NHacker Next
- new
- past
- show
- ask
- show
- jobs
- submit
login
The paper maps two energy-based methods to each other. I'm not sure what the innovation is.
The novelty is in identifying spurious samples at the memorization-generalization threshold that act like memorized samples even though they don't appear in the training data.
Rendered at 05:30:37 GMT+0000 (Coordinated Universal Time) with Vercel.