Reliably Learning the ReLU
We give the first dimension-efficient algorithms for learning Rectified Linear Units (ReLUs), which are functions of the form x 7→ max(0, w · x) with w ∈ S<sup>n−1</sup> . Our algorithm works in the challenging Reliable Agnostic learning model of Kalai et al. (2012) where the learner is...
| Main Authors: | , , , |
|---|---|
| Format: | Conference item |
| Published: |
Journal of Machine Learning Research
2017
|