4 ms·
I mean in the sense that using ML for a problem often requires just trying a dozen different modeling techniques, then a bunch of a hyper-parameter searching, t
by gok 6y ago
I mean in the sense that using ML for a problem often requires just trying a dozen different modeling techniques, then a bunch of a hyper-parameter searching, then a bunch of stochastic tuning…
- SoSoRoCoCo 6y agoOh. I see what you mean. Yeah, I guess by definition backwards propagation is trial-end-error. Huh, I never thought of it that way. Thanks for clarifying, I thought you were being saucy: my apologies for being snarky.
- gok 6y agoIt's HN; we're all here for the snark!