ECCV 2020 Tutorial on

From NAS to HPO: Automated Deep Learning



Researchers have spent tremendous time in optimizing hyper-parameters and tweaking architectures. Can we alleviate the efforts in developing deep learning algorithms and make the researchers focus more on innovative areas? In this tutorial, we encourage the researchers to design the hyper-parameter ranges and possible network architecture combinations, and pass the workload to the machines. This tutorial will cover the important concepts in automatic machine learning, and the applications in computer vision. The audience will be able to reproduce large scale experiments through hands-on section using Jupiter Notebooks.


Tentative Agenda

08:30 - 09:15 AutoML Background and Overview

09:15 - 10:00 AutoML Toolkit and HPO for Deep Learning

10:00 - 10:30 Coffee Break

10:30 - 11:15 Neural Architecture Search

11:15 - 12:00 Hands-on Section

Contact: Hang Zhang