Time Series Simulations - Revision history

Chris at 20:53, 26 February 2019

2019-02-26T20:53:50Z

← Older revision		Revision as of 16:53, 26 February 2019
Line 50:		Line 50:
	Ps(r)=pval(2);		Ps(r)=pval(2);
	end		end

			===Implement Autoencoder with Classifier===
			A complementary autoencoder can be created that also includes 2 classifier units to assess the benefits and costs that the additional constraints bring. 40 simulations were completed using the AE network and 40 were completed using the AE network with embedded classifier. It appears that adding the classifier negatively impacts the AE reconstruction accuracy in the synthetic data (t(78)=-6.10, p<2E-8). Classifier accuracy was uncorrelated with the AE error in the CAT/AE networks (r(39)=-.15, p>.35).

			Having done this, we can extract the autoencoder weights for the CAT/AE networks and see whether they are structured differently than the —/AE networks.

	===Extract Autoencoder Weights===		===Extract Autoencoder Weights===
Line 107:		Line 112:
	np.savetxt(savename, output_weights[0], delimiter=",", fmt="%10.5f")		np.savetxt(savename, output_weights[0], delimiter=",", fmt="%10.5f")

	With 40 ~~model files~~ to process, I used a shell script:		With 40 models for each type of network to process, I used a shell script:
	#!/bin/bash		#!/bin/bash
	for i in $(seq -f "%02g" 1 40)		for i in $(seq -f "%02g" 1 40)
Line 118:		Line 123:
	Note that there was a bug when trying to apply the h5tocsv.py script to the &ndash/AE models because I was inconsistent with my layer names. The autoencoder output layer in these models was called '''ae_out''' rather than '''ae_output'''. The python code had to be modified to reflect this difference.		Note that there was a bug when trying to apply the h5tocsv.py script to the &ndash/AE models because I was inconsistent with my layer names. The autoencoder output layer in these models was called '''ae_out''' rather than '''ae_output'''. The python code had to be modified to reflect this difference.

	~~===Implement Autoencoder with Classifier===~~
	A complementary autoencoder can be created that also includes 2 classifier units to assess the benefits and costs that the additional constraints bring. 40 simulations were completed using the AE network and 40 were completed using the AE network with embedded classifier. It appears that adding the classifier negatively impacts the AE reconstruction accuracy in the synthetic data (t(78)=-6.10, p<2E-8). Classifier accuracy was uncorrelated with the AE error in the CAT/AE networks (r(39)=-.15, p>.35).

	~~Having done this, we can extract the autoencoder weights for the CAT/AE networks and see whether they are structured differently than the —/AE networks.~~

	[[Category: Networks]]		[[Category: Networks]]
	[[Category: Network Analyses]]		[[Category: Network Analyses]]
	[[Category: Time Series]]		[[Category: Time Series]]

Chris: /* Extract Autoencoder Weights */

2019-02-26T20:52:03Z

Extract Autoencoder Weights

← Older revision		Revision as of 16:52, 26 February 2019
Line 115:		Line 115:
	python h5tocsv.py ${fname}		python h5tocsv.py ${fname}
	done		done

			Note that there was a bug when trying to apply the h5tocsv.py script to the &ndash/AE models because I was inconsistent with my layer names. The autoencoder output layer in these models was called '''ae_out''' rather than '''ae_output'''. The python code had to be modified to reflect this difference.

	===Implement Autoencoder with Classifier===		===Implement Autoencoder with Classifier===

Chris: /* Extract Autoencoder Weights */

2019-02-26T20:45:20Z

Extract Autoencoder Weights

← Older revision		Revision as of 16:45, 26 February 2019
Line 106:		Line 106:
	savename=modelfile+".ae_output.csv"		savename=modelfile+".ae_output.csv"
	np.savetxt(savename, output_weights[0], delimiter=",", fmt="%10.5f")		np.savetxt(savename, output_weights[0], delimiter=",", fmt="%10.5f")

			With 40 model files to process, I used a shell script:
			#!/bin/bash
			for i in $(seq -f "%02g" 1 40)
			do
			echo $i
			fname=catAE_model_${i}.h5
			python h5tocsv.py ${fname}
			done

	===Implement Autoencoder with Classifier===		===Implement Autoencoder with Classifier===

Chris: /* Extract Autoencoder Weights */

2019-02-26T20:38:49Z

Extract Autoencoder Weights

@@ Line 84: / Line 84: @@
      imagesc(path);
   end
 ===Implement Autoencoder with Classifier===

Chris: /* Implement Autoencoder with Classifier */

2019-02-26T20:14:26Z

Implement Autoencoder with Classifier

← Older revision		Revision as of 16:14, 26 February 2019
Line 88:		Line 88:
	A complementary autoencoder can be created that also includes 2 classifier units to assess the benefits and costs that the additional constraints bring. 40 simulations were completed using the AE network and 40 were completed using the AE network with embedded classifier. It appears that adding the classifier negatively impacts the AE reconstruction accuracy in the synthetic data (t(78)=-6.10, p<2E-8). Classifier accuracy was uncorrelated with the AE error in the CAT/AE networks (r(39)=-.15, p>.35).		A complementary autoencoder can be created that also includes 2 classifier units to assess the benefits and costs that the additional constraints bring. 40 simulations were completed using the AE network and 40 were completed using the AE network with embedded classifier. It appears that adding the classifier negatively impacts the AE reconstruction accuracy in the synthetic data (t(78)=-6.10, p<2E-8). Classifier accuracy was uncorrelated with the AE error in the CAT/AE networks (r(39)=-.15, p>.35).

