Deep Learning with Hadoop by unknow

Author:unknow
Language: eng
Format: epub
Publisher: Packt Publishing

Fully connected layer

The fully connected layer is the final layer of a CNN. The input volume for this layer comes from the output of the preceding Convolutional layer, ReLU, or Pooling layer. The fully connected layer takes this input and outputs an N dimensional vector, where N is the number of different classes present in the initial input datasets. The basic idea on which a fully connected layer works is that it works on the output received from the preceding layer, and identifies the specific feature that mostly correlates to a particular class. For example, if the model is predicting whether an image contains a cat or bird, it will have high values in the activation maps, which will represent some high-level features such as four legs or wings, respectively.

Download

Copyright Disclaimer:
This site does not store any files on its server. We only index and link to content provided by other sites. Please contact the content providers to delete copyright contents if any and email us, we'll remove relevant links or contents immediately.