Corruption that supervises how to corrupt the input matrix. (Default : kr.ac.kaist.ir.deep.train.NoCorruption)
An objective function (Default: kr.ac.kaist.ir.deep.fn.SquaredErr)
Corruption that supervises how to corrupt the input matrix.
Corruption that supervises how to corrupt the input matrix. (Default : kr.ac.kaist.ir.deep.train.NoCorruption)
Corrupt input
Corrupt input
input to be corrupted
corrupted input
Check whether given two are same or not.
Check whether given two are same or not.
Out-type object
Out-type object
True if they are different.
An objective function (Default: kr.ac.kaist.ir.deep.fn.SquaredErr)
An objective function (Default: kr.ac.kaist.ir.deep.fn.SquaredErr)
Apply given input and compute the error
Apply given input and compute the error
A network that gets input
(Input, Real output) for error computation.
error of this network
Normalization layer
Apply given single input as one-way forward trip.
Apply given single input as one-way forward trip.
A network that gets input
input to be computed
output of the network.
Apply & Back-prop given single input
Apply & Back-prop given single input
A network that gets input
Input for error computation.
Real Output for error computation.
Make validation output
Make validation output
A network that gets input
input as string
Input Operation : VectorTree as Input & Recursive Auto-Encoder Training (no output type)
This implementation designed as a replica of the standard RAE (RAE + normalization) in this paper
,We recommend that you should not apply this method to non-AutoEncoder tasks