libeblearn
Public Member Functions
ebl::net_cscsc< T, Tstate > Class Template Reference

#include <ebl_machines.h>

Inheritance diagram for ebl::net_cscsc< T, Tstate >:
ebl::layers< T, Tstate > ebl::module_1_1< T, Tstate, Tstate > ebl::module ebl::lenet_cscsc< T, Tstate >

List of all members.

Public Member Functions

 net_cscsc ()
 net_cscsc (parameter< T, Tstate > &prm, intg ini, intg inj, intg ki0, intg kj0, idx< intg > &tbl0, intg si0, intg sj0, intg ki1, intg kj1, idx< intg > &tbl1, intg si1, intg sj1, intg ki2, intg kj2, idx< intg > &tbl2, bool norm=false, bool mirror=false, bool tanh=false, bool shrink=false, bool diag=false, bool norm_pos=false)
void init (parameter< T, Tstate > &prm, intg ini, intg inj, intg ki0, intg kj0, idx< intg > &tbl0, intg si0, intg sj0, intg ki1, intg kj1, idx< intg > &tbl1, intg si1, intg sj1, intg ki2, intg kj2, idx< intg > &tbl2, bool norm=false, bool mirror=false, bool tanh=false, bool shrink=false, bool diag=false, bool norm_pos=false)

Detailed Description

template<typename T, class Tstate = bbstate_idx<T>>
class ebl::net_cscsc< T, Tstate >

Standard LeNet5-type architecture without the final e-dist RBF layer.

Constructor & Destructor Documentation

template<typename T , class Tstate >
ebl::net_cscsc< T, Tstate >::net_cscsc ( )

Empty constructor, awaiting for initialization by the user via the init() function.

template<typename T , class Tstate >
ebl::net_cscsc< T, Tstate >::net_cscsc ( parameter< T, Tstate > &  prm,
intg  ini,
intg  inj,
intg  ki0,
intg  kj0,
idx< intg > &  tbl0,
intg  si0,
intg  sj0,
intg  ki1,
intg  kj1,
idx< intg > &  tbl1,
intg  si1,
intg  sj1,
intg  ki2,
intg  kj2,
idx< intg > &  tbl2,
bool  norm = false,
bool  mirror = false,
bool  tanh = false,
bool  shrink = false,
bool  diag = false,
bool  norm_pos = false 
)

Complete constructor, calls the init() function. See the init() description for complete arguments description.

Parameters:
norm_posPosition of normalization: before subsampling (false) or after (true).

Member Function Documentation

template<typename T , class Tstate >
void ebl::net_cscsc< T, Tstate >::init ( parameter< T, Tstate > &  prm,
intg  ini,
intg  inj,
intg  ki0,
intg  kj0,
idx< intg > &  tbl0,
intg  si0,
intg  sj0,
intg  ki1,
intg  kj1,
idx< intg > &  tbl1,
intg  si1,
intg  sj1,
intg  ki2,
intg  kj2,
idx< intg > &  tbl2,
bool  norm = false,
bool  mirror = false,
bool  tanh = false,
bool  shrink = false,
bool  diag = false,
bool  norm_pos = false 
)

The init function creates the machine by stacking the modules in this order (c-s-c-s-c-f): convolution_layer, subsampling_layer, convolution_layer, subsampling_layer, convolution_layer, full_layer. <ini> <inj>: expected max size of input for preallocation of internal states <ki0> <kj0>: kernel size for first convolutional layer <tbl0>: table of connections between input anf feature maps for first layer <si0> <sj0>: subsampling for first layer <ki1> <kj1> <tbl1> <si1> <sj1>: same for next 2 layers <ki2> <kj2> <tbl2>: same for last convolution layer <outthick>: number of outputs. <prm> an idx1-ddparam in which the parameters will be allocated.

Parameters:
norm_posPosition of normalization: before subsampling (false) or after (true).
The documentation for this class was generated from the following files: