Skip to main content
. Author manuscript; available in PMC: 2020 Sep 1.
Published in final edited form as: Hum Mutat. 2019 Aug 7;40(9):1579–1592. doi: 10.1002/humu.23826

Table 4.

Comparing the three versions of weakly supervised regression (WSR) models on their loss functions, thresholds, label transformation, and feature embedding.

Model Loss function l(pi,yi) Threshold Label Transformation Feature Embedding
WSR1 Fixed polynomial Constant b Linear
WSR2 Parameterized ε–insensitive Parameter τ Linear
WSR3 Parameterized ε–insensitive Parameter τ Nonlinear (Kernelized)