Mind Reading - Revision history

Chris: /* Graphing Continuous Performance */

2019-07-18T01:32:39Z

Graphing Continuous Performance

← Older revision		Revision as of 21:32, 17 July 2019
Line 134:		Line 134:
	== Graphing Continuous Performance ==		== Graphing Continuous Performance ==
	'''<span style="color:red;">Note:</span> this section only pertains to outputs of continuous networks, which you are likely not running. It's 99.9% likely that this section does not have information that is relevant to whatever it is you're working on.'''		'''<span style="color:red;">Note:</span> this section only pertains to outputs of continuous networks, which you are likely not running. It's 99.9% likely that this section does not have information that is relevant to whatever it is you're working on.'''

	Load the network output into a matrix, '''''A'''''. The first two columns contain information about the trial and the time point; the remaining columns contain the output activations, and there should be one column for each classification condition.		Load the network output into a matrix, '''''A'''''. The first two columns contain information about the trial and the time point; the remaining columns contain the output activations, and there should be one column for each classification condition.
	%Isolate output activations		%Isolate output activations

Chris: /* Graphing Performance */

2019-07-18T01:32:22Z

Graphing Performance

← Older revision		Revision as of 21:32, 17 July 2019
Line 132:		Line 132:
	%Note: the UniformOutput argument allows cellfun to return a cell array with differently-sized matrices		%Note: the UniformOutput argument allows cellfun to return a cell array with differently-sized matrices

	== Graphing Performance ==		== Graphing Continuous Performance ==
			'''<span style="color:red;">Note:</span> this section only pertains to outputs of continuous networks, which you are likely not running. It's 99.9% likely that this section does not have information that is relevant to whatever it is you're working on.'''
	Load the network output into a matrix, '''''A'''''. The first two columns contain information about the trial and the time point; the remaining columns contain the output activations, and there should be one column for each classification condition.		Load the network output into a matrix, '''''A'''''. The first two columns contain information about the trial and the time point; the remaining columns contain the output activations, and there should be one column for each classification condition.
	%Isolate output activations		%Isolate output activations

Chris: /* Measuring Classification Accuracy */

2019-07-18T01:29:47Z

Measuring Classification Accuracy

← Older revision		Revision as of 21:29, 17 July 2019
Line 131:		Line 131:
	TASK_TARGETS=cellfun(@noNulls, TASK_TARGETS, 'UniformOutput', 0);		TASK_TARGETS=cellfun(@noNulls, TASK_TARGETS, 'UniformOutput', 0);
	%Note: the UniformOutput argument allows cellfun to return a cell array with differently-sized matrices		%Note: the UniformOutput argument allows cellfun to return a cell array with differently-sized matrices

	~~== Measuring Classification Accuracy ==~~
	A testing module takes as input the continuous time-series for an entire run (i.e., with a window set to ''all'') and generates an output value for each unit in the classification output layer for every time point in the example file. Note that the number of time points will likely exceed the number of volumes, effectively oversampling the time series. There is a mathematical relationship between the oversampling and the clamp/target parameters, but I'm not positive what it is just now.

	== Graphing Performance ==		== Graphing Performance ==

Chris: /* Obsolete: MikeNet */

2019-07-18T01:29:31Z

Obsolete: MikeNet

@@ Line 102: / Line 102: @@
   sed -i .bak 's/,22/,1/g' ${SUB}_lo_post.csv
   sed -i .bak 's/,32/,2/g' ${SUB}_lo_post.csv
 == Match Targets to Inputs ==

Chris: /* "...But I only want events from specific conditions" */

2019-06-28T15:27:49Z

"...But I only want events from specific conditions"

← Older revision		Revision as of 11:27, 28 June 2019
Line 74:		Line 74:
	A MATLAB function, [[TSTagger]] has been written to assign a label to a 5-second activity window from the fMRI time series. The first step is to marry the event information from PTBParser with the time series by creating a cell array of structs. This cell array is used to create a series of .csv files (1 for each time series) where the activity associated with the trials with the designated condition codes is tagged with the associated condition code. Activity values are calculated as the median activity within a 5-second window, offset by 1 second from the trial onset. For example, if an event happened at the 10 second mark, we calculate the median activity for each region within the 11- and 16-second mark time window.		A MATLAB function, [[TSTagger]] has been written to assign a label to a 5-second activity window from the fMRI time series. The first step is to marry the event information from PTBParser with the time series by creating a cell array of structs. This cell array is used to create a series of .csv files (1 for each time series) where the activity associated with the trials with the designated condition codes is tagged with the associated condition code. Activity values are calculated as the median activity within a 5-second window, offset by 1 second from the trial onset. For example, if an event happened at the 10 second mark, we calculate the median activity for each region within the 11- and 16-second mark time window.

