talk.bib
@misc{Ning2007,
author = {\textbf{X. Ning} and Karypis, George},
title = {Evaluation of 3D Descriptors in Virtual Screening},
howpublished = {Oral presentation, ACS annual meeting},
month = {Sept.},
year = {2007},
abstract = {In recent years there has been an increased interest in using structural
descriptors in conjunction with advanced supervised learning algorithms
(e.g., support vector machines and neural networks) for solving various
problems arising in virtual screening. This research resulted in
the development of highly effective activity and/or property prediction
methods and has provided an objective and data-driven assessment
of the characteristics that a descriptor set should have in order
to achieve good performance. Unfortunately, this research has primarily
focused on topological descriptors and to a large extent has ignored
the various 3D descriptors. In this talk we discuss our results in
evaluating the various parameters of the design space for 3D descriptors
and how they impact the machine learning based virtual screening
approaches. Specifically, our work focuses on the questions like:
What kinds of 3D elements of the compound structures are the most
significant for bioactivity and how to efficiently extract them?
How to quantitatively measure and represent these significant 3D
elements in descriptors so as to optimally balance the trade-off
between generality and specificity of structure representation? What
is the best way of using the 3D descriptors in kernel-based machine
learning approaches in order to take great advantage of boththe descriptors
and the learning method? We address these questions by performing
a comprehensive experimental evaluation using different 3D descriptors
on a wide-range of datasets.},
file = {papers/Ning2007.pdf},
owner = {xia},
timestamp = {2011.03.27}
}