Welcome to Changwei Hu's Homepage.


About me: I am a senior applied scientist at Amazon sponsored product ads, focusing on ads recommendation and LLM. Before joining Amazon, I worked at XPENG and Yahoo Research, working on behavior prediction in autonomous driving, contents (text & image) moderation, user profiling, and time series analysis. Before that I completed my PhD in Electrical and Computer Engineering at Duke University, advised by Professor Lawrence Carin. I was also affiliated with iiD (Information Initiative at Duke). Before that, I obtained Master's degree in Physics, and Bachelor's degree in Software Engineering, both from Xiamen University in China.

Research: My research lies at the intersection of Bayesian statistics and machine learning, with an emphasis on large language models, autonomous driving and time series analysis.

Professional Activities:

  • Reviewer for IEEE PAMI
  • PC member for AAAI 2019, reviewer for AAAI 2019, 2017, 2016
  • PC member for ICMLA 2019, 2018
  • Reviewer for NeurIPS 2019, 2016
  • Reviewer for Techpulse 2018, 2017
  • PC member and reviewer for SMARTCOMP 2017
  • PC member for SmartMM 2017
  • Reviewer for IJCAI 2016
  • Reviewer for UAI 2016
  • Reviewer for IEEE Transactions on Knowledge and Data Engineering (TKDE)
  • Reviewer for Knowledge and Information Systems (KAIS)
  • Reviewer for IEEE Access

  • News:

  • 09/2019: Three papers were accepted by EMNLP 2020.
  • 09/2019: Our paper "A Deep Structural Model for Analyzing Correlated Multivariate Time Series" was accepted by ICMLA 2019.
  • 09/2019: Our paper "Large-scale Gender/Age Prediction of Tumblr Users" was accepted by ICMLA 2019.
  • 06/2017: The code for binary tensor factorization is released. See link
  • 04/2017: Our paper "Deep Generative Models for Relational Data with Side Information" was accepted by ICML 2017.
  • 01/2017: I won the 3rd prize in data visualization challenge held at Duke University.
  • 12/2015: I had two papers accepted by AISTATS 2016.
  • 09/2015: Our paper "Large-Scale Bayesian Multi-Label Learning via Topic-Based Label Embeddings" was accepted by NIPS 2015 for spotlight presentation.
  • 08/2015: Our paper "Scalable Bayesian Non-Negative Tensor Factorization for Massive Count Data" submitted to ECML PKDD 2015 received the Best Student Paper Award.
  • 08/2015: We received a research grant from Accenture for our research on machine learning for analyzing healthcare data, and I was honored as "Accenture Fellow".
  • 05/2015: Our paper "Zero-Truncated Poisson Model for Scalable Bayesian Factorization of Massive Binary Tensors with Mode-Networks" was accepted by UAI 2015 for plenary oral presentation.
  • 04/2015: Our paper "Scalable Probabilistic Tensor Factorization for Binary and Count Data" was accepted by IJCAI 2015.