2 Apr
2022
2 Apr
'22
10:15 p.m.
[i guess it's nice to just think of this. i'm imagining features emerging from the data, based on their utility for predicting it. one approach is to actually train a prediction model.] [another approach is what? remember this data has precisely-same audio components, scattered everywhere in patterns to be mapped.] [features emerge from data, based on utility for predicting it ... without sequentialness?] [there are a number ways of approaching this. what's important is that it isn't being publicly discussed, and would be incredibly useful.] [features in the data ...]