fastica_main.cc File Reference

Demonstrates usage of fastica.h. More...

Include dependency graph for fastica_main.cc:

This graph shows which files directly or indirectly include this file:

Go to the source code of this file.

Functions
int	main (int argc, char *argv[])
Variables
const fx_module_doc	fastica_main_doc
const fx_entry_doc	fastica_main_entries []
	Here are usage instructions for this implementation of FastICA.
const fx_submodule_doc	fastica_main_submodules []

---

Detailed Description

Demonstrates usage of fastica.h.

See also:

fastica.h

Author:

Nishant Mehta

Definition in file fastica_main.cc.

---

Variable Documentation

const fx_module_doc fastica_main_doc

Initial value:

 {
  fastica_main_entries, fastica_main_submodules,
  "This program performs fastica.\n"
}

Definition at line 95 of file fastica_main.cc.

const fx_entry_doc fastica_main_entries[]

Initial value:

 {
  {"data", FX_REQUIRED, FX_STR, NULL,
   "  A file containing data.\n"},
  {"ic_filename", FX_PARAM, FX_STR, NULL,
   "  Filename to which independent components are written.\n"},
  {"unmixing_filename", FX_PARAM, FX_STR, NULL,
   "  Filename to which unmixing matrix is written.\n"},
  FX_ENTRY_DOC_DONE
}

Here are usage instructions for this implementation of FastICA.

Default values are given to the right in parentheses.

Parameters specific to this driver:

Parameters:

	data	= data file with each row being one sample (REQUIRED PARAMETER)
	ic_filename	= independent components results filename (ic.dat)
	unmixing_filename	= unmixing matrix results filename (unmixing.dat)

Parameters specific to fastica.h (should be preceded by 'fastica/' on the command line):

Parameters:

	seed	= (long) seed to the random number generator (clock() + time(0))
	approach	= {deflation, symmetric} (deflation)
	nonlinearity	= {logcosh, gauss, kurtosis, skew} (logcosh)
	fine_tune	= {true, false} (false)
	a1	= numeric constant for logcosh nonlinearity (1)
	a2	= numeric constant for gauss nonlinearity (1)
	mu	= numeric constant for fine-tuning Newton-Raphson method (1)
	stabilization	= {true, false} (false)
	epsilon	= threshold for convergence (0.0001)
	max_num_iterations	= maximum number of fixed-point iterations
	max_fine_tune	= maximum number of fine-tuning iterations
	percent_cut	= number in [0,1] indicating percent data to use in stabilization updates (1)

Example use:

 ./fastica --data=X_t.dat --ic_filename=ic.dat --unmixing_filename=W.dat
 --fastica/approach=symmetric --fastica/nonlinearity=gauss
 --fastica/stabilization=true --fastica/epsilon=0.0000001 --percent_cut=0.5

Note: Compile with verbose mode to display convergence-related values

Definition at line 79 of file fastica_main.cc.

const fx_submodule_doc fastica_main_submodules[]

Initial value:

 {
  {"fastica", &fastica_doc,
   " Responsible for performing fastica.\n"},
  FX_SUBMODULE_DOC_DONE
}

Definition at line 89 of file fastica_main.cc.