Successfully reported this slideshow.
We use your LinkedIn profile and activity data to personalize ads and to show you more relevant ads. You can change your ad preferences anytime.

ブラックボックス最適化とその応用

2,938 views

Published on

『CCSE2019』で発表された資料です。

https://ccse.jp/2019/

Published in: Engineering
  • Hi there! I just wanted to share a list of sites that helped me a lot during my studies: .................................................................................................................................... www.EssayWrite.best - Write an essay .................................................................................................................................... www.LitReview.xyz - Summary of books .................................................................................................................................... www.Coursework.best - Online coursework .................................................................................................................................... www.Dissertations.me - proquest dissertations .................................................................................................................................... www.ReMovie.club - Movies reviews .................................................................................................................................... www.WebSlides.vip - Best powerpoint presentations .................................................................................................................................... www.WritePaper.info - Write a research paper .................................................................................................................................... www.EddyHelp.com - Homework help online .................................................................................................................................... www.MyResumeHelp.net - Professional resume writing service .................................................................................................................................. www.HelpWriting.net - Help with writing any papers ......................................................................................................................................... Save so as not to lose
       Reply 
    Are you sure you want to  Yes  No
    Your message goes here

ブラックボックス最適化とその応用

  1. 1. Copyright © GREE, Inc. All Rights Reserved.
  2. 2. Copyright © GREE, Inc. All Rights Reserved. • 
 AI • 
 • • Automated Machine Learning (AutoML) https://y0z.github.io/about/
  3. 3. Copyright © GREE, Inc. All Rights Reserved. • 
 
 ! 
 • ! • • Minimize f(x) subject to x ∈ X f(x)
  4. 4. Copyright © GREE, Inc. All Rights Reserved. • • ! ( ) ! • • AutoML 1 (Feurer and Hutter, 2019) • ! ! f(x) f(x) f(x) x ” ”
  5. 5. Copyright © GREE, Inc. All Rights Reserved. • • ! • • • • • • • Grey-box Bayesian Optimization for AutoML
 https://slideslive.com/38916582/keynote-greybox-bayesian- optimization-for-automl f(x)
  6. 6. Copyright © GREE, Inc. All Rights Reserved. • 
 GP-EI SMAC TPE • 
 Population-based methods CMA-ES • 
 Nelder–Mead MADS • • Google Vizier (Google) • Optuna (PFN) • Nevergrad (Facebook)
  7. 7. Copyright © GREE, Inc. All Rights Reserved. • 
 GP-EI SMAC TPE • 
 Population-based methods CMA-ES • 
 Nelder–Mead MADS • • Google Vizier (Google) • Optuna (PFN) • Nevergrad (Facebook)
  8. 8. Copyright © GREE, Inc. All Rights Reserved. • • (Cohen et al., 2005; Ozaki et al., 2017) Nelder–Mead Nelder and Mead, 1965 CNN (Ozaki et al., 2017)
  9. 9. Copyright © GREE, Inc. All Rights Reserved. Nelder–Mead reflect, expand, inside contract, outside contract, shrink 5 reflect, expand, inside contract, outside contract shrink
  10. 10. Copyright © GREE, Inc. All Rights Reserved. • Nelder–Mead • Nelder–Mead Nelder–Mead Accelerating the Nelder–Mead Method with Predictive Parallel Evaluation Yoshihiko Ozaki, Shuhei Watanabe, and Masaki Onishi
 6th ICML Workshop on Automated Machine Learning, Jun 2019. ! !f(x) ∼ GP(m(x), k(x, x′)) g(x)
  11. 11. Copyright © GREE, Inc. All Rights Reserved. 1. 2. Nelder–Mead 3. P 4. 2. Nelder–Mead ! !f(x) ∼ GP(m(x), k(x, x′)) g(x)
  12. 12. Copyright © GREE, Inc. All Rights Reserved. • • 6 (Klein et al., 2018) • ! ! • Baseline 1 shrink ( ) • Baseline 2 • • Baseline 1 49% 2 13% P = 10 J = 1,2,3,4,5 Nelder–Mead Method J Average # of eval steps Average # of evaluations Baseline 1 - 590.27 (±141.42) 614.10 (±142.82) Baseline 2 - 347.27 (±89.32) 3469.67 (±893.21) Proposed 1 406.20 (±97.24) 1534.20 (±427.69) 2 314.13 (±72.26) 2307.83 (±558.02) 3 304.97 (±54.57) 2679.13 (±464.80) 4 310.60 (±67.58) 2948.20 (±642.62) 5 301.90 (±58.70) 2942.33 (±567.27)
  13. 13. Copyright © GREE, Inc. All Rights Reserved. • ! ! • ! P = 10,20,30,40 J = 1,2,3,4,5 P, J Nelder–Mead
  14. 14. Copyright © GREE, Inc. All Rights Reserved. • • • • 8 5 KDD AutoML Workshop • Yoshihiko Ozaki and Masaki Onishi,
 “Practical Deep Neural Network Performance Prediction for Hyperparameter Optimization,”
 To appear. • https://sites.google.com/view/automl2019-workshop/
  15. 15. Copyright © GREE, Inc. All Rights Reserved.

×