Trjtdtk

What is the relationship between relativity and the Doppler effect?

Is there a term for accumulated dirt on the outside of your hands and feet?

What does Jesus mean regarding "Raca," and "you fool?" - is he contrasting them?

What are substitutions for coconut in curry?

Help rendering a complicated sum/product formula

Do US professors/group leaders only get a salary, but no group budget?

Asserting that Atheism and Theism are both faith based positions

Existence of a celestial body big enough for early civilization to be thought of as a second moon

In Aliens, how many people were on LV-426 before the Marines arrived​?

Is honey really a supersaturated solution? Does heating to un-crystalize redissolve it or melt it?

HP P840 HDD RAID 5 many strange drive failures

Is it insecure to send a password in a `curl` command?

Why is indicated airspeed rather than ground speed used during the takeoff roll?

What does "mu" mean as an interjection?

Is there a hypothetical scenario that would make Earth uninhabitable for humans, but not for (the majority of) other animals?

Should I be concerned about student access to a test bank?

Matrix using tikz package

World War I as a war of liberals against authoritarians?

Generic TVP tradeoffs?

How to terminate ping <dest> &

Using Past-Perfect interchangeably with the Past Continuous

Can you move over difficult terrain with only 5 feet of movement?

How could an airship be repaired midflight?

Practical application of matrices and determinants

Why do we share parameters between two different inputs in the embeddings layer?

Recurrent neural network multiple types of input KerasHow to fix these vanishing gradients?Basic encoder-decoder architectureUsing RNN (LSTM) for Gesture Recognition SystemTwo-class classification model with multi-type input dataWays to Encode context for text classification?Best practice for short sentences in a deep learning networkMulti-input Convolutional Neural Network for Images ClassificationCombining different features as input to Neural NetworkHow to optimally train deep learning model using output as new input

0

$begingroup$

I noticed in some deep learning networks that have two inputs to the network, they use one embeddings layer to share the parameters between these two different inputs.

As an example, in Keras:

input_target = Input((1,))
input_context = Input((1,))
embedding = Embedding(vocab_size, embed_size, input_length=1, name='embedding')
target = embedding(input_target)
context = embedding(input_context)

Why do they use this way?

To make everything clear, the other case is: for each input we have different embeddings layer before moving to the RNN or CNN layers.

deep-learning keras word-embeddings embeddings

share|improve this question

asked 2 days ago

GhanemGhanem

1186

$endgroup$

• 
  
  
  

  
  

  
  
  
  
  
  

  
  $begingroup$
  It depends on the use case. Sometimes you have parameter sharing in order to decrease parameters or because all inputs need to be embedded in the same manner.
  $endgroup$
  – Andreas Look
  2 days ago
  

  
  
  
  


• 
  
  
  

  
  

  
  
  
  
  
  

  
  $begingroup$
  @Andreas Look , could you give an example?
  $endgroup$
  – Ghanem
  2 days ago
  

  
  
  
  


• 
  
  
  

  
  

  
  
  
  
  
  

  
  $begingroup$
  e.g. you embedd two images in a low dimensional space where distance is interpretable and want to calculate their similarity afterwards. like siamese networks
  $endgroup$
  – Andreas Look
  2 days ago
  

  
  
  
  

add a comment | 

0

$begingroup$

I noticed in some deep learning networks that have two inputs to the network, they use one embeddings layer to share the parameters between these two different inputs.

As an example, in Keras:

input_target = Input((1,))
input_context = Input((1,))
embedding = Embedding(vocab_size, embed_size, input_length=1, name='embedding')
target = embedding(input_target)
context = embedding(input_context)

Why do they use this way?

To make everything clear, the other case is: for each input we have different embeddings layer before moving to the RNN or CNN layers.

deep-learning keras word-embeddings embeddings

share|improve this question

asked 2 days ago

GhanemGhanem

1186

$endgroup$

• 
  
  
  

  
  

  
  
  
  
  
  

  
  $begingroup$
  It depends on the use case. Sometimes you have parameter sharing in order to decrease parameters or because all inputs need to be embedded in the same manner.
  $endgroup$
  – Andreas Look
  2 days ago
  

  
  
  
  


• 
  
  
  

  
  

  
  
  
  
  
  

  
  $begingroup$
  @Andreas Look , could you give an example?
  $endgroup$
  – Ghanem
  2 days ago
  

  
  
  
  


• 
  
  
  

  
  

  
  
  
  
  
  

  
  $begingroup$
  e.g. you embedd two images in a low dimensional space where distance is interpretable and want to calculate their similarity afterwards. like siamese networks
  $endgroup$
  – Andreas Look
  2 days ago
  

  
  
  
  

add a comment | 

$begingroup$

I noticed in some deep learning networks that have two inputs to the network, they use one embeddings layer to share the parameters between these two different inputs.

As an example, in Keras:

input_target = Input((1,))
input_context = Input((1,))
embedding = Embedding(vocab_size, embed_size, input_length=1, name='embedding')
target = embedding(input_target)
context = embedding(input_context)

Why do they use this way?

To make everything clear, the other case is: for each input we have different embeddings layer before moving to the RNN or CNN layers.

deep-learning keras word-embeddings embeddings

share|improve this question

asked 2 days ago

GhanemGhanem

1186

$endgroup$

input_target = Input((1,))
input_context = Input((1,))
embedding = Embedding(vocab_size, embed_size, input_length=1, name='embedding')
target = embedding(input_target)
context = embedding(input_context)

share|improve this question

asked 2 days ago

GhanemGhanem

1186

share|improve this question

asked 2 days ago

GhanemGhanem

1186

asked 2 days ago

GhanemGhanem

1186

asked 2 days ago

GhanemGhanem

1186