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.

20181214 Minecraftで深層強化学習 ハンズオン

900 views

Published on

Minecraftで深層強化学習ハンズオンの資料

Published in: Engineering
  • DOWNLOAD THIS BOOKS INTO AVAILABLE FORMAT (Unlimited) ......................................................................................................................... ......................................................................................................................... Download Full PDF EBOOK here { https://soo.gd/qURD } ......................................................................................................................... Download Full EPUB Ebook here { https://soo.gd/qURD } ......................................................................................................................... Download Full doc Ebook here { https://soo.gd/qURD } ......................................................................................................................... Download PDF EBOOK here { https://soo.gd/qURD } ......................................................................................................................... Download EPUB Ebook here { https://soo.gd/qURD } ......................................................................................................................... Download doc Ebook here { https://soo.gd/qURD } ......................................................................................................................... ......................................................................................................................... ................................................................................................................................... eBook is an electronic version of a traditional print book THIS can be read by using a personal computer or by using an eBook reader. (An eBook reader can be a software application for use on a computer such as Microsoft's free Reader application, or a book-sized computer THIS is used solely as a reading device such as Nuvomedia's Rocket eBook.) Users can purchase an eBook on diskette or CD, but the most popular method of getting an eBook is to purchase a downloadable file of the eBook (or other reading material) from a Web site (such as Barnes and Noble) to be read from the user's computer or reading device. Generally, an eBook can be downloaded in five minutes or less ......................................................................................................................... .............. Browse by Genre Available eBooks .............................................................................................................................. Art, Biography, Business, Chick Lit, Children's, Christian, Classics, Comics, Contemporary, Cookbooks, Manga, Memoir, Music, Mystery, Non Fiction, Paranormal, Philosophy, Poetry, Psychology, Religion, Romance, Science, Science Fiction, Self Help, Suspense, Spirituality, Sports, Thriller, Travel, Young Adult, Crime, Ebooks, Fantasy, Fiction, Graphic Novels, Historical Fiction, History, Horror, Humor And Comedy, ......................................................................................................................... ......................................................................................................................... .....BEST SELLER FOR EBOOK RECOMMEND............................................................. ......................................................................................................................... Blowout: Corrupted Democracy, Rogue State Russia, and the Richest, Most Destructive Industry on Earth,-- The Ride of a Lifetime: Lessons Learned from 15 Years as CEO of the Walt Disney Company,-- Call Sign Chaos: Learning to Lead,-- StrengthsFinder 2.0,-- Stillness Is the Key,-- She Said: Breaking the Sexual Harassment Story THIS Helped Ignite a Movement,-- Atomic Habits: An Easy & Proven Way to Build Good Habits & Break Bad Ones,-- Everything Is Figureoutable,-- What It Takes: Lessons in the Pursuit of Excellence,-- Rich Dad Poor Dad: What the Rich Teach Their Kids About Money THIS the Poor and Middle Class Do Not!,-- The Total Money Makeover: Classic Edition: A Proven Plan for Financial Fitness,-- Shut Up and Listen!: Hard Business Truths THIS Will Help You Succeed, ......................................................................................................................... .........................................................................................................................
       Reply 
    Are you sure you want to  Yes  No
    Your message goes here