	Having done this, we can extract the autoencoder weights for the CAT/AE networks and see whether they are structured differently than the &em;/AE networks.		Having done this, we can extract the autoencoder weights for the CAT/AE networks and see whether they are structured differently than the —/AE networks.

	[[Category: Networks]]		[[Category: Networks]]
	[[Category: Network Analyses]]		[[Category: Network Analyses]]
	[[Category: Time Series]]		[[Category: Time Series]]

Chris: /* Implement Autoencoder with Classifier */

2019-02-26T20:13:40Z

Implement Autoencoder with Classifier

← Older revision		Revision as of 16:13, 26 February 2019
Line 88:		Line 88:
	A complementary autoencoder can be created that also includes 2 classifier units to assess the benefits and costs that the additional constraints bring. 40 simulations were completed using the AE network and 40 were completed using the AE network with embedded classifier. It appears that adding the classifier negatively impacts the AE reconstruction accuracy in the synthetic data (t(78)=-6.10, p<2E-8). Classifier accuracy was uncorrelated with the AE error in the CAT/AE networks (r(39)=-.15, p>.35).		A complementary autoencoder can be created that also includes 2 classifier units to assess the benefits and costs that the additional constraints bring. 40 simulations were completed using the AE network and 40 were completed using the AE network with embedded classifier. It appears that adding the classifier negatively impacts the AE reconstruction accuracy in the synthetic data (t(78)=-6.10, p<2E-8). Classifier accuracy was uncorrelated with the AE error in the CAT/AE networks (r(39)=-.15, p>.35).

	Having done this, we can extract the autoencoder weights for the CAT/AE networks and see whether they are structured differently than the &~~emdash~~;AE networks.		Having done this, we can extract the autoencoder weights for the CAT/AE networks and see whether they are structured differently than the &em;/AE networks.

	[[Category: Networks]]		[[Category: Networks]]
	[[Category: Network Analyses]]		[[Category: Network Analyses]]
	[[Category: Time Series]]		[[Category: Time Series]]

Chris: /* Implement Autoencoder with Classifier */

2019-02-26T20:13:27Z

Implement Autoencoder with Classifier

← Older revision		Revision as of 16:13, 26 February 2019
Line 87:		Line 87:
	===Implement Autoencoder with Classifier===		===Implement Autoencoder with Classifier===
	A complementary autoencoder can be created that also includes 2 classifier units to assess the benefits and costs that the additional constraints bring. 40 simulations were completed using the AE network and 40 were completed using the AE network with embedded classifier. It appears that adding the classifier negatively impacts the AE reconstruction accuracy in the synthetic data (t(78)=-6.10, p<2E-8). Classifier accuracy was uncorrelated with the AE error in the CAT/AE networks (r(39)=-.15, p>.35).		A complementary autoencoder can be created that also includes 2 classifier units to assess the benefits and costs that the additional constraints bring. 40 simulations were completed using the AE network and 40 were completed using the AE network with embedded classifier. It appears that adding the classifier negatively impacts the AE reconstruction accuracy in the synthetic data (t(78)=-6.10, p<2E-8). Classifier accuracy was uncorrelated with the AE error in the CAT/AE networks (r(39)=-.15, p>.35).

			Having done this, we can extract the autoencoder weights for the CAT/AE networks and see whether they are structured differently than the &emdash;AE networks.

	[[Category: Networks]]		[[Category: Networks]]
	[[Category: Network Analyses]]		[[Category: Network Analyses]]
	[[Category: Time Series]]		[[Category: Time Series]]

Chris: /* Multivariate Connectivity Analysis */

2019-02-26T20:11:51Z

Multivariate Connectivity Analysis

← Older revision		Revision as of 16:11, 26 February 2019
Line 85:		Line 85:
	end		end

			===Implement Autoencoder with Classifier===
			A complementary autoencoder can be created that also includes 2 classifier units to assess the benefits and costs that the additional constraints bring. 40 simulations were completed using the AE network and 40 were completed using the AE network with embedded classifier. It appears that adding the classifier negatively impacts the AE reconstruction accuracy in the synthetic data (t(78)=-6.10, p<2E-8). Classifier accuracy was uncorrelated with the AE error in the CAT/AE networks (r(39)=-.15, p>.35).

	[[Category: Networks]]		[[Category: Networks]]
	[[Category: Network Analyses]]		[[Category: Network Analyses]]
	[[Category: Time Series]]		[[Category: Time Series]]

Chris at 19:01, 26 February 2019

2019-02-26T19:01:20Z

← Older revision		Revision as of 15:01, 26 February 2019
Line 84:		Line 84:
	imagesc(path);		imagesc(path);
	end		end


			[[Category: Networks]]
			[[Category: Network Analyses]]
			[[Category: Time Series]]

Chris: /* Extract Autoencoder Weights */

2019-02-26T04:01:10Z

Extract Autoencoder Weights

← Older revision		Revision as of 00:01, 26 February 2019
Line 75:		Line 75:

	The resultant .csv files can be loaded into MATLAB using <code>csvread()</code>. In the simple autoencoder example, two .csv files were produced: one was a 24 × 6 array and the other was a 6 × 24 array. Multiply the two arrays together to get a 24 × 24 matrix.		The resultant .csv files can be loaded into MATLAB using <code>csvread()</code>. In the simple autoencoder example, two .csv files were produced: one was a 24 × 6 array and the other was a 6 × 24 array. Multiply the two arrays together to get a 24 × 24 matrix.
			for m=1:4
			modelname=sprintf('model_%02d',m);
			l1=csvread([modelname '.h503.csv']);
			l2=csvread([modelname '.h504.csv']);
			path=l1*l2;
			paths(m,:,:)=path;
			figure(m);
			imagesc(path);
			end