A Python Implementation of GLAD.
Algorithm for aggregating labels given by labeler more acculately than majority vote.
- Whitehill, Jacob, et al. "Whose vote should count more: Optimal integration of labels from labelers of unknown expertise." Advances in neural information processing systems. 2009.
- numpy
- scipy
$ python glad.py data/data.txt
First line:
<numGivenLabels> <numLabelers> <numTasks> <Prior p(Z=1)>
Following <numGivenLabels> lines:
<taskId> <labelerId> <label:0|1>
- The task IDs must be integers and must be 0....
- The labeler IDs must be integers and must be 0....