Completed
Best performing networks: comparison between GPs and SGD-NNS
Class Central Classrooms beta
YouTube videos curated by Class Central.
Classroom Contents
Computation in Very Wide Neural Networks
Automatically move to the next video in the Classroom when playback concludes
- 1 Intro
- 2 Outline
- 3 Starting point
- 4 Single-hidden layer (shallow) neural networks of infinite width Consider a NN which
- 5 What GP does it correspond to?
- 6 Properties of the NNGP
- 7 Bayesian inference with a GP prior (Review)
- 8 Experiments from original work
- 9 Performance comparison
- 10 NNGP performance across hyperparameters
- 11 Large depth behavior & fixed points
- 12 Phase diagrams: experiments vs. theory
- 13 Performance trends with width and dataset size
- 14 Empirical comparison of various NN-GPS
- 15 Empirical trends
- 16 Best performing networks: comparison between GPs and SGD-NNS
- 17 Partway summary
- 18 What dynamics occurs in parameter space?
- 19 Closing Remarks
