Dept. of Computer Science
Columbia University
Mailcode 0401
1214 Amsterdam Ave.
New York, NY 10027-7003
Office: CEPSR 612

Stuart Andrews
Postdoctoral Scientist in Machine Learning @ Columbia

I work in the Columbia Machine Learning Laboratory directed by Tony Jebara.

My current research concerns a problem called structured network completion which has applications in social networks as well as computational molecular and systems biology. In the example below, the goal is to infer the complete graph on the right using geometric and topological cues from the incomplete graph on the left.

  • Microarray Preprocessing
    (pdf) with Uri-David Akavia and Bo-Juen Chen
  • Structured Prediction of Generalized Matchings
    (pdf) with Tony Jebara, Submitted to JMLR
  • Generalized Matching Code
  • Graph reconstruction with degree-constrained subgraphs
    (abstract, poster) with Tony Jebara, Workshop on Statistical Models of Networks (spotlight + poster), NIPS 2007
  • A Transductive Max-Margin Framework for Completion of Structured Variables with Application to Semi-Supervised Graph Inference
    (pdf) with Tony Jebara, New York Academy of Sciences, Machine Learning Symposium (poster), October 2007
  • Structured Network Learning
    (pdf) with Tony Jebara, Workshop on Learning to Compare Examples (talk) NIPS 2006
  • Learning from Ambiguous Examples
    (pdf) Doctoral Thesis, Brown University, 2007
  • A Cutting-Plane Algorithm for Learning from Ambiguous Examples
    (abstract, paper) with Thomas Hofmann, Technical Report, Brown University, 2006
  • MIL Data Multiple-instance learning data sets from our 2003 paper.
  • MIL Bibliography An incomplete list of papers on multiple-instance learning.