Neural network training relies on our ability to find "good" minimizers of
highly non-convex loss functions. It is well-known that certain network
architecture designs (e.g., skip connections) produce loss functions that train
easier, and well-chosen training parameters (batch size, learning rate,
optimizer) produce minimizers that generalize better. However, the reasons for
these differences, and their effects on the underlying loss landscape, are not
well understood. In this paper, we explore the structure of neural loss
functions, and the effect of loss landscapes on generalization, using a range
of visualization methods. First, we introduce a simple "filter normalization"
method that helps us visualize loss function curvature and make meaningful
side-by-side comparisons between loss functions. Then, using a variety of
visualizations, we explore how network architecture affects the loss landscape,
and how training parameters affect the shape of minimizers.
%0 Generic
%1 li2017visualizing
%A Li, Hao
%A Xu, Zheng
%A Taylor, Gavin
%A Studer, Christoph
%A Goldstein, Tom
%D 2017
%K loss_surface skip_connection
%T Visualizing the Loss Landscape of Neural Nets
%U https://papers.nips.cc/paper/7875-visualizing-the-loss-landscape-of-neural-nets.pdf
%X Neural network training relies on our ability to find "good" minimizers of
highly non-convex loss functions. It is well-known that certain network
architecture designs (e.g., skip connections) produce loss functions that train
easier, and well-chosen training parameters (batch size, learning rate,
optimizer) produce minimizers that generalize better. However, the reasons for
these differences, and their effects on the underlying loss landscape, are not
well understood. In this paper, we explore the structure of neural loss
functions, and the effect of loss landscapes on generalization, using a range
of visualization methods. First, we introduce a simple "filter normalization"
method that helps us visualize loss function curvature and make meaningful
side-by-side comparisons between loss functions. Then, using a variety of
visualizations, we explore how network architecture affects the loss landscape,
and how training parameters affect the shape of minimizers.
@misc{li2017visualizing,
abstract = {Neural network training relies on our ability to find "good" minimizers of
highly non-convex loss functions. It is well-known that certain network
architecture designs (e.g., skip connections) produce loss functions that train
easier, and well-chosen training parameters (batch size, learning rate,
optimizer) produce minimizers that generalize better. However, the reasons for
these differences, and their effects on the underlying loss landscape, are not
well understood. In this paper, we explore the structure of neural loss
functions, and the effect of loss landscapes on generalization, using a range
of visualization methods. First, we introduce a simple "filter normalization"
method that helps us visualize loss function curvature and make meaningful
side-by-side comparisons between loss functions. Then, using a variety of
visualizations, we explore how network architecture affects the loss landscape,
and how training parameters affect the shape of minimizers.},
added-at = {2020-02-18T18:04:40.000+0100},
author = {Li, Hao and Xu, Zheng and Taylor, Gavin and Studer, Christoph and Goldstein, Tom},
biburl = {https://www.bibsonomy.org/bibtex/2a2e4c6ef1ee6521656e8b2ed05e5b284/konstantinkobs},
description = {Visualizing the Loss Landscape of Neural Nets},
interhash = {11b08ffd9f6ee740a9a17da8e6e4c46e},
intrahash = {a2e4c6ef1ee6521656e8b2ed05e5b284},
keywords = {loss_surface skip_connection},
timestamp = {2020-02-18T18:05:36.000+0100},
title = {Visualizing the Loss Landscape of Neural Nets},
url = {https://papers.nips.cc/paper/7875-visualizing-the-loss-landscape-of-neural-nets.pdf},
year = 2017
}