Anchored Discrete Factor Analysis
This work addresses semi-supervised learning for complex latent variable models, but it appears incremental as it builds on existing method-of-moment techniques with a specific anchor assumption.
The authors tackled the problem of learning discrete factor analysis models with arbitrary latent variable structures by assuming each latent variable has an observed 'anchor' variable, enabling consistent recovery of latent moments and complete model learning, with evaluation on real-world tasks like tag prediction and medical diagnosis.
We present a semi-supervised learning algorithm for learning discrete factor analysis models with arbitrary structure on the latent variables. Our algorithm assumes that every latent variable has an "anchor", an observed variable with only that latent variable as its parent. Given such anchors, we show that it is possible to consistently recover moments of the latent variables and use these moments to learn complete models. We also introduce a new technique for improving the robustness of method-of-moment algorithms by optimizing over the marginal polytope or its relaxations. We evaluate our algorithm using two real-world tasks, tag prediction on questions from the Stack Overflow website and medical diagnosis in an emergency department.