2
amoux
3y

One-hot encoding is fucking garbage. Everyone loves using this useless stupid shit that doesn’t work.

Comments
  • 2
    How do you know that "One-hot encoding" doesn't work? Is your accuracy dropping?

    We need more details.
  • 4
    Garbage in, garbage out.

    The architecture is also important.

    It works fine for most simple classification problems
  • 2
    It's not garbage, but like anything in ML/DL/NLP, if you throw garbage at a model you get garbage out (even if you have clean data with the wrong model).
Add Comment