Summary: Specification and Simulation of Statistical Query Algorithms
for E#ciency and Noise Tolerance
Javed A. Aslam # Scott E. Decatur +
Aiken Computation Laboratory
Harvard University
Cambridge, MA 02138
Abstract
A recent innovation in computational learning theory is
the statistical query (SQ) model. The advantage of specify­
ing learning algorithms in this model is that SQ algorithms
can be simulated in the PAC model, both in the absence and
in the presence of noise. However, simulations of SQ algo­
rithms in the PAC model have non­optimal time and sample
complexities. In this paper, we introduce a new method for
specifying statistical query algorithms based on a type of
relative error and provide simulations in the noise­free and
noise­tolerant PAC models which yield e#cient algorithms.
Requests for estimates of statistics in this new model take
the form: ``Return an estimate of the statistic within a 1±µ
factor, or return `#', promising that the statistic is less than

Source: Aslam, Javed - College of Computer Science, Northeastern University

Collections: Computer Technologies and Information Sciences