Different types of layers in CAFE

This is a post classifying different layers in CAFE, a popular library implementation for convolutional neural networks. Cafe can be found here https://github.com/BVLC/caffe.

  • neuron layers
  1. cudnn_sigmoid
  2. cudnn_relu
  3. cudnn_tanh
  4. sigmoid
  5. relu (rectified linear units)
  6. tanh
  7. absolute value
  8. power
  9. bnll (binomial normal log likelihodd)
  • loss layers
  1. soft max
  2. contrastive_loss_layer
  3. euclidean_loss_layer (sum of squares)
  4. hinge_loss_layer
  5. loss
  6. inforgain loss
  7. multinomial logistic loss
  8. accuracy and Top-k
  • pooling
  1. pooling
  2. cudn_pooling
  • data layers
  1. data layer
  2. hdf5_layer
  3. memory
  4. image data layer
  5. windows
  • Regularization
  1. drop out
  • Convolution
  1. conv
  2. cudn_conv
  • Common layers
  1. inner product
  2. splitting
  3. flattening
  4. concatenation
  5. slicing
  6. elementiwise
  7. argmax
  8. softmax
  9. mean-variance normalization (mV)

A similar less comprehensive classification can be found here

http://caffe.berkeleyvision.org/tutorial/layers.html

Advertisements
This entry was posted in Convoluted Neural Nets. Bookmark the permalink.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s