Deep Learning Semantic Category Classification - Revision history

Chris: /* Event Tagging */

2019-06-13T02:20:04Z

Event Tagging

← Older revision		Revision as of 22:20, 12 June 2019
Line 21:		Line 21:
	== Event Tagging ==		== Event Tagging ==
	The TSTagger() function is used to assign conditions to event windows. It is run separately for the category conditions for high familiarity and low familiarity items:		The TSTagger() function is used to assign conditions to event windows. It is run separately for the category conditions for high familiarity and low familiarity items:
	hiconds=[11,21,31]		hiconds=[11,21,31];
	loconds=[12,22,32];		loconds=[12,22,32];
	TSTagger('tr', 2.047, 'condition', hiconds, 'volumes_dropped', 4, 'mat', SCALED, 'expinfo', eio);		TSTagger('tr', 2.047, 'condition', hiconds, 'volumes_dropped', 4, 'mat', SCALED, 'expinfo', eio);

Chris: /* Normalize and Rescale Matrices */

2019-06-13T00:59:16Z

Normalize and Rescale Matrices

← Older revision		Revision as of 20:59, 12 June 2019
Line 12:		Line 12:

	I ran into a problem after going through the remaining steps of this process and appending the tagged time series values to an existing training/testing set. Python was having a problem parsing some of the scaled activation values (it was complaining that some lines had 998 columns instead of the expected 999 columns). My first attempt at a fix seemed to work: rather than use MATLAB's default precision, use the <code>round()</code> function to round off the scaled activation values to a reasonable level of precision:		I ran into a problem after going through the remaining steps of this process and appending the tagged time series values to an existing training/testing set. Python was having a problem parsing some of the scaled activation values (it was complaining that some lines had 998 columns instead of the expected 999 columns). My first attempt at a fix seemed to work: rather than use MATLAB's default precision, use the <code>round()</code> function to round off the scaled activation values to a reasonable level of precision:
	SCALED=cellfun(@(x) round(x,4), SCALED, 'UniformOutput', false); %round the values to 4 decimal places, which seems plenty precise for our purposes.		SCALED=cellfun(@(x) round(x,4), SCALED, 'UniformOutput', false);
			%round the values to 4 decimal places, which seems plenty precise for our purposes.

	== Event Onset Extraction ==		== Event Onset Extraction ==

Chris: /* Normalize and Rescale Matrices */

2019-06-13T00:58:58Z

Normalize and Rescale Matrices

← Older revision		Revision as of 20:58, 12 June 2019
Line 12:		Line 12:

	I ran into a problem after going through the remaining steps of this process and appending the tagged time series values to an existing training/testing set. Python was having a problem parsing some of the scaled activation values (it was complaining that some lines had 998 columns instead of the expected 999 columns). My first attempt at a fix seemed to work: rather than use MATLAB's default precision, use the <code>round()</code> function to round off the scaled activation values to a reasonable level of precision:		I ran into a problem after going through the remaining steps of this process and appending the tagged time series values to an existing training/testing set. Python was having a problem parsing some of the scaled activation values (it was complaining that some lines had 998 columns instead of the expected 999 columns). My first attempt at a fix seemed to work: rather than use MATLAB's default precision, use the <code>round()</code> function to round off the scaled activation values to a reasonable level of precision:
	SCALED=round(SCALED, 4); %round the values to 4 decimal places, which seems plenty precise for our purposes.		SCALED=cellfun(@(x) round(x,4), SCALED, 'UniformOutput', false); %round the values to 4 decimal places, which seems plenty precise for our purposes.

	== Event Onset Extraction ==		== Event Onset Extraction ==

Chris: /* Training */

2018-06-06T20:10:10Z

Training

← Older revision		Revision as of 16:10, 6 June 2018
Line 53:		Line 53:
	Save trained networks using <code>python readout.py</code>		Save trained networks using <code>python readout.py</code>

	Get region cluster activations using dummy.csv (generated using <code>eye(998)</code> with a dummy column 999 tacked on at the end. ~~Node activation values were set to the average per-trial summed activation value in the training set (~490.0).~~ Run <code>python printact.py dummy</code> to get the hidden-layer activity associated with each region.		Get region cluster activations using dummy.csv (generated using <code>eye(998)</code> with a dummy column 999 tacked on at the end. Run <code>python printact.py dummy</code> to get the hidden-layer activity associated with each region.

	[[Category: Machine Learning]]		[[Category: Machine Learning]]
	[[Category: Neural Networks]]		[[Category: Neural Networks]]

Chris: /* Training */

2018-06-06T19:30:19Z

Training

← Older revision		Revision as of 15:30, 6 June 2018
Line 53:		Line 53:
	Save trained networks using <code>python readout.py</code>		Save trained networks using <code>python readout.py</code>

	Get region cluster activations using dummy.csv (generated using <code>eye(998)</code> with a dummy column 999 tacked on at the end. Node activation values were set to the average per-trial summed activation value in the training set (~490.0). Run <code>python printact.py</code> to get the hidden-layer activity associated with each region.		Get region cluster activations using dummy.csv (generated using <code>eye(998)</code> with a dummy column 999 tacked on at the end. Node activation values were set to the average per-trial summed activation value in the training set (~490.0). Run <code>python printact.py dummy</code> to get the hidden-layer activity associated with each region.

