👋 关于我
我是 东北大学（中国） 与 邓迪大学（英国） 的中英联合培养本科生，当前正在苏格兰邓迪进行本科第四年的学习。
我的研究兴趣包括：
- 🤖 人工智能（AI）
- 🧬 智能影像组学（Intelligent Radiomics）
- 🧠 神经科学（Neuroscience）
- 🏃♂️ 活动识别（Activity Recognition）
- 💓 医学生理信号分析（Biomedical Signal Processing）
🎓 学历
📝 学术成果
已发表论文
Wang, X.¹, Liu, H.¹, Zhang, Y., Zhao, B., Duan, H., Hu, W., ... & Li, C. (2025). Joint Modelling Histology and Molecular Markers for Cancer Classification. Medical Image Analysis, 102(4): 103505.
论文简介
This paper introduces a novel approach for cancer classification by integrating histological images with molecular marker data. The proposed model utilizes a multi-modal deep learning architecture to effectively combine features from different data sources, resulting in improved diagnostic accuracy compared to single-modality methods. Our framework demonstrates significant performance improvements across multiple cancer types and provides interpretable results that can aid clinicians in treatment planning.
Liu, H.¹, Wang, Z.*,¹, Zhao, B., Shen, Q., Li, M., Que, N., ... & Xin, J.* (2024, August). Multi-scale context-aware networks based on fragment association for human activity recognition. In Proceedings of the Thirty-Third International Joint Conference on Artificial Intelligence (pp. 3169-3177).
论文简介
This work presents a novel approach to human activity recognition through multi-scale context-aware networks. We introduce a fragment association mechanism that enables the model to capture temporal relationships across different time scales. Our method achieves state-of-the-art performance on multiple benchmark datasets by effectively modeling both local and global temporal patterns in sensor data, leading to more robust recognition of complex human activities.
Liu, H.¹, Zhao, B.¹, Dai, C., Sun, B., Li, A., & Wang, Z* (2023). MAG-Res2Net: A novel deep learning network for human activity recognition. Physiological Measurement, 44(11), 115007.
论文简介
In this paper, we propose MAG-Res2Net, an innovative deep learning architecture specifically designed for human activity recognition using wearable sensor data. Our network combines multi-attention gating mechanisms with a hierarchical residual structure to capture both fine-grained motion details and high-level activity patterns. Extensive experiments demonstrate that MAG-Res2Net outperforms existing methods, particularly for complex activities and in challenging real-world scenarios with noise and variability.
在投论文
-
Tang H^1^,
Liu, H.*^1^, Li C. (2025). FOD-Diff: 3D Multi-Channel Patch Diffusion Model for Fiber Orientation Distribution. MICCAI2025 -
Liu, H.*, Li, X.^1^, Jiang, Y.^1^, Tang, H., Wu, D., & Guo, Y. (2025). Redundant feature screening method for human activity recognition based on attention purification mechanism. arXiv preprint arXiv:2503.23537. [网页] [预览] [[下载]] -
Liu, H.*, Li, S.^1^, Yu, Y.^1^, Jiang, Y., Xiao, H., Long, J., & Tang, H. (2025). CMD-HAR: Cross-Modal Disentanglement for Wearable Human Activity Recognition. arXiv preprint arXiv:2503.21843. [网页] [预览] [[下载]] -
Yu Y.^1^, Wang H., Wang J., Yan M., Han X., Wu D., Zhao X.,
Liu, H.*.(2024). Optimizing End-to-End Sensor-Based Human Activity Recognition Through Multi-Attention Interaction. [网页] [预览] [[下载]]
学术服务
- AAAI2024 会议审稿
- IJCAI2024 会议审稿
- MICCAI2025 会议审稿
- 为MICCAI-CMMCA2025大会提供技术支持 [网页]
- IEEE Transactions on Industrial Informatics 期刊审稿人
- Engineering Applications of Artificial Intelligence 期刊审稿人
专利
刘涵瑜, 赵伯阳, 王之琼. 基于多尺度度量学习的人类活动识别综合优化方法. （已受理）
软件著作权
- 赵伯阳,
刘涵瑜, 王之琼. 基于多尺度神经网络的动作识别系统V1.0.（已受理）
💻 科研项目
2024.09 - Now
剑桥大学 | 临床医学院 - 临床神经科学系（英国剑桥）
2024.04 - 2025.01
剑桥大学 | 数学院 - 应用数学与理论物理系（英国剑桥）
2023.12 - 2025.03
东北大学 | 信息科学与工程学院（辽宁沈阳）
2023.07 - 2024.04
东北大学 | 医学与生物信息工程学院 & 计算机科学与工程学院（辽宁沈阳）
2022.10 - 2023.04
东北大学 | 医工学院（辽宁沈阳）
🏅 荣誉奖项
- 2024.8 获得 中国大学生计算机博弈大赛——第十八届中国计算机博弈锦标赛
一等奖 - 2024.4 获得 中国国家大学生创新设计训练计划项目
专业第一名, 国家级优秀结题 - 2023.8 获得 中国大学生生物医学工程创新设计大赛
第一名, 二等奖 - 2023.8 获得 中国大学生计算机博弈大赛——第十七届中国计算机博弈锦标赛
二等奖
🏅 学术会议
- 2024.8, International Joint Conferences on Artificial Intelligence, 韩国济州岛, 受邀报告