Classifiers are often used to detect miscre-ant activities. We study how an adversary can efficiently query a classifier to elicit information that allows the adversary to evade detection at near-minimal cost. We generalize results of Lowd and Meek (2005) to convex-inducing classifiers. We present algorithms that construct undetected instances of… (More)
BACKGROUND Recombinant protein production in the methylotrophic yeast Pichia pastoris largely relies on integrative vectors. Although the stability of integrated expression cassettes is well appreciated for most applications, the availability of reliable episomal vectors for this host would represent a useful tool to expedite cloning and high-throughput… (More)
Awarded for Lantern. The first time ever the award went to a digital project instead of a book. th Century Media, http://projectarclight.org (2013-present) • Co-authored successful $200,000 Digging into Data grant proposal. • Developed a new approach for data mining more than 2 million pages of newspapers and film and television magazines. • Leader of the… (More)
Search has been unfairly maligned within digital humanities big data research. While many digital tools lack a wide audience due to the uncertainty of researchers regarding their operation and/or skepticism towards their utility, search offers functions already familiar and potentially transparent to a range of users. To adapt search to the scale of Big… (More)
The Macademia website promotes faculty collaboration and research by visualizing faculty research interests as a dynamic "constellation."Semantic similarity inference algorithms power the site's visualization, allowing users to spatially browse related research interests, and researchers who have those interests.