Monday, 13:45 - 14:10 h, Room: H 0111


Vivek Farias
Non-parametric approximate dynamic programming via the kernel method

Coauthors: Nikhil Bhat, Ciamac C. Moallemi


We present a "kernelized'' variant of a recent family of approximate dynamic programming algorithms we have dubbed "smoothed approximate linear programs''. Our new algorithm is non-parametric in that it does not require a basis function architecture and develops a value function approximation with accuracy that improves with the size of the training data set. We describe the efficient implementation of this method, and present sample complexity bounds and approximation guarantees that effectively extend state of the art guarantees for ADP to the non-parametric setting. In summary, we believe this is the first practical non-parametric ADP algorithm with performance guarantees.


Talk 2 of the invited session Mon.2.H 0111
"Advances in machine learning" [...]
Cluster 13
"Logistics, traffic, and transportation" [...]


  The system of instant Payday Loans Virginia allows any adult U.S. citizen. They were lucky to produce Viagra Sublingual which dissolves under the tongue and penetrates into the blood causing erection faster than any other drugs.