Piyush Rai

Assistant Professor
Dr. Deep Singh and Daljeet Kaur Faculty Fellow
Computer Science & Engineering, IIT Kanpur

Adjunct Assistant Professor
Electrical and Computer Engineering, Duke University


Email: firstname AT cse DOT iitk DOT ac DOT in
Office: KD-319, CSE Department, IIT Kanpur


I'm an Assistant Professor in Computer Science & Engineering at Indian Institute of Technology, Kanpur. I'm also an Adjunct Faculty in the Electrical & Computer Engineering department at Duke University. I did my Ph.D. (2007-2012) in Computer Science from School of Computing, University of Utah, and B.Tech. in Computer Science and Engineering from IIT-BHU, Varanasi.

I work in the area of machine learning and Bayesian statistics. My research is primarily on probabilistic modeling of massive and complex data. My research focuses on inferring compact latent structures and feature representations from such data and leveraging these to better understand/summarize the data and make better predictions/decisions.


Recent Teaching:

- Bayesian Machine Learning (2016-17, Semester II)
- Machine Learning (2016-17, Semester I)
- Probabilistic Machine Learning (2015-16, Semester II)

Recent Publications (full list)

  1. Non-negative Inductive Matrix Completion for Discrete Dyadic Data [pdf]
    AAAI 2017, San Francisco, USA

  2. Earliness-Aware Deep Convolutional Networks for Early Time Series Classification [pdf]
    With: Wenlin Wang, Changyou Chen, Wenqi Wang, and Lawrence Carin
    (arXiv pre-print, 2016)

  3. Deep Distance Metric Learning with Data Summarization [pdf]
    With: Wenlin Wang, Changyou Chen, Wenlin Chen, and Lawrence Carin
    ECML 2016, Riva del Garda, Italy

  4. Topic-Based Embeddings for Learning from Large Knowledge Graphs [pdf]
    With: Changwei Hu and Lawrence Carin
    AISTATS 2016, Cadiz, Spain

  5. Non-negative Matrix Factorization for Discrete Data with Hierarchical Side-Information [pdf]
    With: Changwei Hu and Lawrence Carin
    AISTATS 2016, Cadiz, Spain

  6. Large-Scale Bayesian Multi-Label Learning via Topic-Based Label Embeddings [pdf]
    With: Changwei Hu, Ricardo Henao, and Lawrence Carin
    NIPS 2015, Montreal, Canada
    Spotlight Presentation

  7. Zero-Truncated Poisson Tensor Factorization for Massive Binary Tensors [pdf]
    With: Changwei Hu and Lawrence Carin
    UAI 2015, Amsterdam, The Netherlands
    Plenary Oral Presentation

  8. Scalable Bayesian Non-Negative Tensor Factorization for Massive Count Data [pdf][slides]
    With: Changwei Hu, Changyou Chen, Matthew Harding, and Lawrence Carin
    ECML 2015, Porto, Portugal
    Best Student Paper Award