A Mean Field Theory of Batch Normalization

February 21, 2019 ยท Declared Dead ยท ๐Ÿ› International Conference on Learning Representations

๐Ÿ‘ป CAUSE OF DEATH: Ghosted
No code link whatsoever

"No code URL or promise found in abstract"

Evidence collected by the PWNC Scanner

Authors Greg Yang, Jeffrey Pennington, Vinay Rao, Jascha Sohl-Dickstein, Samuel S. Schoenholz arXiv ID 1902.08129 Category cs.NE: Neural & Evolutionary Cross-listed cond-mat.dis-nn, cs.LG, math.DS Citations 186 Venue International Conference on Learning Representations Last Checked 2 months ago
Abstract
We develop a mean field theory for batch normalization in fully-connected feedforward neural networks. In so doing, we provide a precise characterization of signal propagation and gradient backpropagation in wide batch-normalized networks at initialization. Our theory shows that gradient signals grow exponentially in depth and that these exploding gradients cannot be eliminated by tuning the initial weight variances or by adjusting the nonlinear activation function. Indeed, batch normalization itself is the cause of gradient explosion. As a result, vanilla batch-normalized networks without skip connections are not trainable at large depths for common initialization schemes, a prediction that we verify with a variety of empirical simulations. While gradient explosion cannot be eliminated, it can be reduced by tuning the network close to the linear regime, which improves the trainability of deep batch-normalized networks without residual connections. Finally, we investigate the learning dynamics of batch-normalized networks and observe that after a single step of optimization the networks achieve a relatively stable equilibrium in which gradients have dramatically smaller dynamic range. Our theory leverages Laplace, Fourier, and Gegenbauer transforms and we derive new identities that may be of independent interest.
Community shame:
Not yet rated
Community Contributions

Found the code? Know the venue? Think something is wrong? Let us know!

๐Ÿ“œ Similar Papers

In the same crypt โ€” Neural & Evolutionary

R.I.P. ๐Ÿ‘ป Ghosted

LSTM: A Search Space Odyssey

Klaus Greff, Rupesh Kumar Srivastava, ... (+3 more)

cs.NE ๐Ÿ› IEEE TNNLS ๐Ÿ“š 6.0K cites 11 years ago

Died the same way โ€” ๐Ÿ‘ป Ghosted