20181214 Minecraftで深層強化学習 ハンズオン

  1. 1. マインクラフトで深層強化学習 ハンズオン Preferred Networks / Chainer Evangelist Keisuke Umezawa
  2. 2. 2018/02/23 自己紹介 2 • 梅澤 慶介 • 金融マーケット予測モデルの研究開発 @AlpacaJapan • Chainer Evangelist @Preferred Networks • Twitter:@kmechann • GitHub:@keisuke-umezawa
  3. 3. 2018/11/25 Alpaca Forecast AI Prediction Matrix 3 • 最近、Bloombergアプリとしてリリースした金融商品予測システム • 30分後に価格が上がるか下がるか予測する
  4. 4. 2018/02/23 マインクラフトで深層強化学習 4
  5. 5. 2018/11/25 概要 5 • Minecraftをシミュレーション環境として深層強化学習のハンズオン を行います • 現在、マインクラフトを活用したMARLOという深層強化学習のコ ンテストが行われています • そこで使用されている、marLoというOpenAI's Gym互換の Minecraft環境を活用します
  6. 6. 2018/11/25 ハンズオン内容 6 • marLoでは下記のような環境が用意されており、今回はこのうち一 つを選んでハンズオンを行います。 • その後、皆さんに好きな課題にチャレンジしていただきます。 • だいたいのタイムスケジュール • 18:30〜 Chainer/CuPy/ChainerRLの紹介 • 18:45〜 環境設定・問題解説 • 20:00〜 演習 • 21:00 終了
  7. 7. 2018/02/23 Chainer/CuPy/ChainerRLの紹介 7
  8. 8. 2018/11/25 Chainer • Chainerとは (http://chainer.org/) • Preferred Networks製Deep Learningフレームワーク 8
  9. 9. 2018/11/25 Chainer 9 • Google社製TensorFlowなどの同類
  10. 10. 2018/11/25 CuPy 10 ChainerにおけるGPU計算を全て担当するライブラリが独立 NumPy互換APIで低コストにCPUコードをGPUへ移行 特異値分解などの線形代数アルゴリズムをGPU実行 KMeans, Gaussian Mixture ModelなどのExampleの充実 import numpy as np x = np.random.rand(10) W = np.random.rand(10, 5) y = np.dot(x, W) import cupy as cp x = cp.random.rand(10) W = cp.random.rand(10, 5) y = cp.dot(x, W) GPU https://github.com/cupy/cupy
  11. 11. 2018/11/25 拡大するChainerファミリー 11 Chainer UI Chainer Chemistry 強化学習 画像認識 可視化 グラフ構造 大規模分散 Menoh 推論特化
  12. 12. 強化学習? ChainerRL: 深層強化学習ライブラリ エージェントが環境とのインタラクションを通じて報酬を最大化 する行動を学習する
  13. 13. ChainerRLによる強化学習の流れ 1. 環境を構築
  14. 14. ChainerRLによる強化学習の流れ 2. モデルを定義 (contd.) Q-Function: 観測→各行動の価値(将来の報酬の和の期待値) ActionValue: Discrete, Quadratic
  15. 15. ChainerRLによる強化学習の流れ Distribution: Softmax, Mellowmax, Gaussian,… Policy: 観測→行動の確率分布 2. モデルを定義
  16. 16. ChainerRLによる強化学習の流れ 3. エージェントを定義
  17. 17. ChainerRLによる強化学習の流れ 4. インタラクションさせる
  18. 18. 実装済みのアルゴリズム • Deep Q-Network (Mnih et al., 2015) • Double DQN (Hasselt et al., 2016) • Normalized Advantage Function (Gu et al., 2016) • (Persistent) Advantage Learning (Bellemare et al., 2016) • Deep Deterministic Policy Gradient (Lillicrap et al., 2016) • SVG(0) (Heese et al., 2015) • Asynchronous Advantage Actor-Critic (Mnih et al., 2016) • Asynchronous N-step Q-learning (Mnih et al., 2016) • Actor-Critic with Experience Replay (Wang et al., 2017) <- NEW! • Path Consistency Learning (Nachum et al., 2017) <- NEW! • etc.
  19. 19. ChainerRL Quickstart Guide • Jupyter notebookでQ-functionを定義し、Double DQNでCart Pole Balancingを学習 https://chainer-colab-notebook.readthedocs.io/ja/latest/notebook/hands_on/chainerrl/quickstart.html
  20. 20. 2018/02/23 ハンズオンをはじめよう! 20
  21. 21. 2018/11/25 今回の強化学習の問題 21 • MarLo-FindTheGoal-v0 • 7x7 room • Goal: find the goal  (yellow block) • Rewards: • -0.01 per command • 100 commands max • 1.0 find goal • -0.1 out of time
  22. 22. 2018/11/25 ハンズオン開始! 22 • https://bit.ly/2Bfmi9z
  23. 23. 2018/02/23 chug (Chainer User Group)の紹介 23
  24. 24. 2018/02/23 Chainer User Groupの紹介 • connpass • Slack chainer-jp • Twitter @chug_jp • 活動内容 • Meetup・ハンズオンの開催 • Chainerに関する情報・ドキュメントの拡充 • Web チュートリアルの作成 • 一緒に活動できる方を募集しています !Slack #chug-jp-management に是非 24
  25. 25. 2018/02/23
  26. 26. 2018/02/23 APPENDIX 26

×