Stimulation Voter

Summary

Doc_BoxAlgorithm_StimulationVoter.png
  • Plugin name : Stimulation Voter
  • Version : 1.0
  • Author : Jussi T. Lindgren
  • Company : Inria
  • Short description : Performs majority vote on the input stimuli
  • Documentation template generation date : Jul 24 2014

Description

Votes the most frequent stimulus ID in a given time window. Outputs the winning stimulus type. Several options are possible. To process multiple inputs, use Stimulation Multiplexer first.

This box reads all the incoming stimuli a stimulus channel and returns the stimulus ID with the highest number of occurrences. Some constraints can be specified as box parameters.

Inputs

Collects the stimuli from a single input. To use several inputs, combine them first with Stimulation Multiplexer.

1. Stimulus input

Input stream, can include any stimuli.

  • Type identifier : Stimulations (0x6f752dd0, 0x082a321e)

Outputs

1. Selected stimulus

The output sends a new stimulation as soon as the most frequent stimulus type passes the criteria specified by the box parameters.

  • Type identifier : Stimulations (0x6f752dd0, 0x082a321e)

Settings

1. Number of stimuli required for vote

Minimum number of incoming stimulus required to be inside the time window to carry out a vote.

  • Type identifier : Integer (0x007deef9, 0x2f3e95c6)
  • Default value : [ 4 ]

2. Time window (secs)

The length of the time window. Stimuli are discarded after they drop outside the time window.

  • Type identifier : Float (0x512a166f, 0x5c3ef83f)
  • Default value : [ 2 ]

3. Clear votes

You can specify that stimuli can either vote as long as they remain inside the time window, or that they are discarded after they participate in a single successful vote.

  • Type identifier : Clear votes (0x17ae30f8, 0x40b57661)
  • Default value : [ After output ]

4. Output timestamp

The output stimulus can be tagged with either the time of vote, the time of the last occurrence of the winning stimulus type, or the time of the last stimulus that participated in the vote.

  • Type identifier : Output time (0x48583e8f, 0x47f22462)
  • Default value : [ Time of last voting stimulus ]

5. Reject class label

Specifies the class interpreted as the 'rejected' class.

  • Type identifier : Stimulation (0x2c132d6e, 0x44ab0d97)
  • Default value : [ OVTK_StimulationId_Label_00 ]

6. Reject class can win

If set to 'no', 'rejected' class never wins the votes and is never returned as output.

  • Type identifier : Reject can win (0x442f2f14, 0x7a17336c)
  • Default value : [ No ]

Examples

Possible usages include aggregating classifier outputs by making a majority vote over them.

Miscellaneous

This algorithm is similar to the Voting Classifier but it allows more parameters. Also, all stimulus available at time step t are used to carry out the vote at time step t. On the other hand, Stimulation Voter only supports input streams of type Stimulus.