TY  - RPRT
TI  - Algorithm Selection as a Bandit Problem with Unbounded Losses
AU  - Matteo Gagliolo
AU  - Juergen Schmidhuber
PY  - 2008
DO  - 10.1007/978-3-642-13800-3_7
UR  - https://arxiv.org/abs/0807.1494
ID  - 0807.1494
ER  -