q_1pl_constraint: A custom kernel constraint function that forces nonzero...

In ML2Pvae: Variational Autoencoder Models for IRT Parameter Estimation

A custom kernel constraint function that forces nonzero weights to be equal to one, so the VAE will estimate the 1-parameter logistic model. Nonzero weights are determined by the Q matrix.

Description

A custom kernel constraint function that forces nonzero weights to be equal to one, so the VAE will estimate the 1-parameter logistic model. Nonzero weights are determined by the Q matrix.

Usage

q_1pl_constraint(Q)

Arguments

Q	a binary matrix of size num_skills by num_items

Value

returns a function whose parameters match keras kernel constraint format

ML2Pvae documentation built on May 23, 2022, 9:05 a.m.