• August 13th 2016
    NIPS & Allerton

    Two machine-learning papers on which I collaborated with other researchers have been accepted by NIPS 2016 and Allerton 2016, respectively. I was fortunate to be able to help on these papers a little bit from a systems perspective.

    NIPS 2016: Xinghao Pan (Berkeley) et al.: CYCLADES: Conflict-free Asynchronous Machine Learning

    Allerton 2016: Ioannis and Stefan (Stanford) have had a paper accepted by Allerton 2016. In this paper, Ioannis articulates elegantly the relationship between momentum and asynchrony for distributed stochastic gradient descent. The theoretical result is inspired by Stefan's distributed deep-learning system, called Omnivore, which is described here.

  • August 3rd 2016
    Device Donation from NVIDIA

    Thank you to NVIDIA for the generous donation of one Jetson TX1 and one Titan X to support my group. These devices will enable the study of the following question: How fast can we make a subset of machine-learning algorithms, using just your everyday laptops with energy-efficient, credit-card-sized coprocessors?

    Big Data in Small Pockets?--see this very preliminary one-pager for our vision! At the end, we hope to support GPUs, FPGAs, or even specially designed hardware. But first, we plan to understand the system trade-offs better with GPUs.

    Students--if you are excited by this project, send me an email and let's chat!

  • June-Sep 2016
    Visting Peking & Tsinghua

    I will spend my summer visiting Peking University (Institute of Network) and Tsinghua University (School of Software). It is exciting to work with professors and students back in my hometown (one of which is also my alma mater)!

  • June 2016
    ETH Zurich

    I am joining ETH Zurich as a tenure-track assistant professor in computer science.

  • 2010-2016
    Stanford University & UW-Madison

    I spent six years working with Christopher Ré at Stanford University and the University of Wisconsin-Madison. For a summary of our work, see my research statement and PhD dissertation.

    With indispensable help from many collaborators, my PhD thesis produced DeepDive, a trained system for automatic knowledge base construction. DeepDive has been used to achieve human-level quality in a range of applications, such as paleontology, anti-human trafficking, and genomics etc. DeepDive is being commercialized as Lattice.

    I would like to thank my advisor, Chris. I owe Chris my career as a researcher. Since the day I first met Chris and told him about my dream, he has done everything he could, as a scientist, an educator, and a friend to help me.

  • 2006-2010
    Peking University

    I spent three years working with Bin Cui at Peking University as an undergraduate student. My research focused on managing user-generated content on social media. For a summary our work, see my B.Sc. dissertation (in Chinese) or the SIGIR 2010 and SIGMOD 2010 paper.

    I would like to thank my advisor, Bin. The training I received from his group helped me realize that doing research is exciting and fun and that data management is awesome.

  • ce.zhang@inf.ethz.ch
  • CAB F 71.2 @ ETH
  • TBA
Travel Log
  • VLDB, 09/05-09/09
  • NVIDIA, 07/22
  • IBM Watson, 06/30-07/01
  • SIGMOD, 06/27-06/28
  • NVIDIA: Device Gifts