KDD) 2020. Oral Presentation. [pdf] [code] | |
H. Song, M. Kim, D. Park, JG. Lee. How Does Early Stopping Help Generalization against Label Noise? . International Conference on Machine Learning (UDL Workshop) 2020. [pdf] [code] | |
S. Kim, H. Song, S. Kim, B. Kim, JG. Lee. Revisit Prediction by Deep Survival Analysis. Pacific-Asia Conference on Knowledge Discovery and Data Mining (PAKDD) 2020. [pdf] | |
D. Park, H. Song, M. Kim, JG. Lee. TRAP: Two-level Regularized Autoencoder-based Embedding for Power-law Distributed Data. TheWebConf (WWW) 2020. Oral Presentation. [pdf] [code] |
D. Park, S. Yoon, H. Song, JG. Lee. MLAT: Metric Learning for kNN in Streaming Time Series. International Conference on Knowledge Discovery and Data Mining (MileTs Workshop) 2019. [pdf] | |
H. Song, M. Kim, JG. Lee. SELFIE: Refurbishing Unclean Samples for Robust Deep Learning. International Conference on Machine Learning (ICML) 2019. Short Presentation. [pdf] [code] [dataset] |
H. Song, JG. Lee. RP-DBSCAN: A Superfast Parallel DBSCAN Algorithm based on Random Partitioning. International Conference on Management of Data (SIGMOD) 2018. Top 2% of the submitted papers (accepted without revision round). Oral Presentation. [pdf] [code] |
H. Song, JG. Lee, WS. Han. PAMAE: Parallel k-Medoids Clustering with High Accuracy and Efficiency. International Conference on Knowledge Discovery and Data Mining (KDD) 2017. Selected one of the outstanding research among Microsoft Azure Supporting Projects. [blog] [pdf] [code] |
Services
Co-organized Workshop on Machine Learning Robustness, Fairness, and their Convergence at KDD 2021
Reviewer for ICML, NeurIPS, ICLR, CVPR, ICCV, ECCV, PAMI, IJCV, TNNLS since 2020
Seminar / Techtalk Data Robustness and Efficiency, Vision Transformers, and Continual Learning, GIST and IEEE Seminar, Dec 2022 [slides] ML Robustness against Label Noise, UNIST Graduate School of AI, Sep 2022 Robust Deep Learning and Extension to Real-world Applications, Database Society Summer School, Aug 2022 An Extendable, Efficient and Effective Tranformer-based Object Detector, NAVER CLOVA, Jun 2022 ML Robustness against Label Noise, Amazon AWS AI / Responsible AI, Mar 2022 Transformers for Computer Vision, Electronics and Telecommunications Research Institute, Feb 2022 Machine Learning Robustness, Fairness, and their Convergence, KDD Tutorial, Aug 2021 Robust Learning by Self-transition for Handling Noisy Labels, NAVER CLOVA, May 2021 Learning from Noisy Labels for Classification, Google Research, May 2021 Exploiting Scene Depth for Object Detection, Google Research & NAVER AI Lab, Dec 2020 Robust Learning under Label Noise, Institute of Basic Science, Dec 2019 Parallel Clustering and Large-Scale Data Analytics, NAVER CLOVA, Aug 2018
Students and Interns
I have been fortunate to work with many gifted students:
© 2022 Hwanjun Song Thanks Dr. Deqing Sun and Dr. Ce Liu for the template.