Sequential Prediction Under Log-Loss with Side Information
Abstract
The problem of online prediction with sequential side information under logarithmic loss is studied, and general upper and lower bounds on the minimax regret incurred by the predictor is established. The upper bounds on the minimax regret are obtained by constructing and analyzing a probability assignment based on mixture probability assignments in universal compression, and the lower bounds are obtained by way of a redundancy–capacity theorem. A tight characterization of the regret is provided in some special settings.
Cite
Text
Bhatt and Kim. "Sequential Prediction Under Log-Loss with Side Information." Proceedings of the 32nd International Conference on Algorithmic Learning Theory, 2021.Markdown
[Bhatt and Kim. "Sequential Prediction Under Log-Loss with Side Information." Proceedings of the 32nd International Conference on Algorithmic Learning Theory, 2021.](https://mlanthology.org/alt/2021/bhatt2021alt-sequential/)BibTeX
@inproceedings{bhatt2021alt-sequential,
title = {{Sequential Prediction Under Log-Loss with Side Information}},
author = {Bhatt, Alankrita and Kim, Young-Han},
booktitle = {Proceedings of the 32nd International Conference on Algorithmic Learning Theory},
year = {2021},
pages = {340-344},
volume = {132},
url = {https://mlanthology.org/alt/2021/bhatt2021alt-sequential/}
}