	====="...But I only want events ~~from~~ specific ~~conditions~~"=====		====="...But I only want events that meet specific criteria"=====
	No problem. We need to select the rows from the PTB Log file that we care about. In the example below, we're going to filter for trials with a modality code of 1 (VIS) or 2 (AUD). Assume we've already extracted the experiment info and stored it in dat.expinfo as in the above example. We simply iterate through the dat cell array of structs, find the column of interest (in this case, '''modality''') and find the rows that have the values of interest to make a filter. Then, replace the original expinfo.data with the filtered data structure.		No problem. We need to select the rows from the PTB Log file that we care about. In the example below, we're going to filter for trials with a modality code of 1 (VIS) or 2 (AUD). Assume we've already extracted the experiment info and stored it in dat.expinfo as in the above example. We simply iterate through the dat cell array of structs, find the column of interest (in this case, '''modality''') and find the rows that have the values of interest to make a filter. Then, replace the original expinfo.data with the filtered data structure.
	nruns=length(DAT);		nruns=length(DAT);

Chris: /* "...But I only want events from specific conditions" */

2019-06-28T15:24:33Z

"...But I only want events from specific conditions"

← Older revision		Revision as of 11:24, 28 June 2019
Line 76:		Line 76:
	====="...But I only want events from specific conditions"=====		====="...But I only want events from specific conditions"=====
	No problem. We need to select the rows from the PTB Log file that we care about. In the example below, we're going to filter for trials with a modality code of 1 (VIS) or 2 (AUD). Assume we've already extracted the experiment info and stored it in dat.expinfo as in the above example. We simply iterate through the dat cell array of structs, find the column of interest (in this case, '''modality''') and find the rows that have the values of interest to make a filter. Then, replace the original expinfo.data with the filtered data structure.		No problem. We need to select the rows from the PTB Log file that we care about. In the example below, we're going to filter for trials with a modality code of 1 (VIS) or 2 (AUD). Assume we've already extracted the experiment info and stored it in dat.expinfo as in the above example. We simply iterate through the dat cell array of structs, find the column of interest (in this case, '''modality''') and find the rows that have the values of interest to make a filter. Then, replace the original expinfo.data with the filtered data structure.
	nruns=length(~~dat~~);		nruns=length(DAT);
	for i=1:nruns		for i=1:nruns
	tmp=~~dat~~{i}.expinfo.data;		tmp=DAT{i}.expinfo.data;
	modalities=cell2mat(cellfun(@(x) double(x), {tmp.modality}, 'UniformOutput', false)); %This ugliness is because the script breaks if a run terminated early		modalities=cell2mat(cellfun(@(x) double(x), {tmp.modality}, 'UniformOutput', false)); %This ugliness is because the script breaks if a run terminated early
	filter=(modalities=='''1'''\|modalities=='''2'''); %which rows have the modalities of interest (1 or 2)?		filter=(modalities=='''1'''\|modalities=='''2'''); %which rows have the modalities of interest (1 or 2)?
	tmp=tmp(filter); %retain only the rows with the modalities of interest		tmp=tmp(filter); %retain only the rows with the modalities of interest
	~~dat~~{i}.expinfo.data=tmp; %~~eplaces~~ the original expinfo.data with the filtered data		DAT{i}.expinfo.data=tmp; %replaces the original expinfo.data with the filtered data
	end		end

Chris: /* TSTagger */

2019-06-28T15:19:01Z

TSTagger

@@ Line 72: / Line 72: @@
 ==== TSTagger ====
-A MATLAB function, <code>TSTagger</code> has been written to assign a label to a 5-second activity window from the fMRI time series. The first step is to marry the event information from PTBParser with the time series by creating a cell array of structs. This cell array is used to create a series of .csv files (1 for each time series) where the activity associated with the trials with the designated condition codes is tagged with the associated condition code. Activity values are calculated as the median activity within a 5-second window, offset by 1 second from the trial onset. For example, if an event happened at the 10 second mark, we calculate the median activity for each region within the 11- and 16-second mark time window.
+A MATLAB function, [[TSTagger]] has been written to assign a label to a 5-second activity window from the fMRI time series. The first step is to marry the event information from PTBParser with the time series by creating a cell array of structs. This cell array is used to create a series of .csv files (1 for each time series) where the activity associated with the trials with the designated condition codes is tagged with the associated condition code. Activity values are calculated as the median activity within a 5-second window, offset by 1 second from the trial onset. For example, if an event happened at the 10 second mark, we calculate the median activity for each region within the 11- and 16-second mark time window.
 ====="...But I only want events from specific conditions"=====

