How to Win the NeurIPS BBO ML Competition| Bayesian Optimisation| Fitting ML|Tune AI|Learn Params

NeurIPS 2020

How to Win the NeurIPS BBO ML Competition| Bayesian Optimisation| Fitting ML|Tune AI|Learn Params

Feb 16, 2021
|
30 views
Details
In this video, Ali @ImanisMind tells us how the Noah's Ark team from Huawei in London in collaboration with colleagues abroad in China won the NeurIPS 2020 black-box optimisation competition organised by #Twitter, #SigOpt, #Facebook, #ChaLearn, #Valohai, and #4Paradigm. Very short and concise video showing that: 1) Gaussian Processes are NOT out-dated (in fact they produce SOTA) 2) Multi-Objective Acquisitions matter 3) Noise processes are important 4) Acquisition functions can be misspecified needing robustness. Our code is also open source if you want to try it out: https://github.com/huawei-noah/noah-research/tree/master/HEBO You can also find a fork here: https://pypi.org/project/HEBO/ -- thanks Antoni for the help!!

Comments
loading...