Unit 9 17 Value Iteration 1YouTube↗  IA↗(long shot)
Howto & Style  24.8K views 27 likes 24 dislikes   01 Nov 2011
knowitvideos 11.8K subscribers N/A YouTube↗


Please select the problem
You must specify your email and describe the problem in detail
Describe your problem in detail if it's not obvious

Click to Play Video Click to Play Video
The definition of the expected sum of future possible discounted reward that it has given you
将来の割引報酬の和の期待値の定義を示しましたが
allows me to define a value function.
これは価値関数の定義に用いることができます
For each status, my value of the state
状態の価値はsが開始状態であり
is the expected sum of future discounted reward provided that I start in state S, then I execute policy pi.
ポリシーπを実行した時の 割引報酬の総和の期待値となります
This expression looks really complex, but it really means something really simple,
この式は一見複雑に見えますが 実際はとても単純です
which is suppose we start in the state over here,
例えばここから始めて
and you get +100 over here, -100 over here.
ここで＋100としこちらで－100としましょう
And suppose for now, every other state costs you -3.
他の場所は－3の報酬とします
For any possible policy that assigns actions to the non-absorbing states, you can now
吸収状態でないすべての状態に行為を割り当てた ポリシーを与え
simulate the agent for quite a while and compute empirically what is the average reward that is being received until you finally hit a goal state.
実際にこれをシミュレートすることで 実験的に平均報酬を求めることができます
For example, for the policy that you like,
このようなポリシーは
the value would, of course, for any state depend on how much you make progress towards the goal,
価値や状態に対してゴールまでの進捗の程度や
or whether you bounce back and forth.
壁に跳ね返されたかどうかで決まります
In fact, in this state over here, you might bounce down and have to do the loop again.
実際この状態では下にそれてしまい 回り道をする可能性もあります
But there's a well defined expectation over any possible execution of the policy pi that is generic to each state and each policy pi.
しかしすべての状態とポリシーについて 一意に定まった期待値があり
That's called a value.
これが“価値”と呼ばれます
And value functions are absolutely essential to MDP, so the way we're going to plan is we're going to iterate and compute value functions, and it will turn out
価値関数はマルコフ決定過程で 極めて重要な役割を果たします 計画を求めるためには今示した手順を シミュレートすることで価値関数を計算し
that by doing this, we're going to find better and better policies as well.
ポリシーを改善していけばよいのです
Similar videos:

Gundarr (Ep 11): Skygoth (subtitulado al castellano)
daniborr
12.3K views 52 likes 5 dislikes 02 Jul 2012100% match

Gundarr (Ep 12): Sausage Party Poopers (subtitulado al castellano)
daniborr
35.1K views 110 likes 10 dislikes 03 Jul 2012100% match

Unit 9 18 Value Iteration 2

13.4K views 16 likes 1 dislikes 01 Nov 2011100% match

CEOS Chatter
Fernando Garcia
1.3K views 2 likes 1 dislikes 14 Mar 2012100% match

Écosystèmes
cris210885
27.6K views 71 likes 23 dislikes 25 May 2013100% match

Mr. Peabody & Sherman
TheCaptionedInternet
43.6K views 74 likes 9 dislikes 03 Mar 2014100% match

Instrucciones para llorar, de Julio Cortázar
Inés Andrés
5.9K views 10 likes 0 dislikes 10 Jan 2012100% match

Unit 3 2b Question

18.4K views 1 likes 1 dislikes 08 Oct 2011100% match



Horrible Bosses (2011)
Captionwire
13.5K views 22 likes 2 dislikes 27 May 2011100% match

Unit 3 12b Question

7.6K views 2 likes 0 dislikes 08 Oct 2011100% match


Unit 9 19 Value Iteration 3

48.2K views 89 likes 90 dislikes 01 Nov 2011100% match

Unit 2, Topic 31, Sliding Blocks Puzzle

21.1K views 15 likes 2 dislikes 05 Oct 2011100% match



Unit 5 47 Gradient Descent Implementation

14K views 16 likes 17 dislikes 17 Oct 2011100% match

Recomendación kuentalibros - pizarra digital y niños con discapacidad auditiva
anitamc20
933 views 1 likes 0 dislikes 21 Dec 2012100% match

(ENG/CHI SUB) 140216 SM True Thailand Facebook EXO
Baby Xiumin
783 views 10 likes 0 dislikes 16 Feb 2014100% match
0 views 0 likes 0 dislikes 06 Jun 2026100% match

Unit 7 3c Vacuum World

8.7K views 15 likes 10 dislikes 26 Oct 2011100% match

Unit 2, Topic 20, Storage Requirements, Completeness

18.1K views 19 likes 3 dislikes 05 Oct 2011100% match




Unit 2, Topic 2, Route Finding Question

22.6K views 45 likes 1 dislikes 05 Oct 2011100% match





Efecto Desplazamiento (Crowding_out effect)
canaloie24
10.2K views 21 likes 1 dislikes 24 Nov 2011100% match

La double origine de l'être humain Part.1/2
Selim Aïssel
20.2K views 125 likes 16 dislikes 31 Jan 2013100% match

Unit 5 2 What is Machine Learning

8.2K views 7 likes 0 dislikes 17 Oct 2011100% match


¿Cuál es la función de los Bancos de Inversión?
canaloie24
17.4K views 97 likes 5 dislikes 22 Nov 2011100% match

Unit 5 8 SPAM Detection

8.6K views 15 likes 1 dislikes 17 Oct 2011100% match

TED & TEDx explained
Kyo young Chu
1.1K views 7 likes 0 dislikes 02 Apr 2012100% match

Unit 9 04 MDP Grid World

12.5K views 28 likes 2 dislikes 01 Nov 2011100% match

Unit 11 17.mp4 HMM Happy Grumpy Problem.mp4

27.9K views 164 likes 10 dislikes 11 Nov 2011100% match

DokuWiki 02 - Creating Links and Pages
Andreas Gohr
12.1K views 22 likes 5 dislikes 25 May 2014100% match

Unit 3 1 Introduction

36.6K views 95 likes 6 dislikes 08 Oct 201150% match



Unit 3 2d Question

17.5K views 3 likes 0 dislikes 08 Oct 201150% match




Settings Privacy About Contact Donate