MC
NARMA
M-G
Lorenz
Fig.1 recalculated, updated
Theoretical background
The whole theoretical background chapter. Hopefully didn’t forget about anything. theory_master.pdf
Neuron activations in MC, NARMA, Lorenz and MG tasks
Information measures
First draft of the chapter on information measures from master thesis.
57: Reservoir TE without significance testing & Lorenz task v4
Initial network setup
All 4 instances share the same initial W and Win matrices
Estimators setup
Transfer entropy
Active information storage
TE matrix
Measures
Significance test for differences in mean square error (MSE) using Wilcoxon signed-rank test
Reservoir TE distribution
AIS
56: Reservoir TE without significance testing & M-G task v4
Initial network setup
All 4 instances share the same initial W and Win matrices
Estimators setup
Transfer entropy
Active information storage
TE matrix
Measures
Significance test for differences in mean square error (MSE) using Wilcoxon signed-rank test
Reservoir TE distribution
AIS
55: Reservoir TE without significance testing & NARMA task v4
h2>Initial network setup
All 4 instances share the same initial W and Win matrices
Estimators setup
Transfer entropy
Active information storage
TE matrix
Measures
Significance test for differences in mean square error (MSE) using Wilcoxon signed-rank test.
Reservoir TE distribution
AIS
54: Reservoir TE without significance testing & Memory capacity task v4
Initial network setup
All 4 instances share the same initial W and Win matrices
Estimators setup
Transfer entropy
Active information storage
TE matrix
Measures
Significance test for differences in MC using Mann-Whitney rank test. Sample size N=100.
Reservoir TE distribution
AIS
Reservoir embedding MC, MG, NARMA, Lorenz kHistory_max=6, kTau_max=6
Initial network setup
Memory capacity task
Parameters with max counts
NARMA
Parameters with max counts
M-G
Parameters with max counts
Lorenz
Parameters with max counts
