machine learning is an inelegant hack to solve problems without first understanding them. rule-based expert systems are better in basically every way. unfortunately, if there's one thing the tech industry loves more than inelegant hacks, it's not understanding things, so here we are.
Doing embedded stuff
- he/him
- jamesmunns.com/
Be nice, work hard. Build tools, tell stories. Start five year fights. Kein Ort für Nazis.
A list of all the other services I'm on around the internet
Posts with code highlighted with codehost by @wavebeem.
All posts licensed under CC-BY-SA 4.0, unless otherwise stated.
You must log in to comment.
in reply to @cactus's post:
i've been doing data plumbing for a while and now that i'm getting more into 'solving problems with ML' part of it i've honestly been sort of shocked how little understanding goes into our process. we're trying to make our spellchecker better and our existing solution doesn't have a concept of a word, let alone a misspelled one.
i keep saying "maybe we could even just use a normal dictionary and some basic heuristics and see how that goes" and the response is more or less "well that isn't very ML, is it"
One can use ML approaches to develop rules-based systems. I’ve worked projects where that approach was successful.
Pinned Tags