Normalization of Neural Networks using Analytic Variance Propagation

Result code in IS VaVaI

<a href="https://www.isvavai.cz/riv?ss=detail&h=RIV%2F68407700%3A21230%2F18%3A00326237" target="_blank" >RIV/68407700:21230/18:00326237 - isvavai.cz</a>

Result on the web

<a href="https://arxiv.org/abs/1803.10560" target="_blank" >https://arxiv.org/abs/1803.10560</a>

DOI - Digital Object Identifier

—

Result language

angličtina

Original language name

Normalization of Neural Networks using Analytic Variance Propagation

Original language description

We address the problem of estimating statistics of hidden units in a neural network using a method of analytic moment propagation. These statistics are useful for approximate whitening of the inputs in front of saturating non-linearities such as a sigmoid function. This is important for initialization of training and for reducing the accumulated scale and bias dependencies (compensating covariate shift), which presumably eases the learning. In batch normalization, which is currently a very widely applied technique, sample estimates of statistics of hidden units over a batch are used. The proposed estimation uses an analytic propagation of mean and variance of the training set through the network. The result depends on the network structure and its current weights but not on the specific batch input. The estimates are suitable for initialization and normalization, efficient to compute and independent of the batch size. The experimental verification well supports these claims. However, the method does not share the generalization properties of BN, to which our experiments give some additional insight.

Czech name

—

Czech description

—

Type

D - Article in proceedings

CEP classification

—

OECD FORD branch

10201 - Computer sciences, information science, bioinformathics (hardware development to be 2.2, social aspect to be 5.8)

Project

Result was created during the realization of more than one project. More information in the Projects tab.

Continuities

P - Projekt vyzkumu a vyvoje financovany z verejnych zdroju (s odkazem do CEP)

Publication year

2018

Confidentiality

S - Úplné a pravdivé údaje o projektu nepodléhají ochraně podle zvláštních právních předpisů

Article name in the collection

Proceedings of the 23rd Computer Vision Winter Workshop

ISBN

978-80-270-3395-9

ISSN

—

e-ISSN

—

Number of pages

9

Pages from-to

45-53

Publisher name

Czech Society for Cybernetics and Informatics

Place of publication

Praha

Event location

Český Krumlov

Event date

Feb 5, 2018

Type of event by nationality

WRD - Celosvětová akce

UT code for WoS article

—