ACQUIRE THE CURRENT VARIANCE OF BASELINE --- Syntax 1. [v2,sx2] = vb_baseline_variance_acquisition(bayes_parm,nGall); 2. [v2,sx2] = vb_baseline_variance_acquisition(bayes_parm,nGall,Cov,sx); 1. Update sx and v using a spherical covariance Cov = I (identity matrix) 2. Fix sx using a given covariance matrix Cov --- Input (Field of bayes_parm) .megfile : MEG file for current variance estimation .twin_baseline : Time window for current variance estimation .megfile_baseline: MEG file for noise calculation .twin_noise : Time window for noise estimation .noise_reg : Regularization parameter for noise covariance .a_min : .a_max : .Fdmin : .update_v : (本当に必要か？) .update_sx : (本当に必要か？) .basisfile_global: (本当に必要か？) .basisfile : (本当に必要か？) .variance_orientation: .Tv0 : .Ta0 : .temporal_filter (optional): .trial_average (optional): --- History 2005-08-17 O. Yamashita, ver.30b 2005-08-17 O. Yamashita 2006-08-10 M. Sato Use 'vbmeg_multi_fmri' for estimation 2008-06-30 Taku Yoshioka twin_global -> twin_baseline 2011-08-26 taku-y [minor] 'bayes_parm.megfile_baseline' became effective. Copyright (C) 2011, ATR All Rights Reserved. License : New BSD License(see VBMEG_LICENSE.txt)
0001 function [v2,sx2] = vb_baseline_variance_acquisition(bayes_parm,nGall,Cov,sx); 0002 % ACQUIRE THE CURRENT VARIANCE OF BASELINE 0003 % 0004 % --- Syntax 0005 % 1. [v2,sx2] = vb_baseline_variance_acquisition(bayes_parm,nGall); 0006 % 2. [v2,sx2] = vb_baseline_variance_acquisition(bayes_parm,nGall,Cov,sx); 0007 % 0008 % 1. Update sx and v using a spherical covariance Cov = I (identity matrix) 0009 % 2. Fix sx using a given covariance matrix Cov 0010 % 0011 % --- Input 0012 % (Field of bayes_parm) 0013 % .megfile : MEG file for current variance estimation 0014 % .twin_baseline : Time window for current variance estimation 0015 % .megfile_baseline: MEG file for noise calculation 0016 % .twin_noise : Time window for noise estimation 0017 % .noise_reg : Regularization parameter for noise covariance 0018 % .a_min : 0019 % .a_max : 0020 % .Fdmin : 0021 % .update_v : (本当に必要か？) 0022 % .update_sx : (本当に必要か？) 0023 % .basisfile_global: (本当に必要か？) 0024 % .basisfile : (本当に必要か？) 0025 % .variance_orientation: 0026 % .Tv0 : 0027 % .Ta0 : 0028 % .temporal_filter (optional): 0029 % .trial_average (optional): 0030 % 0031 % --- History 0032 % 2005-08-17 O. Yamashita, ver.30b 0033 % 2005-08-17 O. Yamashita 0034 % 2006-08-10 M. Sato 0035 % Use 'vbmeg_multi_fmri' for estimation 0036 % 2008-06-30 Taku Yoshioka 0037 % twin_global -> twin_baseline 0038 % 2011-08-26 taku-y 0039 % [minor] 'bayes_parm.megfile_baseline' became effective. 0040 % 0041 % Copyright (C) 2011, ATR All Rights Reserved. 0042 % License : New BSD License(see VBMEG_LICENSE.txt) 0043 0044 if nargin ~= 2 & nargin ~=4 0045 error('Input arguments must be 2 or 4') 0046 end 0047 0048 Njall = size(nGall{1},2); 0049 0050 bayes_parm2 = bayes_parm; 0051 0052 bayes_parm2.twin_meg = bayes_parm2.twin_baseline; 0053 bayes_parm2.Tperiod = ... 0054 bayes_parm2.twin_meg(2) - bayes_parm2.twin_meg(1) + 1; 0055 bayes_parm2.Tnext = bayes_parm2.Tperiod; 0056 0057 bayes_parm2.Npre_train = 100; % = Ntrain 0058 bayes_parm2.Ntrain = 100; % : Do not use accelerated-update rule 0059 bayes_parm2.skip = 20; 0060 bayes_parm2.noise_model = 1; % Scaler noise model 0061 0062 % modified by TY 2011-08-26 0063 bayes_parm2.megfile = bayes_parm.megfile_baseline; 0064 0065 if nargin == 2 0066 % Estimate sensor noise variance 0067 wiener.update_sx = ON; 0068 [Cov] = vb_observation_noise_specification(bayes_parm2); 0069 [B] = vb_megdata_preparation(bayes_parm2); 0070 vb_parm = vb_set_vb_parm(bayes_parm2, B, [], Njall); 0071 elseif nargin == 4 0072 % Fix sensor noise variance 0073 wiener.update_sx = OFF; 0074 [B] = vb_megdata_preparation(bayes_parm2); 0075 vb_parm = vb_set_vb_parm(bayes_parm2, B, [], Njall, [], sx); 0076 end 0077 0078 wiener.update_v = ON; % Estimate global current variance 0079 vb_parm.wiener = wiener; 0080 0081 % No fMRI bias information 0082 vb_parm.Ta0 = 0; 0083 vb_parm.Tv0 = 0; 0084 0085 %[Model,Info] = vbmeg_ard_estimate2(B, nGall, [], Cov, vb_parm); 0086 [Model,Info] = vbmeg_multi_fmri(B, nGall, [], Cov, vb_parm); 0087 0088 v2 = Model.v(1); % global current variance 0089 sx2 = Model.sx; % noise variance