	[[Category: Machine Learning]]		[[Category: Machine Learning]]
	[[Category: Neural Networks]]		[[Category: Neural Networks]]

Chris at 19:28, 6 June 2018

2018-06-06T19:28:44Z

← Older revision		Revision as of 15:28, 6 June 2018
Line 53:		Line 53:
	Save trained networks using <code>python readout.py</code>		Save trained networks using <code>python readout.py</code>

	Get region cluster activations using dummy.csv (generated using <code>eye(998)</code> with a dummy column 999 tacked on at the end. Node activation values were set to the average per-trial summed activation value in the training set (~490.0). Run <code>python printact.py<code> to get the hidden-layer activity associated with each region.		Get region cluster activations using dummy.csv (generated using <code>eye(998)</code> with a dummy column 999 tacked on at the end. Node activation values were set to the average per-trial summed activation value in the training set (~490.0). Run <code>python printact.py</code> to get the hidden-layer activity associated with each region.

	[[Category: Machine Learning]]		[[Category: Machine Learning]]
	[[Category: Neural Networks]]		[[Category: Neural Networks]]

Chris at 19:28, 6 June 2018

2018-06-06T19:28:18Z

← Older revision		Revision as of 15:28, 6 June 2018
Line 47:		Line 47:
	sed -i .bak 's/,22/,1/g' 0${SUB}_lo_pre.csv		sed -i .bak 's/,22/,1/g' 0${SUB}_lo_pre.csv
	sed -i .bak 's/,32/,2/g' 0${SUB}_lo_pre.csv		sed -i .bak 's/,32/,2/g' 0${SUB}_lo_pre.csv

			== Training ==
			Last attempt was done using 5 subjects with T1 and T2. Reduced network size to 998 -> 32 -> 12 -> 3

			Save trained networks using <code>python readout.py</code>

			Get region cluster activations using dummy.csv (generated using <code>eye(998)</code> with a dummy column 999 tacked on at the end. Node activation values were set to the average per-trial summed activation value in the training set (~490.0). Run <code>python printact.py<code> to get the hidden-layer activity associated with each region.

	[[Category: Machine Learning]]		[[Category: Machine Learning]]
	[[Category: Neural Networks]]		[[Category: Neural Networks]]

Chris: /* Time Series Extraction and Normalization */

2018-06-05T19:42:27Z

Time Series Extraction and Normalization

← Older revision		Revision as of 15:42, 5 June 2018
Line 11:		Line 11:
	[BIN, SCALED]=binarizeMatrix(ZM, 'thresh', thresh);		[BIN, SCALED]=binarizeMatrix(ZM, 'thresh', thresh);

	I ran into a problem after going through the remaining steps of this process, and appending the tagged time series values to an existing training/testing set. Python was having a problem parsing some of the scaled activation values. My first attempt at a fix seemed to work: rather than use MATLAB's default precision, use the <code>round()</code> function to round off the scaled activation values to a reasonable level of precision:		I ran into a problem after going through the remaining steps of this process and appending the tagged time series values to an existing training/testing set. Python was having a problem parsing some of the scaled activation values (it was complaining that some lines had 998 columns instead of the expected 999 columns). My first attempt at a fix seemed to work: rather than use MATLAB's default precision, use the <code>round()</code> function to round off the scaled activation values to a reasonable level of precision:
	SCALED=round(SCALED, 4); %round the values to 4 decimal places, which seems plenty precise for our purposes.		SCALED=round(SCALED, 4); %round the values to 4 decimal places, which seems plenty precise for our purposes.

Chris: /* Normalize and Rescale Matrices */

2018-06-05T19:41:27Z

Normalize and Rescale Matrices

← Older revision		Revision as of 15:41, 5 June 2018
Line 10:		Line 10:
	ZM=normalizeMatrix(M)		ZM=normalizeMatrix(M)
	[BIN, SCALED]=binarizeMatrix(ZM, 'thresh', thresh);		[BIN, SCALED]=binarizeMatrix(ZM, 'thresh', thresh);

			I ran into a problem after going through the remaining steps of this process, and appending the tagged time series values to an existing training/testing set. Python was having a problem parsing some of the scaled activation values. My first attempt at a fix seemed to work: rather than use MATLAB's default precision, use the <code>round()</code> function to round off the scaled activation values to a reasonable level of precision:
			SCALED=round(SCALED, 4); %round the values to 4 decimal places, which seems plenty precise for our purposes.

	== Event Onset Extraction ==		== Event Onset Extraction ==

Chris at 18:12, 5 June 2018

2018-06-05T18:12:12Z

← Older revision		Revision as of 14:12, 5 June 2018
Line 46:		Line 46:

	[[Category: Machine Learning]]		[[Category: Machine Learning]]
			[[Category: Neural Networks]]