Chris: /* "...But I only want events from specific conditions" */

2018-07-06T00:33:53Z

"...But I only want events from specific conditions"

← Older revision		Revision as of 20:33, 5 July 2018
Line 96:		Line 96:
	tmp=dat{i}.expinfo.data;		tmp=dat{i}.expinfo.data;
	modalities=cell2mat(cellfun(@(x) double(x), {tmp.modality}, 'UniformOutput', false)); %This ugliness is because the script breaks if a run terminated early		modalities=cell2mat(cellfun(@(x) double(x), {tmp.modality}, 'UniformOutput', false)); %This ugliness is because the script breaks if a run terminated early
	filter=(modalities==1\|modalities==2); %which rows have the modalities of interest?		filter=(modalities=='''1'''\|modalities=='''2'''); %which rows have the modalities of interest (1 or 2)?
	tmp=tmp(filter); %retain only the rows with the modalities of interest		tmp=tmp(filter); %retain only the rows with the modalities of interest
	dat{i}.expinfo.data=tmp; %eplaces the original expinfo.data with the filtered data		dat{i}.expinfo.data=tmp; %eplaces the original expinfo.data with the filtered data

Chris: /* But I only want events from specific conditions */

2018-07-06T00:29:58Z

But I only want events from specific conditions

← Older revision		Revision as of 20:29, 5 July 2018
Line 90:		Line 90:
	This will produce a series of nruns=12 .csv files (${SUB}_0''xx''.csv), where the last column in the file is tagged with 11, 21 or 31.		This will produce a series of nruns=12 .csv files (${SUB}_0''xx''.csv), where the last column in the file is tagged with 11, 21 or 31.

	=====But I only want events from specific conditions=====		====="...But I only want events from specific conditions"=====
	No problem. We need to select the rows from the PTB Log file that we care about. In the example below, we're going to filter for trials with a modality code of 1 (VIS) or 2 (AUD). Assume we've already extracted the experiment info and stored it in dat.expinfo as in the above example. We simply iterate through the dat cell array of structs, find the column of interest (in this case, '''modality''') and find the rows that have the values of interest to make a filter. Then, replace the original expinfo.data with the filtered data structure.		No problem. We need to select the rows from the PTB Log file that we care about. In the example below, we're going to filter for trials with a modality code of 1 (VIS) or 2 (AUD). Assume we've already extracted the experiment info and stored it in dat.expinfo as in the above example. We simply iterate through the dat cell array of structs, find the column of interest (in this case, '''modality''') and find the rows that have the values of interest to make a filter. Then, replace the original expinfo.data with the filtered data structure.
	nruns=length(dat);		nruns=length(dat);

Chris: /* But I only want events from specific conditions */

2018-06-26T19:20:27Z

But I only want events from specific conditions

← Older revision		Revision as of 15:20, 26 June 2018
Line 91:		Line 91:

	=====But I only want events from specific conditions=====		=====But I only want events from specific conditions=====
	No problem. ~~First we~~ need to select the rows from the PTB Log file that we care about. In the example below, we're going to filter for trials with a modality code of 1 (VIS) or 2 (AUD)		No problem. We need to select the rows from the PTB Log file that we care about. In the example below, we're going to filter for trials with a modality code of 1 (VIS) or 2 (AUD). Assume we've already extracted the experiment info and stored it in dat.expinfo as in the above example. We simply iterate through the dat cell array of structs, find the column of interest (in this case, '''modality''') and find the rows that have the values of interest to make a filter. Then, replace the original expinfo.data with the filtered data structure.
	~~filter~~=(~~cell2mat({eio~~{1}.data.modality}))==1\|(~~cell2mat(~~{~~eio{1~~}.data.~~modality}))==2~~		nruns=length(dat);
			for i=1:nruns
			tmp=dat{i}.expinfo.data;
			modalities=cell2mat(cellfun(@(x) double(x), {tmp.modality}, 'UniformOutput', false)); %This ugliness is because the script breaks if a run terminated early
			filter=(modalities==1\|modalities==2); %which rows have the modalities of interest?
			tmp=tmp(filter); %retain only the rows with the modalities of interest
			dat{i}.expinfo.data=tmp; %eplaces the original expinfo.data with the filtered data
			end

	==== csv File Cleanup ====		==== csv File Cleanup ====