A Filter is an estimate of the probability density of the inputs.
Attributes
| shape |
Methods
| distances(cue) | |
| flat_to_coords(i) | |
| learn(cue, **kwargs) | |
| neuron(coords) | |
| reset([f]) | |
| sample(n) | |
| smallest(distances) | |
| weights(distances) | |
| winner(cue) |
Initialize this Filter with an underlying Map implementation.
These values determine how much new cues influence the activation state of the Filter.
A 0 value would mean that no history is preserved (i.e. each new cue stored in the Filter completely determines the activity of the Filter) while a 1 value would mean that new cues have no impact on the activity of the Filter (i.e. the initial activity is the only activity that is ever used).
Methods
| __init__(map[, history]) | Initialize this Filter with an underlying Map implementation. |
| distances(cue) | |
| flat_to_coords(i) | |
| learn(cue, **kwargs) | |
| neuron(coords) | |
| reset([f]) | |
| sample(n) | |
| smallest(distances) | |
| weights(distances) | |
| winner(cue) |
Attributes
| shape |