Publications
This page lists peer-reviewed publications. For surveys and works currently under review, please refer to the homepage.
Published in CIKM 2025, 2025
Shangqing Xu, Harshavardhan Kamarthi, Haoxin Liu, and B. Aditya Prakash analyze how in-context learning enables time-series foundation models to adapt to novel forecasting tasks without gradient updates.
Recommended citation: Shangqing Xu, Harshavardhan Kamarthi, Haoxin Liu, and B. Aditya Prakash. "In-context Pre-trained Time-Series Foundation Models Adapt to Unseen Tasks." In Proceedings of CIKM 2025.
Published in KDD 2025, 2025
Zhiyuan Zhao, Haoxin Liu, Alexander RodrÃguez, and B. Aditya Prakash introduce a framework for performative time-series forecasting that adapts to feedback loops between predictions and the environment.
Recommended citation: Zhiyuan Zhao, Haoxin Liu, Alexander RodrÃguez, and B. Aditya Prakash. "Performative Time-Series Forecasting." In Proceedings of the 2025 ACM SIGKDD Conference on Knowledge Discovery and Data Mining (KDD 2025).
Published in NAACL 2025 (Main Conference), 2025
Haoxin Liu, Chenghao Liu, and B. Aditya Prakash propose a visualization-first pipeline that helps large language models reason about time-series data in natural language tasks.
Recommended citation: Haoxin Liu, Chenghao Liu, and B. Aditya Prakash. "A Picture is Worth A Thousand Numbers: Enabling LLMs Reason about Time Series via Visualization." NAACL 2025 Main Conference.
Published in NeurIPS 2024, 2024
Haoxin Liu and collaborators release Time-MMD, a benchmark multimodal dataset for time-series analysis that spans multiple domains and tasks (acceptance rate 25.8%).
Recommended citation: Haoxin Liu, Shangqing Xu, Zhiyuan Zhao, Lingkai Kong, Harshavardhan Kamarthi, Aditya B. Sasanur, Megha Sharma, Jiaming Cui, Qingsong Wen, Chao Zhang, and B. Aditya Prakash. "Time-MMD: A New Multi-Domain Multimodal Dataset for Time Series Analysis." Advances in Neural Information Processing Systems 2024.
Published in ACL 2024 Findings, 2024
Haoxin Liu, Zhiyuan Zhao, Jindong Wang, Harshavardhan Kamarthi, and B. Aditya Prakash present LSTPrompt, enabling large language models to perform zero-shot time-series forecasting via structured prompting.
Recommended citation: Haoxin Liu, Zhiyuan Zhao, Jindong Wang, Harshavardhan Kamarthi, and B. Aditya Prakash. "LSTPrompt: Large Language Models as Zero-Shot Time Series Forecasters by Long-Short-Term Prompting." Findings of ACL 2024.
Published in ICML 2024, 2024
Haoxin Liu and collaborators design invariant learning principles that improve the out-of-distribution robustness of time-series forecasting models (acceptance rate 27.5%).
Recommended citation: Haoxin Liu, Harshavardhan Kamarthi, Lingkai Kong, Zhiyuan Zhao, Chao Zhang, and B. Aditya Prakash. "Time-Series Forecasting for Out-of-Distribution Generalization Using Invariant Learning." International Conference on Machine Learning 2024.
Published in WWW 2023 Industry Track, 2023
Haoxin Liu and co-authors develop HAPENS, a hardness-personalized negative sampling strategy that strengthens large-scale recommender systems (acceptance rate 19.8%).
Recommended citation: Haoxin Liu, Pu Zhao, Si Qin, Yong Shi, Mirror Xu, Qingwei Lin, and Dongmei Zhang. "HAPENS: Hardness-Personalized Negative Sampling for Implicit Collaborative Filtering." The Web Conference 2023 Industry Track.
Published in CIKM 2022 DL4SR, 2022
Qian Yu, Xiangdong Wu, Chen Yang, Zihao Zhao, Haoxin Liu, and Jingping Shao explore global behavioral context for sequential recommendation, presented at the CIKM 2022 DL4SR workshop.
Recommended citation: Qian Yu, Xiangdong Wu, Chen Yang, Zihao Zhao, Haoxin Liu, and Jingping Shao. "Exploiting Global Behavior Contextual Correlation in Sequential Recommendation." CIKM 2022 DL4SR.
Published in SIGIR 2022, 2022
Haoxin Liu (sole author) presents LightSGCN, a simplified signed graph convolutional network for link sign prediction accepted to SIGIR 2022 (acceptance rate 24.7%).
Recommended citation: Haoxin Liu. "LightSGCN: Powering Signed Graph Convolution Network for Link Sign Prediction with Simplified Architecture Design." SIGIR 2022.
Published in CVPR 2022, 2022
Xingxuan Zhang, Linjun Zhou, Renzhe Xu, Peng Cui, Zheyan Shen, and Haoxin Liu propose methods for unsupervised domain generalization, published at CVPR 2022 (acceptance rate 25.3%).
Recommended citation: Xingxuan Zhang, Linjun Zhou, Renzhe Xu, Peng Cui, Zheyan Shen, and Haoxin Liu. "Towards Unsupervised Domain Generalization." CVPR 2022.
Published in KDD 2021, 2021
Haoxin Liu, Ziwei Zhang, Peng Cui, Yafeng Zhang, Qiang Cui, Jiashuo Liu, and Wenwu Zhu propose a signed graph neural network that leverages latent groups, appearing in the KDD 2021 Research Track (acceptance rate 15.4%).
Recommended citation: Haoxin Liu, Ziwei Zhang, Peng Cui, Yafeng Zhang, Qiang Cui, Jiashuo Liu, and Wenwu Zhu. "Signed Graph Neural Network with Latent Groups." KDD 2021 Research Track.
Published in ICCSSE 2019, 2019
Bin Jiang, Haoxin Liu, Qingwei Li, Shuhua Cao, Zhaoyu Chen, Liyun Cheng, and Meixia Qu design improved CNN-based feature extraction for SDSS spectra, earning Best Oral Presentation at ICCSSE 2019.
Recommended citation: Bin Jiang, Haoxin Liu, Qingwei Li, Shuhua Cao, Zhaoyu Chen, Liyun Cheng, and Meixia Qu. "Feature Extraction of SDSS Spectra With Improved CNN." ICCSSE 2019.
