About Me

Welcome to my homepage! My name is Bojian Hou (also Bo-Jian Hou), a postdoctoral researcher of CBICA (The Center for Biomedical Image Computing and Analytics) at the University of Pennsylvania advised by Prof. Yong Fan. Before that, I was a postdoctoral associate in the Department of Poplulation Health and Science at Cornell University, advised by Prof. Fei Wang.

I received my B.Sc. and Ph.D. degree in the Department of Computer Science at Nanjing University in 2014 and 2020 separately. I was a member of LAMDA Group led by Prof. Zhi-Hua Zhou during my doctoral study. My Ph.D. supervisor is Prof. Zhi-Hua Zhou.

I have broad interests in machine learning and data mining, and their potential applications to biomedical data such as medical images, medical literature, and electronic health records (EHR).

During my doctoral studies, I developed a novel learning scenario known as feature evolvable learning, where data features would evolve in an open and dynamic environment. The goal was to keep optimal online learning performance in dynamic feature space. I also studied semi-supervised learning and interpretability problems, such as storage-fit learning with unlabeled data and learning the interpretable structure from RNNs, respectively.

At my postdoctoral position, I mainly conducted multimodal survival analysis for medical images and clinical data, built natural language processing models to do medical literature mining, and investigated the potential issues of the interpretability methodologies for medical data.

In summary, my research interests include:

  • Interpretability: studying the interpretability of the black-box machine learning models.
  • Feature Evolvable Learning: studying learning scenarios where data features evolve.
  • Semi-Supervised Learning: learning models from both labeled and unlabeled data.
  • Online Learning: learning models continuously from online streaming data.
  • Natural Language Processing: utilizing pre-trained model to understand natural language.
  • Deep Learning: using deep neural networks to handle complex spatial and temporal data.
  • Learnware: towards reusable, evolvable and comprehensible machine learning models.

Recent Highlights

  • 06-29-2022: Our paper “Online Deep Learning from Doubly-Streaming Data” with Heng Lian, John Scovil Atwood, Jian Wu and Yi He was accepted by ACMMM’22.
  • 11-22-2021: Winning the Excellent Doctoral Dissertation Award of Jiangsu Province.
  • 09-13-2021: Winning the Excellent Doctoral Dissertation Award of Nanjing University.
  • 08-31-2021: Our paper “Online Learning in Variable Feature Spaces with Mixed Data” with Yi He, Jiaxian Dong, Yu Wang, and Fei Wang was accepted by ICDM’21.
  • 04-01-2021: Our paper “Prediction with Unpredictable Feature Evolution” with Prof. Lijun Zhang and Prof. Zhi-Hua Zhou was accepted by IEEE Transactions on Neural Networks and Learning Systems.
  • 12-24-2020: Winning the JSAI Excellent Doctoral Dissertation Award.
  • 12-02-2020: Our paper “Storage Fit Learning with Feature Evolvable Streams” with Yu-Hu Yan, Peng Zhao and Zhi-Hua Zhou was accepted by AAAI’21.
  • 06-18-2020: Winning the CS Excellent Doctoral Dissertation Award of Nanjing University.
  • 05-27-2020: I have successfully defended my PhD dissertation and became a Ph.D.
  • 04-21-2020: Winning the Outstanding Graduate Student Award of Nanjing University.