Charles Pippin
Georgia Institute of Technology
College of Computing
cepippin@cc
Charles Pippin is a first year Phd student in the College of Computing at Georgia Tech.
He is currently employed by the
Georgia Tech Research Institute as a Research Scientist, and is part
of the development team for a collaborative mapping and planning system. Charles
worked as a software consultant before returning to graduate school.
Resume: charlespippin.pdf
His research interests are in machine learning and robotics, see below for more information.
Signature Verification using Stroke Features
Signature verification is a biometric technique that is useful because
signatures are in many practices accepted as a means of verification.
This work approaches signature verification using filters of varying
sensitivity. In the first, global features of the signature, such as
average velocity are considered using a weak learner. In the second,
strokes are segmented using the minima of the velocity and encoded before
comparing them using dynamic time warping.
This research was performed as part of my College of Computing masters
project with Dr. Charles Isbell at Georgia Tech.
Handwritten SignatureVerification (pdf)
More On Signature Segmentation (pdf)
Financial Prediction - Hierarchical Confidence based clustering
This work seeks to improve on the use of neural networks by using them
in a hierarchical fashion. As networks become specialized over subsets
of the data, the datasets are split and new networks trained on the more
difficult set. Our hope is that through this approach, each network becomes
specialized to a type of data in the possible space. We have applied
this problem to the area of financial prediction and found slight improvements
over the prediction accuracy of a single network.
This research was performed as part of my machine learning course with
Dr. Charles Isbell at Georgia Tech.
Hierarchical Confidence Based Clustering
Biotracking of Snails
This work seeks to apply the video tracking tools from the Biotracking
team and Dr. Tucker Balch at Georgia Tech, to the problem of observing
the behavior of multiple snails in a tank environment. Snails are videotaped
in their environment, and the Biotracking color segmentation tools from
the GA Tech Biotracking
Lab are applied to the video to derive spatial coordinate information
over time.
This research was performed as part of an independent study in concert
with Dr. Robert Butera in the school of Biomedical Engineering at Georgia
Tech.
A Philosophical View on Free Will
Can Free Will exist if the actions of an agent are known in advance? This paper seeks to address that issue in the context of philosophy of cognition.
Free Will In a Deterministic Machine
Genetic Algorithm Based Multiprocessor Scheduling Algorithm
This work seeks to apply the use of a Genetic algorithm to the problem
of multiprocessor scheduling in the Linux kernel. Each iteration of the
genetic algorithm dynamically replaces the kernel at runtime and runs
an application performance test. At the end of a number of iterations,
our hope is that an optimal scheduling policy can be found.
This research was performed as part of my advanced operating systems
course with Dr. Karsten Schwan at Georgia Tech.