4 ms·
Just like with spam filters, the problem isn't when they work; it's the false positives that are really the concern, and what I've seen with all the machine-lea
by userbinator 8y ago
Just like with spam filters, the problem isn't when they work; it's the false positives that are really the concern, and what I've seen with all the machine-learning stuff these days is that they tend to fail in mysterious ways.