The Kindest Cut: Minimum Message Length Segmentation
Abstract
We consider some particular instances of the segmentation problem. We derive minimum message length (MML) expressions for stating the region boundaries for some one and two dimensional examples. It is found the message length cost of stating region boundaries is dependent on the noise of the data in the separated regions and also the ‘degree of separation’ of the two regions. The framework given here can be extended to different shaped cuts and also non-constant fits for the regions. Possible applications for the work presented here include its use in tree (i.e. CART) regression and in image segmentation.
Cite
Text
Baxter and Oliver. "The Kindest Cut: Minimum Message Length Segmentation." International Conference on Algorithmic Learning Theory, 1996. doi:10.1007/3-540-61863-5_36Markdown
[Baxter and Oliver. "The Kindest Cut: Minimum Message Length Segmentation." International Conference on Algorithmic Learning Theory, 1996.](https://mlanthology.org/alt/1996/baxter1996alt-kindest/) doi:10.1007/3-540-61863-5_36BibTeX
@inproceedings{baxter1996alt-kindest,
title = {{The Kindest Cut: Minimum Message Length Segmentation}},
author = {Baxter, Rohan A. and Oliver, Jonathan J.},
booktitle = {International Conference on Algorithmic Learning Theory},
year = {1996},
pages = {83-90},
doi = {10.1007/3-540-61863-5_36},
url = {https://mlanthology.org/alt/1996/baxter1996alt-kindest/}
}