Problems of the "Sampling problem" / low Hz modulation nature are very hard/impossible model,
one approach is to use a MultiSensory encoder that would:
for one/each signal, do:
[ ] scalar encoding (different range & precision)
[ ] frequency domain (fourier filter)
[ ] derivation
[ ] ???
One idea is to use whole regions instead of an encoder, the region could be HTM or other "smart neuron"(an algorithm), like @jvitku idea
Related:
Problems of the "Sampling problem" / low Hz modulation nature are very hard/impossible model, one approach is to use a
MultiSensoryencoder that would:One idea is to use whole regions instead of an encoder, the region could be HTM or other "smart neuron"(an algorithm), like @jvitku idea