Unit 9 02 Planning Under Uncertainty MDP knowitvideos English Italian

Unit 9 02 Planning Under Uncertainty MDPYouTube↗ IA↗(long shot)
Howto & Style 10.6K views 30 likes 0 dislikes 01 Nov 2011

knowitvideos 11.8K subscribers N/A YouTube↗

Click to Play Video

[Narrator] Planning under uncertainty.

[Thrun] Pianificazione sotto incertezza.

In this class so far

In questo corso fino ad ora

we talked a good deal about planning.

abbiamo parlato abbastanza di pianificazione.

We talked about uncertainty and probabilities,

Abbiamo parlato di incertezza e probabilità,

and we also talked about learning,

ed abbiamo parlato anche di apprendimento,

but all 3 items were discussed separately.

ma i tre oggetti sono stati affrontati separatamente.

We never brought planning and uncertainty together,

Non abbiamo mai messo insieme pianificazione ed incertezza,

uncertainty and learning, or planning and learning.

incertezza e apprendimento, o pianificazione e apprendimento.

So the class today, we'll fuse planning and uncertainty

Quindi nel corso di oggi, fonderemo pianificazione ed incertezza

using techniques known as Markov decision processes or MDPs,

usando tecniche conosciute come processi decisionali di Markov o MDP,

and partial observer Markov decision processes or POMDPs.

e processi decisionali di Markov parzialmente osservabili o POMDP.

We also have a class coming up on reinforcement learning

Arriverà anche un corso di apprendimento con rinforzo

which combines all 3 of his aspects,

che combina tutti i 3 aspetti,

planning, uncertainty, and machine learning.

pianificazione, incertezza ed apprendimento automatico.

You might remember in the very first class

Potreste ricordare che nella prima lezione in assoluto

we distinguished very different characteristics of agent tasks,

abbiamo distinto differenti caratteristiche degli incarichi dell'agente,

and here are some of those.

e queste sono alcune di esse.

We distinguished deterministic was the casting environments,

Abbiamo distinto ambienti deterministici e stocastici,

and we also talked about photos as partial observable.

ed abbiamo anche parlato di completamente e parzialmente osservabile.

In the area of planning so far all of our evidence falls into this field over here, like A<i>, depth first, right first and so on.</i> The MDP algorithms

Nel campo della pianificazione fino ad ora

which I will talk about first fall into the intersection of fully observable yet stochastic, and just to remind us what the difference was,

stochastic is an environment where the outcome of an action is somewhat random. Whereas an environment that's deterministic where the outcome of an action is predictable and always the same.

An environment is fully observable if you can see the state of the environment which means if you can make all decisions based on the momentary sensory input. Whereas if you need memory,

it's partially observable. Planning in the partially observable case is called POMDP, and towards the end of this class, I'll briefly talk about POMDPs but not in any depth.

So most of this class focuses on Markov decision processes as opposed to partially observable Markov decision processes. So what is a Markov decision process? One way you can specify a Markov decision process by a graph.

Suppose you have states S1, S2, and S3, and you have actions A1 and A2. In a state transition graph, like this, is a finite state machine,

and it becomes Markov if the outcomes of actions are somewhat random. So for example if A1 over here, with a 50% probability, leads to state S2 but with another 50% probability leads to state S3.

So put differently, a Markov decision process of states, actions, a state's transition matrix, often written of the following form which is just about the same as

a conditional state transition probability that a state is prime is the correct posterior state after executing action A in a state S,

and the missing thing is the objective for the Markov decision process. What do we want to achieve? For that we often define a reward function, and for the sake of this lecture,

I will attach rewards just to states. So each state will have a function R attached that tells me how good the state is. So for example it might be worth $10

to be in the state over here, $0 to be in the state over here, and $100 to be in a state over here. So the planning problem is now the problem

which relies on an action to each possible state. So that we maximize our total reward.

Similar videos:

Qué son los SWAPS
canaloie24
55.4K views 259 likes 4 dislikes 24 Nov 2011100% match

Gundarr (Ep 18): Pistol Passing (subtitulado al castellano)
daniborr
14.7K views 63 likes 2 dislikes 10 Aug 2012100% match

Unit 2, Topic 13, Uniform Cost Search

35.8K views 15 likes 28 dislikes 05 Oct 2011100% match

mona lisa smile חלק 1
edvagabbay
48.5K views 66 likes 5 dislikes 30 Aug 2011100% match

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

Henry VIII - documentary BBC
Iris Houbaer
46.6K views 78 likes 7 dislikes 23 Apr 2014100% match

SD Gundam Mk I: Episodio 1 (subtitulado al castellano)
daniborr
107.6K views 377 likes 14 dislikes 12 Jul 2012100% match

[ENG SUB] 110811 Mnet Wide Super Junior Interview Part 3/3
yingxuan0806
38.7K views 332 likes 1 dislikes 16 Oct 2011100% 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 10 10 Passive Temporal Difference Learning.mp4

21.8K views 228 likes 8 dislikes 05 Nov 2011100% match

Unit 3 2 Probabilities

19.5K views 10 likes 0 dislikes 08 Oct 2011100% match

Homework 3 2a Naive Bayes Laplacian Smoothing ANSWER

19.9K views 29 likes 14 dislikes 01 Nov 2011100% match

Els bons modals. Dmaris Gelabert
Lourdes Soler Fernandez
31.6K views 26 likes 10 dislikes 14 Sep 2013100% match

SD Gundam Mk I: Episodio 2 (subtitulado al castellano)
daniborr
88.4K views 251 likes 9 dislikes 14 Jul 2012100% match

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

0 views 0 likes 0 dislikes 05 Jun 2026100% match

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

Akihiro Miwa & Hiroyuki Ehara on Women & Future (Eng sub)
2bfreeism
23.6K views 99 likes 7 dislikes 31 Aug 2013100% match

Unit 2, Topic 30, State Space Diagram and More Complexity

23.8K views 24 likes 7 dislikes 05 Oct 2011100% match

MTV Iggy 2ne1
AsLadies2ne1Mexico
16K views 387 likes 3 dislikes 18 Dec 2012100% match

0 views 0 likes 0 dislikes 05 Jun 2026100% match

Unit 3 7 Bayes Rule

17.5K views 35 likes 4 dislikes 08 Oct 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 16 45 Harris Corner Detector.mp4

54.4K views 340 likes 14 dislikes 22 Nov 2011100% match

Unit 5 8 SPAM Detection

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

DokuWiki 01 - Your first Page
Andreas Gohr
23.8K views 27 likes 19 dislikes 25 May 2014100% match

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

Papa,please, get the moon for me
Ana María
15K views 24 likes 0 dislikes 23 Nov 2014100% match

Unit 1w, 10 AI and Uncertainty

26.3K views 42 likes 2 dislikes 05 Oct 201167% match

Settings Privacy About Contact Donate