Learning and prediction
We have looked at machine learning and prediction in several contexts. First, in
the control of reconfigurable interconnection networks, then in memory access pattern,
and most recently in predictive filtering for radar signal processing applications.
- "Performance of On-Line Learning Methods in Predicting
Multiprocessor Memory Access Patterns," Technical Report UMIACS-TR-59 and
CS-TR-3676 Institute for Advanced Computational Studies.
- M. F. Sakr, C.L. Giles, S. P. Levitan, B. G. Horne, M. Maggini, D. M. Chiarulli,
"Online Prediction of Multiprocessor Memory Access Patterns,"IEEE
International Conference on Neural Networks (ICNN'96), Vol. 3,
pp. 1564-1569, Washington, DC, June 3-6, 1996. Best Paper Award
- M. F. Sakr, S. P. Levitan, C. L. Giles, B. G. Horne, M. Maggini, D. M. Chiarulli,
"Predictive Control of Opto-Electronic
Reconfigurable Interconnection Networks Using Neural Networks," Second
International IEEE Workshop on Massively Parallel Processing Using Optical Interconnections
(MPPOI'95), pp. 326-335, San Antonio, TX, Oct. 23-24, 1995.
- M.F. Sakr, S.P. Levitan, D.M. Chiarulli, B.G. Horn, C.L. Giles, "Predicting
Multiprocessor Memory Access Patterns with Learning Models," Fourteenth
International Conference on Machine Learning (ICML'97), (poster) pp. 305-312,
D. Fisher, Ed., Vanderbilt University, Nashville, TN, Morgan Kaufmann, Pub., July