editing comment on how to implement dynamic correlation exclusions in spiking TE estimator

This commit is contained in:
Joseph Lizier 2022-09-05 14:43:18 +10:00 committed by GitHub
parent e747e8aa93
commit 47be5df74b
No known key found for this signature in database
GPG Key ID: 4AEE18F83AFDEB23
1 changed files with 2 additions and 1 deletions

View File

@ -40,7 +40,8 @@ import infodynamics.utils.ParsedProperties;
* This implementation of the estimator does not implement dynamic exclusion windows. Such windows make sure
* that history embeddings that overlap are not considered in nearest-neighbour searches (as this breaks the
* independece assumption). Getting dynamic exclusion windows working will probably require modifications to the
* KdTree class.
* KdTree class (or may be do-able with existing KdTree class if excluding based on nearest target spikes
* rather than in terms of timing)
*/
public class TransferEntropyCalculatorSpikingIntegration implements TransferEntropyCalculatorSpiking {