Title: NDL: An R Package for Large-Scale Naive Discriminative Learning

Authors: Cyrus Shaoul, Petar Millin, Peter Hendrix, Michael Rascmar,
R. Harald Baayen

Affiliation: University of Tuebingen

Abstract: We introduce an R package for naive discriminative learning called
the NDL package, a modeling tool based on the Rescorla-Wagner equations
simulated using the equilibria approximation proposed by Danks (2003). NDL
allows the consequences of accumulating experience to be explored with
real-world sized datasets. This means that the model learns to use thousands
of cues to discriminate tens of thousands of outcomes, and can be trained on
corpora of up to several billion of events. We will give a short demonstration
of how the package can be used as both a psychological model of learning and
as a general purpose classifier. We will also present results from recent work
predicting human behavior using weights derived from our NDL models.