5 ms·
Interesting, I'd love to see a link to that if you know of it. Here's the original paper: https://aclanthology.org/2023.findings-acl.426.pdf https://aclantholog
by speleo 3y ago
Interesting, I'd love to see a link to that if you know of it. Here's the original paper: https://aclanthology.org/2023.findings-acl.426.pdf https://aclanthology.org/2023.findings-acl.426.pdf In my own work I've successfully classified emergent behavior in Cellular Automata using a similar technique, and the technique has also been used elsewhere with success: https://www.nature.com/articles/s41598-022-12826-w https://www.nature.com/articles/s41598-022-12826-w
- isoprophlex 3y agoThis took me an unreasonable amount of time to find, but here it is https://kenschutte.com/gzip-knn-paper2/ https://kenschutte.com/gzip-knn-paper2/ The moral: the methodology is cool, but implementation details matter, i guess...
- speleo 3y agoThank you for this, I appreciate it! That's unfortunate to hear. I may have to swap out the example I used in this article, and maybe also include a note that this technique has limitations. I think that using compression/Kolmogorov complexity metrics for classification is a fruitful endeavor and that the philosophy of groups like the Hutter Prize are sound, but the kNN + gzip example looks like it has some problems with it. For anyone else following along, I think the GitHub Issue discussion on the paper's repo is really interesting: https://github.com/bazingagin/npc_gzip/issues/3 https://github.com/bazingagin/npc_gzip/issues/3