TY - JOUR
T1 - Predicting nearly as well as the best pruning of a planar decision graph
AU - Takimoto, Eiji
AU - Warmuth, Manfred K.
N1 - Funding Information:
∗Corresponding author. Tel.: +81-22-217-7149; fax: +81-22-263-9414. E-mail address: t2@ecei.tohoku.ac.jp (E. Takimoto). 1This work was done while the author visited University of California, Santa Cruz. 2Supported by NSF grant CCR 9821087.
PY - 2002/10/16
Y1 - 2002/10/16
N2 - We design efficient on-line algorithms that predict nearly as well as the best pruning of a planar decision graph. We assume that the graph has no cycles. As in the previous work on decision trees, we implicitly maintain one weight for each of the prunings (exponentially many). The method works for a large class of algorithms that update its weights multiplicatively. It can also be used to design algorithms that predict nearly as well as the best convex combination of prunings.
AB - We design efficient on-line algorithms that predict nearly as well as the best pruning of a planar decision graph. We assume that the graph has no cycles. As in the previous work on decision trees, we implicitly maintain one weight for each of the prunings (exponentially many). The method works for a large class of algorithms that update its weights multiplicatively. It can also be used to design algorithms that predict nearly as well as the best convex combination of prunings.
UR - http://www.scopus.com/inward/record.url?scp=0037120731&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=0037120731&partnerID=8YFLogxK
U2 - 10.1016/S0304-3975(01)00401-7
DO - 10.1016/S0304-3975(01)00401-7
M3 - Conference article
AN - SCOPUS:0037120731
VL - 288
SP - 217
EP - 235
JO - Theoretical Computer Science
JF - Theoretical Computer Science
SN - 0304-3975
IS - 2
T2 - Algorithmic Learning Theory (ALT 1999)
Y2 - 6 December 1999 through 8 December 1999
ER -