Dependency Structure Misspecification in Multi-Source Weak Supervision Models
Cachay, Salva Rühling, Boecking, Benedikt, Dubrawski, Artur
–arXiv.org Artificial Intelligence
Data programming (DP) has proven to be an attractive alternative to costly hand-labeling of data. In DP, users encode domain knowledge into \emph{labeling functions} (LF), heuristics that label a subset of the data noisily and may have complex dependencies. A label model is then fit to the LFs to produce an estimate of the unknown class label. The effects of label model misspecification on test set performance of a downstream classifier are understudied. This presents a serious awareness gap to practitioners, in particular since the dependency structure among LFs is frequently ignored in field applications of DP. We analyse modeling errors due to structure over-specification. We derive novel theoretical bounds on the modeling error and empirically show that this error can be substantial, even when modeling a seemingly sensible structure.
arXiv.org Artificial Intelligence
Jun-18-2021
- Country:
- North America > United States
- Virginia > Arlington County
- Arlington (0.04)
- Oregon > Multnomah County
- Portland (0.04)
- New York > New York County
- New York City (0.04)
- Virginia > Arlington County
- Europe
- Slovenia > Drava
- Municipality of Benedikt > Benedikt (0.04)
- Germany > Hesse
- Darmstadt Region > Darmstadt (0.04)
- Slovenia > Drava
- North America > United States
- Genre:
- Research Report (0.64)
- Technology: