| Single-spike information and Poisson log-likelihood | This results in the “cross-validated” single-spike in-formation: |
| Single-spike information and Poisson log-likelihood | Cross-validated single-spike information provides a useful measure for comparing models with different numbers of parameters (e.g., a l-filter vs. 2-filter LNP model), |
| avg # of excitatory filters to (a ‘h at | (C) Cross-validated single-spike information for iSTAC, cbf-LNP, and rbf-LNP, as a function of the number of filters, averaged over a population of 16 neurons (selected from [29] for having 2 8 informative filters). |
| PLS features support independent rather than synergy encoding in the DLMs | Reconstruction of decile averages (Fig 7C), cross-validated predictions (Fig 8D) and individual wing-strokes (Fig 9) are all improved by incorporating PLS features. |
| Torque reconstruction improves with independently driven motor features | Finally, we cross-validated (D) the reconstructions performing 1000 replicates with 70% of the wingstrokes as a training set and 30% withheld for a test set. |
| Torque waveform reconstruction | To test the predictive power of the reconstructions, we cross-validated the feature analysis using 70% of each decile of the data as a training set to predict the remaining 30%. |