Latent variable selection in multidimensional item response theory models using the expectation model selection algorithm

Author(s):  
Ping‐Feng Xu ◽  
Laixu Shang ◽  
Qian‐Zhen Zheng ◽  
Na Shan ◽  
Man‐Lai Tang
2021 ◽  
Author(s):  
Gunter Maris ◽  
Timo Bechger

We show that Boltzmann machines can formally be represented as multidimensional item response theory models. This relationship inspired a new learning principle and new ways to regularize Boltzmann machines to make them more interpretable. The core results carry over to a broader class of models including Gaussian-Bernoulli restricted Boltzmann machines.


Sign in / Sign up

Export Citation Format

Share Document