Unconstrained Monotonic Neural Networks
In this talk, I will first introduce the principles of density estimation with neural architectures. Then, I will describe Unconstrained Monotonic Neural Network, a new neural architecture for monotonic function modeling. I will conclude my presentation by showing how such transformations can be used for density estimation and present experimental results.
Antoine Wehenkel is a PhD student (FNRS Research Fellowship) in machine learning at ULiège (Belgium) under the supervision of Professor Gilles Louppe. The subject of his thesis lays at the intersection between likelihood free inference and deep learning.