Deep Learning Course (NYU, Spring 2020) - Week 15 – Practicum part A: Inference for latent variable energy based models (EBMs)

Course website: http://bit.ly/pDL-home Playlist: http://bit.ly/pDL-YouTube Speaker: Alfredo Canziani Week 15: http://bit.ly/pDL-en-15 0:00:00 – Week 15 – Practicum part A PRACTICUM: http://bit.ly/pDL-en-15-1 When encountering the data with multiple outputs for a single input, feed-forward networks cannot capture such implicit dependencies. Instead, latent-variable energy-based models (EBMs) come to the rescue. We developed a toy ellipse example with a fixed input and the optimal model formulation. Then, we applied latent-variable EBMs to inference the best latent variables that can learn the implicit relationships. 0:00:46 – Training data and model definition 0:18:08 – Energy and free energy for two training samples 0:37:21 – Free energy dense estimation